' i: ' - I 7 ' : :] ,,,' - - ! -'I ~ ~ 1: I '9~~~~ . . V'. .).- . (9 Disease Control Priorities in Developing Countries A WORLD BANK BOOK Disease Control Priorities in Developing Countries EDITORS Dean T. Jamison W. Henry Mosley Anthony R. Measham Jose Luis Bobadilla Publshed for the Word Bark Oxford University Press O.ord University Press OXFORD NEW YORK TORONTO DELHI BOMBAY CALCUTT.A MADRAS KARACHI KUALA LUMPLIR SINGAPORE HONG KONCO TOKYO NAIROBI DAR ES SALAAM CAPE TOWN MELBOLIRNE AUICKLAND and associated companies in BERLIN IBADAN 1993 The Intemational Bank for Reconstruction and Development / THE WORLD BANK 1818 H Street, N.W. Washington, D.C. 20433, U.S.A. Published by Oxford University Press, Inc. 200 Madison Avenue, New York, N.Y. 10016 Oxford is a registered trademark of Oxford University Press. All rights reserved. No part of this publication may be reproduced, stored in a retrieval system, or transmitted, in any form or by any means, electronic, mechanical, photocopying, recording, or otherwise, without the prior permission of Oxford University Press. Manufactured in the United States of America First printing October 1993 The findings, interpretations, and conclusions expressed in this study are entirely those of the authors and should not be attrib- uted in any manner to the World Bank, to its affiliated organizations, or to members of its Board of Executive Directors or the countries they represent. The maps that accompany the text have been prepared solely for the convenience of the reader; the designations and presen- tation of material in them do not imply the expression of any opinion whatsoever on the part of the World Bank, its affiliates, or its Board or member countries conceming the legal status of any country, territory, city, or area, or of the authorities thereof, or concerning the delimitation of its boundaries or its national affiliation. Librart of Congress Cataloging-in-Publication Data Disease c(ontrol priorities in developing countries / editors, Dean T. Jamison ... let al.]. p. cm. Includes bibliographical references. ISBN 0-19-520990-7 1. Public health-Developing countries. 2. Medicine, Preventive- Developilig countries. 3. Health planning-Developing counmties. 1. Jamison, Dean T. 11. Intemational Bank for Reconstmuction and Development. [DNLM: 1. Communicable Disease Control-economics. 2. Developing Countries-economics. 3. Health Policv-economics. 4. Health Priorties-economics. WA 110 D6 11] RA441.5.D57 1993 614'.42272'4-dc20 DN LM/DLC for Library of Congress 92-48723 CIP DEDICATION This volume is dedicated to the men and women who, under the leadership of Dr. D. A. Henderson, participated in the World Health Organization's Intensified Smallpox Eradica- tion Programme. The successful completion of that effort in 1977 obviated the need for a chapter on smallpox in this collection. Contents Preface xi Public Health Significance Acknowledgments *. Lowering the Incidence of ARI Case Management Contributors xv Priorities for ARI Control Appendix 4A. Sources of Data and Method Used Part One Introduction to Obtain Cost-Effectiveness Estimates Summarized in Table 4-3 1. Disease Control Priorities in Developing Countries: An Overview Dean T. Jamison 3 5. Diarrheal Diseases Jose Martines, Margaret The World Bank Health Sector Priorities Review Phillips, and Richard G . Feachem 91 Assessing the Cost-Effectiveness of Intervention The Public Health Significance of Diarrhea The Findings: Public Health Intetventions Lowering Disease Incidence The Findings: Clinical Interentions Case Management Conclusions 6. Poliomvelitis Dean T. Jamison, Alberto M. Torres. Appendix IA. Clinical and Public Health Interventions . C Appendix I B. Countries and Territories as Grouped in this Licoln C. Chen, and Joseph L. Melnick 117 Collection Prevention and Eradication 2. Causes of Death in Industrial and Developing Coase Management Countries: Estimates for 1985-1990 Alan D. Appendix 6A. Epidemiology of Polio Lopez 35 7. Helminth Infection Kenneth S. Warren Causes of Death D. A. P. Bundy, Roy M. Anderson, A. R. Davis, Causes of Death in Industrial Countries in 1985 Donald A. Henderson, Dean T. ]amison, Nicholas Estimated Cause-of-Death Patterns in 1985 Summary and Conclusions Prescott, and Alfred Senft 131 3. The Epidemiologic Transition and Health Priorities Publc Health Significance of Helmnth Infection JsLusBobadilla, Julio Frenk, Rafael Lozano, Strategies for Control of Helminth Infection Jos6 Luis Boail,JloFek aalLzn,Chemoprophylaxis as a Primary Strategy Tomas Frejka, and Claudio Stern 5 1 Priorities and Conclusion Conceptual Framework 8. Measles Sle F Mexico: Historic Trends 8tany 0. oster, Deborah A. The Epidemiologic Transition in Mexico McFarland, and A. Meredith John 161 Future Demographic, Epidemiologic, and Social Changes Epidemiology Implications for the Health Care System Measles Infection and Its Cosr Prevention of Measles Part Two The Unfinished Agenda, I Infectious Measles Strategies for the 1990s Disease Conclusions 4. Acute Respiratory Infection Sally K. Stansfield 9. Tetanus Robert Steinglass, Logan Brenzel, and Donald S. Shepard 67 and Allison Percy 189 Risk Factors for ARI Public Health Significance of Tetanus Clinical Syndromes Causing ARI Mortality Prevention of Tetanus vii viii Contents Case Management of Tetanus Prevention Research Agenda for the 1990s Case Management Appendix 9A. Cost-Effectiveness of Tetanus Immunization Cost-Effectiveness of Dengue Control Programs Priorities 10. Rheumatic Heart Disease Catherine Michaud, Appendix 14A. Definitions of Variables in the 10. Rheumatic Heart Disease Cost-Effectiveness Model Jorge Trejo-Gutierrez, Carlos Cruz, and Appendix 14B. Relationships in the Cost-Effectiveness Thomas A. Pearson 221 Model Background Appendix 14C. Numerical Values of Input Parameters Pathogenesis 15. Hepatitis B Mark Kane, John Clements, Clinical Aspects and Dale Hu 321 Diagnosis The Public Health Significance History and Epidemiology of HBV Infection Elements of Preventive Strategy Hepatitis B Vaccines Elements of Case Management Strategy HB Vaccines and Immunization Priorities Strategies for Control Cost-Effectiveness Considerations 11. Tuberculosis Christopher Murray, Karel Styblo, Hepatitis B Vaccine and Future Immunization Policy and Annik Rouillon 233 Transfer of Technology Tuberculosis Incidence and Mortality Conclusions Prevention Curative Care Part Three The Unfinished Agenda, 11 12. Leprosy Myo Thet Htoon, Jeanne Bertolli, Reproductive Health and Malnutriton and Lies D. Kosasih 261 16. Excess Fertility Susan Cochrane and Frederick Distribution and Risk Factors Sai 333 Current Prevalence and Trends The Significance of Excess Fertility Economic Costs Reducing Excess Fertility Prevention Strategy Case Management: Unwanted Pregnancies Case Management Priorities Priorities for Operations Research Appendix 16A. Tables Priorities for Resource Allocation 17. Matemal and Perinatal Health Julia A. Walsh, Conclusion Chris N. Feifer, Anthony R. Measham, and Paul J. Appendix 12A. Tables Appendix 12B. Cost-Effectiveness Analysis Gertler 363 Appendix 12C. Cost-Effectiveness of Multidrug Therapy Public Health Significance Appendix 12D. Second Analysis: Treatment of Risk Reduction prior to Conception Complications Pregnancy, Delivery, and the Neonatal Period Appendix I 2E. Cost Issues in Screening and Strategies for Two Standardized Populations Immunization Priorities Appendix 12F. Effect of Accuracy of Diagnosis Appendix 17A. Regression Equations Used in the Appendix 12G. General Comments About Construction of Tables 17-11 through 17-17 Cost-Effectiveness Modeling . . . Appendix 12H. Total Cost of a Leprosy Control 18. ProteIn-Energy Malnutrition Program Per Pinstrup-Andersen, Susan Burger, Jean-Pierre 13. Malaria Jose A. Najera, Bemhard H. Liese, Habicht, and Karen Peterson 391 13. Mala'ria Hos~ a.mNmera281 Bernhard H.Liese,The Public Health Significance of PEM and Jeffrey Hammer 281 Causes of PEM A Natural History: Parasite and Vector Interventions for Control of PEM Malaria as a Disease Strategies and Priorities The Public Health Significance of Malaria Appendix ISA. Estimating Nutritionial Benefits Malaria Control Estimate of Cost-Effectiveness 19. Micronutrient Deficiency Disorders Henry M. Priorities Levin, Ernesto Pollitt, Rae Galloway, and Judith 14. Dengue (with Notes on Yellow Fever and McGuire 421 Japanese Encephalitis) Donald S. Shepard and Public Health Significance Prevention Scott B. Halstead 303 Case Management Public Health Significance Assessment of the Effect of Interventions Economic Costs Sumrnary and Conclusions Disease Control Piorities in Developing Countnies Ex Cost Estimation Case Management Comparing Costs Priorities Idiosyncratic Differences in Costs 25. Injury Sally K. Stansfield, Gordon S. Smith, and Appendix 19A. Prevalence, Programs, Recommended W P Intakes, and Indicators of Deficiencies William P. McGreevey 609 Appendix 1 9B. Costs of Supplementation Definitions and Fortification Risk Factors Appendix 19C. Criteria of Effectiveness Injurious Events Appendix 19D. Cost-Benefit Analysis The Public Health Significance of Injury Appendix 19E. Tables Lowering the Incidence and Severity of Injury Case Management Part Four Emerging Problems Priorities for Injury Control Appendix 25A. Sources of Data for Effectiveness 20. HIV Infection and Sexually Transmitted Diseases Calculations Mead Over and Peter Piot 455 26. Cataract Jonathan C. Javitt 635 The Epidemiology of STDs Public Health Significance of STDs Public Health Significance of Cataract Lowering or Postponing the Incidence of STDs Lowering or Postponing Disease Incidence Case Management and Secondary Prevention Reducing the Burden of Cataract Blindness Priorities Research Priorities for Ending Cataract Blindness Appendix 20A. The Medical Consequences of Sexually Conclusions Transmitted Diseases 27. Oral Health Douglas Bratthall and David E. Appendix 20B. A Simulation Model of an STD Epidemic Barmes 647 Appendix 20C. Management of Selected Classic Dental Caries Syndromes Periodontal Diseases 21. Cancers Howard Barnum and E. Robert Other Oral Diseases Greenberg 529 Disease Prevalence: Current Levels and Trends Public Health Importance Economic Costs Prevention Elements of Preventive Strategy Treatment Oral Health Personnel Treatmeloping aStrategy forCancerPriorities for Resource Allocation Developing a Strategy for Cancer Appendix 21A. The Ten Most Important Cancers 28. Schizophrenia and Manic-Depressive Illness in the Developing World Peter Cowley and Richard Jed Wyatt 661 22. Diabetes J. Patrick Vaughan, Lucy Gilson, Risk Factors and Anne Mills 561 Incidence and Prevalence: Schizophrenia The Significance of Diabetes to Public Health Incidence and Prevalence: Manic-Depressive Illness ThwerSingnificance Pof Diabetes to cidence Morbidity and Mortality: Schizophrenia LowerianageormPostponing DiabetesIncidenceMorbidity and Mortality: Manic-Depressive Illness Prioaities Burden of Schizophrenia Appendix 22A. Diagnosis and Self-Care of Diabetes Burden of Manic-Depressive Illness Therapeutic Strategy for Schizophrenia 23. Cardiovascular Disease Thomas A. Pearson, Therapeutic Strategy for Manic-Depressive Illness Dean T. Jamison, and Jorge Trejo-Gutierrez 577 Cost-Effective Schizophrenia Case Management Atherosclerosis and Hypertension Manic-Depressive Illness Case Management The Burden of Cardiovascular Disease Gaps between Good and Actual Practice Strategies for Preventing Cardiovascular Disease Priorities for Control Ctof Cardiovascular Conditions Appendix 28A. Sources Used to Obtain Cost-Effectiveness Case Management Estimatesulr onlton Conclusions and Priorities Appendix 23A. Cardiovascular Mortality Differentiated by Sex Part Five Conclusion 24. Chronic Obstructive Pulmonary Disease 29 The Health Transition: Implications for J. Richard Bumgarmer and Frank E. Speizer 595 Health Policy in Developing Countries Description of the Diseases W. Henry Mosley, Jos~ Luis Bobadilla, and Etiology of COPD Lung Function and the Development of COPD Dean T. Jamison 673 Risk Factors The Health Transition The Public Health Significance of the Condition The Demographic Transition and Population Aging Lowering or Postponing Disease Incidence Policy Implications for National Governments x Contents Intemational Aid Types of Child Survival Interventions Conclusions Competing Concems Appendix 29A. Regional Groupings of Countries and Networks for Health Services Development Territories C. Priority Setting for Health Service Efficiency: Appendixes The Role of Measurement of Burden of Illness Gavrin Mooney and Andrew Creese 731 A. Control of Tobacco Production and Use Kenneth Thaudn ofnesa Prit Setting The Burden of Illness Priority Settng Stanley 703 Measuring the Effectiveness of Interventions The Adverse Effects of Tobacco Cost-Effectiveness Comparisons for Priority Setting Tobacco Production and Consumption ConcludingComments Tobacco Industry and Promotion D. Rationales for Choice in Public Health: The Role Tobacco Control Strategies Economic Analysis and Conclusions of Epidemiology Andre Prost and Michel Conclusions Jancloes 741 B. Reducing Mortality in Children Under Five: Validation of the Results A Continuing Priority Carl E. Taylor and Validity of Indicators Vulmiri Ramalingaswami 725 Comprehensive in Contrast to Selective Care Some Priority Interventions The Demand for Health Services Defining Local Priorities The Decisionmaking Process Determinants of Successful Child Survival Programs Concluding Remarks All dollar amounts are current U.S. dollars unless otherwise stated. Preface Between 1950 and 1990, life expectancy in developing coun- offer preventive and case management guidelines critical to tries increased from forty to sixty-three years with a concomi- improving the quality of care; health program managers will tant rise in the incidence of the noncommunicable diseases of find sets of chapters on such subjects as maternal and child adults and the elderly (Feachem and others 1992). Yet there health related conditions to help them set priorities more remains a huge unfinished agenda for dealing with under- objectively. Planners at the district, regional, and national nutrition and the communicable childhood diseases. These level will find information on how to improve the allocation trends lead to increasingly diverse and complicated epidemio- of resources. logical profiles in developing countries. At the same time, new The need for health sector reform is virtually global. Devel- epidemic diseases like AIDS are emerging; and the health of the oped and developing countries, centrally planned and market- poor during economic crisis is a source of growing concern. oriented health systems, successful and flawed health These developments have intensified the need for better in- institutions all seem to share two basic attitudes: a profound formation on the effectiveness and cost of health interven- dissatisfaction with the present organization and financing tions. To assist countries to define essential health service mechanisms of health care delivery, and a conviction that packages, this book provides information on disease control there are ways to obtain better results with the available interventions for the commonest diseases and injuries in de- resources. To be effective, health-sector reformers will need to veloping countries. review existing services and adapt them to provide the most The decision to undertake this review did not come easily. cost-effective interventions available. This book will meet its Several objections were immediately obvious. A review of purpose if it contributes to this exercise. disease control priorities in developing countries, given the The contents of this volume served as a major source of magnitude of that topic, could not simultaneously do justice background information for the World Bank's World Develop- to the equally critical questions of implementation capacity ment Report 1993 on health. It is timely that this collection, and of financing. Nor could it cover the full range of tropical which serves as a companion and a reference to the Report, endemic diseases in any depth, given the likely resources. The should appear so soon after it. decision was made to go ahead for two reasons. First, because the combined insights of economists, epidemiologists, and clinicians could give valuable guidance in the difficult choices References facing decisionmakers in developing countries, aid agencies, and the World Bank. Second, because, for the most part, other Feachem, Richard G. A., Tord Kjellstrom, Christopher J. L. Murray, reviews did not systematically assess the cost-effectiveness of Mead Over, and Margaret Phillips, eds. 1992. Health of Adults in the available interventions. In the one major review that did Developing World. New York: Oxford University Press. provide information on cost-effectiveness (Walsh and Warren Walsh, Julia, and Kenneth S. Warren, eds. 1986. Strategies for PrimarY 1986), the range of diseases covered was restricted predomi- HealthiCare: TechnologtesAppropriatefortheDetelopingWorld.Chicago nantly to the important diseases of childhood. and London: University of Chicago Press. This collection is intended for health practitioners at World Bank. 1993. World Develpnent Report 1993: Investing in Health. every level. For health care providers, individual chapters New York: Oxford University Press. xi Acknowledgments We wish to acknowledge the high degree of support and We are also deeply grateful to the many chapter authors encouragement we have received from many quarters follow- and reviewers who worked extremely hard, throughout the ing the decision to proceed with this review. The World Bank long and difficult revision process, with virtually no com- provided the bulk of the resources and the review was a pensation for their efforts. We wish particularly to acknowl- principal activity of the Population and Human Resources edge David Bell, Richard Feachem, and William Foege, who Department under Ann Hamilton for four years. The World provided extensive and valuable comments. At the risk of Health Organization provided invaluable support, mainly omitting others whose contributions were extensive, we through the contributions of a large number of staff members, also wish to acknowledge major contributions from Jacques who either coauthored or reviewed chapters. We are indebted Baudouy, Alan Berg, Robert Black, John Briscoe, Guy Car- to the Rockefeller Foundation for critical financial support, rin, E. Chigan, Joseph Davis, Nicholas Drager, Davidson which enabled us to hold review meetings in Woods Hole Gwatkin, Alaya Hammad, Ralph Henderson, Kenneth in 1990 and at the Bellagio Study and Conference Center on Hill, Jeffrey Koplan, Jean-Louis Lamboray, Joanne Leslie, Lake Como in 1992. In 1991, the Centers for Disease Control Richard Morrow, Philip Musgrove, Richard Peto, Nancy and Prevention (cDc) hosted a major review meeting for Pielemeier, Barry Popkin, William Reinke, Julia Rushby, selected chapters in Atlanta; the All-India Institute of Medical Ismail Sirageldin, Eleuther Tarimo, Anne Tinker, and David Sciences hosted a similar review meeting in New Delhi, early Werner. Finally, we thank the anonymous peer reviewers at in 1992; and the International Clinical EpidemiologyNetwork the World Bank for their helpful evaluations of the text; (INCLEN) arranged for critical discussion of many chapters at its Coni Benedicto and Christopher Wilson for tactical skills 1992 annual meeting in Indonesia. This extensive series of in keeping work on track; Joanne Ainsworth for splendid reviews contributed greatly to the quality of this collection, and tactful editing; and Jenepher Moseley for coordinating and we are much indebted to the sponsoring institutions. timely publication. xiiL Contributors EDITORS Dean T. Jamison is a professor of public health and of education at the University of California, Los Angeles. He serves as part-time adviser on population, health, and nutrition to the Latin America and the Caribbean Regional Office at the World Bank. W. Henry Mosley is chairman of the Department of Population Dynamics at the Johns Hopkins School of Hygiene and Public Health. Anthony R. Measham is adviser for population, health, and nutrition in the New Delhi resident mission of the World Bank. Jose Luis Bobadilla is a senior health policy specialist in the Population, Health, and Nutrition Department at the World Bank. AUTHORS Roy M. Anderson Susan Burger Center for Infectious Disease Epidemiology Emory University, Atlanta, Georgia Department of Zoology, University of Oxford Lincoln C. Chen Howard Bamum Harvard School of Public Health The World Bank John Clements David E. Barmes World Health Organization World Health Organization Susan Cochrane Jeanne Bertolli The World Bank University of California, Los Angeles Peter Cowley Douglas Bratthall The World Bank World Health Collaborating Center, Malmo, Sweden Andrew Creese Logan Brenzel World Health Organization REACH, Arlington, Virginia Carlos Cruz J. Richard Bumgamer Instituto Nacional de Salud Puiblica, World Health Organization Cuernavaca, Mexico D. A. P. Bundy A. R. Davis Imperial College, London World Health Organization xv xtq Contributors Richard G. Feachem Jose Martines London School of Hygiene and Tropical Medicine World Health Organization Chris N. Feifer Deborah A. McFarland The Rand Corporation, Califomia Emory University and Centers for Disease Control Stanley 0. Foster and Prevention, Atlanta, Georgia Centers for Disease Control and Prevention, Atlanta, Georgia William P. McGreevey Tomas Frejka The World Bank The Population Council, Regional Office Judith McGuire for Latin America, Mexico City The World Bank Julio Frenk Joseph L. Melnick Instituto Nacional de Salud Puiblica, Cuernavaca, Mexico Baylor College of Medicine, Houston Rae Galloway Catherine Michaud The World Bank Harvard School of Public Health Paul J. Gertler Anne Mills The Rand Corporation, California London School of Hygiene and Tropical Medicine Lucy Gilson Gavin Mooney London School of Hygiene and Tropical Medicine University of Aberdeen, Sydney, and Troms0 E. Robert Greenberg Christopher Murray Dartmouth Medical School, Hanover, New Hampshire Harvard School of Public Health Jean-Pierre Habicht Jose A. Najera Cornell University World Health Organization Scott B. Halstead Mead Over The Rockefeller Foundation The World Bank Jeffrey Hammer Thomas A. Pearson The World Bank Imogene Bassett Research Institute and Columbia Donald A. Henderson University United States Department of Health and Human Services Allison Percy Myo Thet Htoon REACH, Arlington, Virginia Myanmar Ministry of Health Karen Peterson Dale Hu Harvard School of Public Health Centers for Disease Control and Prevention, Atlanta, Georgia Margaret Phillips MichelJancloes Mexico City World Health Organization Per Pinstrup-Andersen Jonathan C. Javitt International Food Policy Research Institute (IFPRI), Georgetown University Medical Center Washington, D.C. A. Meredith John Peter Piot Princeton University World Health Organization Mark Kane Emesto Pollitt World Health Organization University of Califomia, Davis Lies D. Kosasih Nicholas Prescott Jakarta The World Bank Henry M. Levin Andre Prost Stanford University World Health OrganiZation Bernhard H. Liese Vulmiri Ramalingaswami The World Bank All India Institute of Medical Sciences, New Delhi Alan D. Lopez Annik Rouillon World Health Organization International Union Against Tuberculosis, Paris Rafael Lozano Frederick Sai Instituto Nacional de Salud Puiblica, Cuernavaca, Mexico International Planned Parenthood Federation, London Disease Control Priorities in Developing Countries xvii Alfred Senft Karel Styblo Brown University International Union Against Tuberculosis, Paris Donald S. Shepard Carl E. Taylor Brandeis University, Waltham, Massachusetts The Johns Hopkins University School of Hygiene Gordon Smith and Public Health, and UNICEF The Johns Hopkins University School of Hygiene Alberto M. Torres and Public Health World Health Organization Frank E. Speizer Jorge Trejo-Gutierrez Harvard Medical School and Harvard School The Regional Hospital, Guadalajara, Mexico of Public Health J. Patrick Vaughan Kenneth Stanley London School of Hygiene and Tropical Medicine Harvard School of Public Health Julia A. Walsh Sally K. Stansfield Harvard School of Public Health McGill-Ethiopia Project, Addis Ababa Kenneth S. Warren Robert Steinglass Biofield, New York REACH, Arlington, Virginia Richard Jed Wyatt Claudio Stem National Institute of Mental Health Neurosciences El Colegio de Mexico, Mexico City Center at St. Elizabeth's, Washington, D.C. PART ONE Introduction Oterview Causes of Death The Epidenmologic Transinon Disease Control Priorities in Developing Countries: An Overview Dean T. Jamison Intemational health policy is at a time of transition. A pre- the allocation of resources to disease control. By addressing a transition environment dominated by high fertility, high mor- broad range of conditions, the World Bank Health Sector tality, infectious disease, and malnutrition is giving way to a Priorities Review was able both to estimate the cost-effective- low-mortality, low-fertility environment. For the past two ness of CCD interventions and to place them into the context decades much of the international public health community of interventions for noncommunicable diseases and for com- has focused attention on the communicable childhood diseases municable diseases of adults.3 (CCDs). Although there are exceptions to this generalization- This chapter has three purposes: to set the context for the continued concern about the tropical diseases, for example- collection as a whole by describing the Health Sector Priorities much of the debate and analysis has concentrated on whether Review and outlining the methods of cost-effectiveness anal- CCD problems would be best addressed by broad-reaching strat- ysis used; to summarize the findings of the condition-specific egies or by selective ones. Significant technical and program- analyses, which are reported in parts 2, 3, and 4 of this collec- matic progress has been made in this period, and the focus of tion; and to draw a few broad conclusions. In chapter 29, concern on CCD has clearly been appropriate: the problems are Mosley, Bobadilla, and Jamison describe the health transition great; the technological and epidemiological tools have be- more fully and provide an amplified set of implications for come powerful; and the payoff for adapting and applying what policy. is known is very high. The success of CCD control efforts Before proceeding with the substance of the chapter, I will combined with large and sustained fertility reductions in many put forth four caveats. First, any general discussion of condi- developing countries has led, however, to the "health transi- tions and priorities for so vast and diverse a set of countries as tion," that is, the change from a pretransition environment those that make up the developing world naturally runs the dominated by high fertility, high mortality, infectious disease, risks of overstating generalities and understating differences. and malnutrition to a low-mortality, low-fertility environment The authors represented in this collection assume the reader with a disease profile that increasingly emphasizes noncommu- to be already familiar with this diversity, and, therefore, we nicable conditions of adults and the elderly.' Figure 1-1 avoid continually repeating caveats about the limits to gener- illustrates the progress of the health transition through demo- alization. Nevertheless, our concern is with generalization- graphic to epidemiologic change. with addressing trends and findings that are important for a This collection reports the findings of the Health Sector sufficiently large number of countries that they assume signif- Priorities Review, a review conducted by the World Bank of icance for the developing world as a whole. That said, the the implications for disease control priorities of the health conclusions of this chapter, and of the collection as a whole, transition. The core of this collection consists of analyses canbestbeviewedasausefulstartingpointforcountry-specific undertaken for the Health Sector Priorities Review that assess analyses and certainly not as a substitute for them. the significance to public health of individual diseases (or A second caveat concerns limits to the coverage of the related clusters of diseases) and of what is now known about World Bank Health Sector Priorities Review. One shortcom- the cost and effectiveness of relevant interventions for their ing is the lack of attention to most mental and neurological control.2 To the extent possible, the cost-effectiveness of in- illness; analyses structured like those in this collection for a tervention has been summarized by estimates of marginal cost broader range of neuropsychiatric conditions may be initiated per disability-adjusted life-year (DALY) gained; although this soon, but results are currently unavailable.4 Many other con- measure is imperfect, and often varies with the scale of the ditions-some minor, some more important-were omitted in control effort and across environments, its estimation, for each order to keep the scope of the review manageable. Likewise, of a large number of interventions, does indicate priorities for interventions associated with the very diverse range of tradi- 3 4 Dean T. Jamison Figure 1-1. Relationships between Demographic, Epidemiologic, and Health Transitions Health Transition Demographic Transition Epidemiologic Transition Urbanization Infectious Chronic and non- Urbanization 10 disease Fertility Population locommunicable Industrialization mortality declines ages diseases emerge Rising incomes dcie Expansion of education Improved medic-al and public health technology v Source: Chapter 28. tional medical practices were not included. These gaps in to be implemented. In either the public or private sector the coverage genuinely limit the scope of the review. But I feel three key design elements for delivery systems-planning de- they do not significantly alter conclusions conceming the velopment of human and physical infrastructure, planning the very broad range of conditions and interventions that were logistical system for drugs and supplies, and planning appropri- included. ate information and incentive structures and financial instru- A third caveat concems the perspective of this document, ments-all depend in important ways on the intervention mix. which is that of govemments or of development assistance At the same time, the cost-effectiveness of interventions will agencies working with govemments. Other perspectives- vary with the capacity of local infrastructure to deliver them those of clinicians or of patients or of nongovernmental orga- (Over 1988), as is well illustrated in chapter 14, "Dengue," by nizations, for example-are also important, and this review's Halstead and Shepard. Ideally, then, the first two tasks of findingsconcerninginterventioncost-effectivenesshavesome health policy analysis should be addressed iteratively rather relevance from those perspectives. But the review's main pur- than sequentially. The third task of policy, choice of the pose is to identify those interventions that, on the basis of their appropriate mix of governmental instruments, deals with what cost-effectiveness, govemmental policy should seek to encour- governments can do through provision of information, taxa- age or discourage. tion, regulation, direct investment, and research.6 The brief Fourth, we assume throughout this collection that the anal- discussion of these instruments in this chapter points to their ysis of health policy can usefully be divided into three tasks: to importance; it does not assess them in any depth. choose attractive interventions; to design delivery systems for If the familiar pattem of problems in developing countries- such interventions; and to choose appropriate govemmental communicable childhood disease, undernutrition, and excess instruments to encourage these interventions. Because of the ferrility-could be expected to continue their dominance of changing pattem of disease, delineated in chapters 3 and 29, the epidemiological profile, then there would be little need for this collection readdresses the first task of health planning, the broad reassessment of objectives attempted in this collec- that of choosing interventions. For this reason I take the time tion. Analyses aimed at improving the infrastructure, logistics, here to consider the three tasks together. The first task, choos- and financial aspects of the delivery systems so they are able to ing interventions, assesses the cost-effectiveness of potential deal with the pretransition conditions could simply proceed. disease control technologies by combining technical analysis But massive epidemiological change is already deeply penetrat- (epidemiological and clinical) with economic considerations; ing the developing world, and response to this change has it will also, occasionally, extend to the task of assessing the typically been to import the high-cost approaches of the indus- benefits of intervention in relation to cost broadly defined.' trialized nations to the (very limited) extent that resources The second task, designing delivery systems, has dominated permit-hence the timeliness of broad assessment of the cost- thinking about health policy. Economically sensible delivery effectiveness of intervention options. Reassessment of the systems, however, mustfollow choice of what interventions are design of delivery systems and the appropriate role for govern- Disease Control Priorities in Developing Countries: An Overview 5 ment also claims priority; but that reassessment must follow Table 1-2 provides definitions of terms that are frequently identification of the interventions to be delivered (or encour- used in this chapter and throughout the collection. After aged) and those to be discouraged. reviewing drafts of all the chapters, I concluded that dividing The World Bank's 1980 policy paper on the health sector interventions between those that are public health oriented (World Bank 1980) focused on communicable diseases of and those that are clinically oriented was more useful than childhood, but shortly after that paper was published another distinguishing between preventive and case management in- widely cited paper prepared at the World Bank (Evans, Hall, terventions, a distinction that the chapter authors had origi- and Warford 1981) was drawing attention to the emergence of nally been asked to make. In table 1-2 I define how we use these behavioral disorders and noncommunicable disease as signifi- terms, and later in this chapter I divide the summaries of the cant problems in many developing countries. Although Evans, findings of the condition-specific chapters into two sections, Hall, and Warford were explicitly pessimistic about prospects one on public health intervention and the other on clinical for cost-effective intervention to address these emerging prob- intervention. lems, World Bank staff members were encouraged to begin In table 1-2 1 also define the objectives of intervention, addressing the issue in their work with individual govern- which, as we categorize them, include primary prevention, ments, and analyses for China (Jamison and others 1984; secondary prevention, cure, rehabilitation, and palliation. Al- Bumgarner and others 1990) and Brazil (Briscoe and others though there is some tendency for public health interventions 1989) attempted to develop relevant policy responses in epi- to have primary prevention as their main objective, this is far demiological environments characterized both by lingering from universally true; by the same token, some clinical inter- problems of childhood disease and malnutrition and by rapid ventions also seek primary prevention. Hence the importance emergence of noncommunicable disease (NCD).7 During the of clarity about objectives in discussing individual interven- progress of these country studies it became clear that very little tions. Table 1-2 also defines the instruments of govemment analysis indeed had gone into the task of selecting from and policy that can be used to encourage or discourage use of adapting the broad range of NCD interventions available in specific interventions. the industrialized countries to generate options relevant to The cost-effectiveness of any given intervention varies ac- the extremely cost-constrained environments of developing cording to circumstances, and the main sources of such varia- countries. tion are discussed in the next section. A particularly important This lack of analysis of appropriate approaches for develop- source of variation, however, results from the general mortality ing countries to take in dealing with NCDs provided the im- level of the environment, and each set of chapter authors was petus, then, for the World Bank Health Sector Priorities asked to consider two environments (if relevant to their con- Review. At the same time it seemed appropriate to reassess the ditions). One of these was a high-mortality, high-fertility cost-effectiveness of interventions addressing communicable environment with low gross national product (GNP) per capita conditions, malnutrition, and excess fertility; such a reassess- and extremely limited resources available for the health sector; ment would both provide the context for judging the relative Nigeria might be an example. The other was an environment cost-effectiveness of interventions for NCDs and allow judg- with relatively low fertility and mortality, a middle-income ments to be made on intervention priority for child survival. GNP, and more substantial resources for the health sector; Thailand would be typical here. Many of the authors of the The World Bank Health Sector Priorities Review disease-specific analyses used both these paradigms, whereas others used only one or found neither appropriate. What is The most important part of the review is a series of analyses on important to note here, though, is that conclusions concerning diseases and conditions of great importance in part or all of the intervention attractiveness can vary quite substantially de- developing world.8 The first part of table I -1 lists the included pending on a country's progress through the health transition topics and indicates which chapter of the collection covers and that this point was very much a starting point for the them; the second part lists potentially important conditions analyses in the World Bank review. that were not included but that, ideally, should be addressed in further work. The authors of each chapter were asked to Assessing the Cost-effectiveness of Intervention undertake three tasks: to assess the current and probable future public health significance of the conditions in developing This section contains a discussion of general issues associated countries; to judge the cost and effectiveness of altemative with choosing interventions, that is, with criteria for cost-ef- approaches to preventing the condition in different contexts; fective choice. The nature of the instruments open to govern- and to judge the cost and effectiveness of altemative ap- ment to promote cost-effective intervention is discussed in proaches to case management for the conditions in various chapter 29. The purpose in both chapters is not to provide an contexts. Ideally each chapter would have been written by an account of methodological issues associated with economic economist, an epidemiologist, and a clinician or biomedical assessment of intervention options; rather we wish simply to scientist. Although each of these categories is well represented describe the basic concepts being applied and refer the reader among the chapter authors, relatively few of the individual to the relevant literature (for example, Drummond, Stoddart, chapters ended up with all three. and Torrance 1987). 6 Dean T. Jamison Table 1-1. Selected Clusters of Diseases and Conditions Status Unfinished agerda Emerging problems Included Infections principally affecting children Human immunodeficiency virus infection and sexually transmitted Acute respiratorv infections (chap. 4) diseases (chap. 20) Diarrheal diseases (chap. 5) Cancers (chap. 21) Poliomyelitis (chap. 6) Diabetes (chap. 22) Helminthic infections (chap. 7) Cardiovascular disease (chap. 23) Measles (chap. 8) Chronic obstructive pulmonary disease (chap. 24) Tetanus (chap. 9) Injury (chap. 25) Cataract (chap. 26) Oral health (chap. 27) Schizophrenia and manic-depressive illness (chap. 28) Other infections Rheumatic heart disease (chap. 10) Tuberculosis (chap. 11) Leprosy (chap. 12) Malaria (chap. 13) Dengue and yellow fever (chap. 14) Hepatitis B (chap. 15) Reproductive health and nalnutrion Excess fertility (chap. 16) Matemal and perinatal health (chap. 17) Protein-energy malnutrition (chap. 18) Micronutrient deficiency disorders (chap. 19), Omitted Lymphatic filariasis Epilepsy African trypanosomiasis Affective disorders Chagas' disease Alcohol and other drug abuse Leishmaniasis Arthritis Skin diseases Influenza Appendicitis Hernia Source: Author. The methods of assessing the cost and effectiveness of inter- cause of their economic benefits or because of some more vention vary from chapter to chapter, complicating the task of proximal effectiveness measure. Ideally, economic benefits cross-chapter comparisons of the cost-effectiveness of inter- would be the criterion; the results of the analysis-phrased in vention. Because of wide variation in the nature of the condi- dollars of output value given the dollar value of inputs-pro- tions and the adequacy of the literature, the chapters vary in vide standards for assessment of interventions across sectors: the extent to which they provide quantitative estimates of immunization as against irrigation as against smaller class sizes, cost-effectiveness; hence the results are sometimes difficult to say. When there are good markets for products, benefits can be compare. Nevertheless, the authors proceeded with similar assessed in monetary terms by using market prices (that is, objectives and methods, and, if one accepts the results as willingness of consumers to pay) to value benefits. Even when reasonable first approximations, the collection does generate willingness-to-pay valuation cannot be assessed directly be- the raw materials for comparative assessment of the cost-effec- cause of lack of market prices, as is typically true in the health tiveness of intervention. Often these first approximations rely sector, questions in surveys are increasingly being used to elicit on epidemiological or clinical judgments combined with the information about hypothetical willingness-to-pay. Briscoe results of published studies, or, in those cases where published and de Ferranti (1988) indicate the potential for this approach data were unavailable, judgments were made on the basis of in valuation of water projects, but applicability in the health discussion with experienced observers. Some reviewers of this sector remains to be assessed. Nonetheless, pervasive problems effort have been uncomfortable with the explicit use of judg- of consumer ignorance of effectiveness of intervention (pre- mental assessments; we have consistently encouraged them to viously discussed) and a widespread tendency for individuals let us know of any cases where current clinical or public health systematically to underestimate risks (Weinstein 1989) suggest decisions are being made on the basis of better information on that willingness-to-pay assessments will probably have limited efficacy or cost than we use. application to health. An alternative approach-sometimes A critical choice in applications of economic analysis to called the human capital approach-is to view health invest- resource allocation is that of whether to value outcomes be- ments as instrumental to improving economic productivity; Disease Control Priorities in Developing Countries: An Overview 7 Table 1-2. Definition of Terrs 1. Interventions. The term "intervention" is used in this chapter to denote actions taken by or for individuals to reduce the risk, duration, or severity of an adverse health condition. Interventions are the proximal cause of deliberate changes in risks, duration, or severity; instruments of policy (see below) encourage, discourage, or undertake interventions. Stopping smoking, for example, is an intervention that an individual can take to reduce risk from a range of diseases; taxing tobacco products is a potential instrument of government policy to encourage this intervention. I divide interventions into those that are "public health" and those that are "clinical." 1.1 Public health interventions. These are interventions sought of or directed toward entire populations or population subgroups; this chapter divides public health interventions into five broad categories-change of personal behavior, control of environmental hazards, immunization, mass chemoprophylaxis, and screening and referral. (Table 1-5 provides a broad range of examples of each of these strategies for population-based intervention.) 1.2 Clinical interventions. These are interventions provided at facilities, usually to individuals. This chapter divides clinical interventions into those that can be provided at the clinic (community, private, work-based, or school-based), at a district hospital, or at a referral hospital. 2. Objectives of lntervention. The objectives of intervention are structured, in this chapter, into five categories:' 2.1 Primary prevention aims to reduce the risk of a condition occurring by lowering the level of risk factors or instituting policies to forestall their emergence. (This latter is sometimes referred to as "primordial prevention.") 2.2 Secondary prevention aims to reduce the duration or severity of a condition or physiological risk factor in order to forestall its leading to more adverse consequences. 2.3 Cure of a condition aims to remove its cause and restore function to the status quo ante. 2.4 Rehabilitation aims to restore (or partially restore) physical, psychological, or social function resulting from a previous or chronic condition. 2.5 Palliation aims to reduce pain and suffering from a condition for which no means of cure or rehabilitation is currently available. (This may range from the use of aspirin for headaches to use of opiates to control terminal cancer pain.) 3. Instruments of Policy. These are the activities that can (potentially) be undertaken by governments or other entities that wish to encourage or discourage interventions, or, importantly, to expand the menu of potential intervention. I distinguish five major instruments or policy: 3.1 Use of information, education, and communication (IEC) seeks to improve the knowledge of individuals (and service providers) about the consequences of their choices. 3.2 Use of taxes and subsidies on commodities, services, and pollutants seeks to effect appropriate behavioral responses. 3.3 Use of regulLationi and legislation seeks to limit availability of certain commodities, to curtail certain practices, and to define the rLIles governing finance and provision of health services. 3.4 Use of direct expenditures seeks to provide (or finance provision of) selected interventions (e.g., immunizations) or to provide infrastructure (e.g., medical schools) that facilitates provision of a range of interventions. 3.5 Undertakingresearchanddevelopment (or encouraging them through subsidies) is an instrument central to the goal of expanding the range of interventions available and reducing their cost. a. The International Epidemiology Association's Dictonar-y of Epidemiology (Last 1988) provides a helpful discussion of different types of prevention but, in- terestingly, has no entries for "cure or "rehabititation." Their term "tertiary prevention." which is nor used here, seems to encompass both "rehabilitation" and "palliation," as we define those terms. Source: Author. estimates of the effect of a health intervention on productivity measure for making such an assessment that allows compari- thus provide a lower bound to total benefits. One example sons across the health sector, even if intersectoral comparisons comes from assessing the effect on the productivity of rubber (cost-benefit analyses) remain infeasible.9 There is now a plantation workers of correcting iron deficiencies (Basta, valuable literature on how effectiveness measures to aggregate Soekirman, and Scrimshaw 1979; Levin and others, chapter the disability-, morbidity-, and premature mortality-averting 19, this collection); other examples come from assessment of effects of interventions across the health sector might be the effect on productivity of malaria control efforts (Najera constructed and applied (Barnum 1987; Zeckhauser and Shep- andothers,chapterl3, thiscollection). It is worthnotingthat ard 1976; Over 1988; and Feachem, Graham, and Timaeus willingness-to-pay and human capital approaches tend to 1989). Such measures, in addition to providing the effective- imply different values to be attached to the life of different ness measures for cost-effectiveness analyses, can be used with individuals of the same age in the same country. Phelps and epidemiological information to assess the burden of disease in Mushlin (1991) discuss relations between cost-effectiveness a population, as has recently been done for the major regions and cost-benefit analyses of health projects; they conclude that of the world (World Bank, 1993). Nonetheless, inherent dif- willingness to set a cutoff level of acceptable cost-effectiveness ficulties remain, and these are usefully discussed in Murray results in equivalence between cost-effectiveness and cost- (1990). Table 1-3 sets forth the characteristics of the main benefit approaches. approaches to effectiveness measurement in the literature; More typically, however, outcomes will be assessed in deaths from a practical perspective, the use of ratings based on expert or disability averted, and the task is to come up with some judgment is probably the best that can now be done if the 8 Dean T. Jamison purpose of the analysis is to compare interventions across the collection is a particular form of the more general concept of sector, although, as Preston (1 991) has noted, these measures "quality-adjusted life-year" (QALY) introduced by Zeckhauser must be used with care. This has been the approach adopted and Shepard (1976). in the World Bank Health Sector Priorities Review. Garber and Phelps (1992) provide the basic theoretical I conclude that a workable measure for effectiveness for most underpinnings for cost-effectiveness analyses in health that of the analysis will be disability-adjusted life-years gained (or adjust life-years for quality. (See Johannesson [19921 for a DALYS). The DALY gain associated with averting a death is, general discussion of discounting healthy life-years, and see simply. the number of years between the age at which the death Cropper, Aydede, and Portney [1992] for empirical assess- would have occurred and the individual's expected age at ments of time preference for saving lives.) Authors of individ- death, given survival to the given age, with years gained in ual chapters assess the losses due to disability or morbidity in future years discounted back to the present at a discount rate ways judged suitable to the conditions with which they are of 3 percent pet annum in all chapters in this collection. dealing. Thus, reduction in morbidity and disability can be Unhealthy life-years are given lower weights than healthy explicitly considered in the analysis, and most chapters do so, ones, depending on degree of disability (by the rating proce- if the conditions they deal with have significant consequences dure described in the preceding paragraph) so that the effec- other than death. The approach used here explicitly values tiveness of interventions to address morbidity or disability can years of healthy life at all ages equally; this assumption can be be measured in terms that permit comparison with interven- readily relaxed, however, to give greater weight to those age tions that avert mortality. The DALY measure used in this groups likely, say, to have more dependents (Musgrove 1991). Table 1 3. Alternative Approaches to Measuring Effectiveness of Intervention Approach to measurement Cost Possible bias Applicabilityu Example Mortality Deaths averted Very low Highly biased conditions Mvedium Assessment of priorities in child involving disability survival (Walsh and Warren 1979) Years of potential life lost Very low Highly biased against conditions Medium Regularly used by Centers for involvingdisability Disease Concrol to assess burden of disease in the United States, (MMwvR 1992) Quality-of-life adjusted life-yearsh Expert ratings assessment Low Unrepresentative experts High Ghana Health Assessment Project Team (1981); this collection Survey-based Medium n.a. Low (in practice) Rosser scale (Rosser and Kind 1978); European quality-of-life assessments (EuroQol Group 1990) Risk tradeoffs High Questionable relevance of Low (in practice) Various qualitv-of-life artificial gambles assessments (Tan-Torres 1990) Quantity-of-life high tradeoffs: Medium/high Probably low for patient-level Medium Various quality-of-life Individual length vs quality of life dectsionmaking assessments (Tan-Torres 1990) Quantitv-of-life tradeoff: Medium Probably low for social Medium/high Vaccine development study Across individuals decisionmaking (Institute of Medicine 1986; Nord 1991) Calibration of preexisting Medium Probably low Low Cairns and Johnston 1991 condition-specific studies n.a. Not applicable. Note: This table does nor review approaches to measuring the economic benefits of changes in health status. Such measures-based, for example, orn will- ingness to pay for reductions in the probability of adverse outcomes or on assessment of health-related determinants of labor productivity (human capital)- allow conclusions to he drawn about rhe artracriveness of particular health interventions relative to their cost, nor simply by comparison with other interventions (for examples, see chapter 19). No usable set of benefit measures is available across conditions. a. Availability for application in health-sector-wide cost-effectiveness studies. b. Each of the methods for quality of life measurement-ratings, risk tradeoffs, quantity-of-life tradeoffs, calibrations-can be undertaken by different groups, possibly with different results. The groups can be of "experts," respondents to a survey, or, in a clinical setting, potential patients. For the ratings method, this table comments on both expert and survey approaches: a similar breakdown could be provided for each method. See Fallowfield (1990) for a gen- eral discussion of these matters. Source: See references in final column of table. Disease Control Priorities in Developing Countres: An Overview 9 Costs are generally assessed at market prices. In some cases, pletely free market for foreign exchange and the costs of however, for some inputs into health care, costs may be lower nontradables are similar to those of the comparator country. in developing countries (for example, for semiskilled labor). Another important issue in cost analysis concerns assess- These costs are typically for inputs that cannot be traded ment of the amount and value of time required of patients or internationally, and their existence undermines attempts to caretakers; the importance of mothers' time, in particular, for estimate costs that are not simply country-specific. Squire compliance with child survival interventions has been stressed (1989) provides a general discussion of approaches to dealing by Leslie (1989). These time costs are potentially difficult to with nontradables in project analysis; his results, though, are value (Briscoe and de Ferranti 1988) and have been neglected more relevant to country-specific assessments than to cross- in this collection. It is hoped that subsequent work will redress national comparisons. this omission. A related issue concerns treatment of costs that The working conclusion of this chapter is that for drugs, for will ensue from intervention success; Levin and others (chap- most equipment, and for high-level manpower, considerations ter 19, this collection) point out that substantial food costs can of cost variability between high- and low-income countries are result from micronutrient supplementation or parasite control. essentially irrelevant. For facilities and lower-level manpower The existence of such costs suggests the importance, in these they are likely to change some numbers, but in most cases costs cases, of broadening the definition of the intervention. can more reasonably be expressed in constant dollars than, say, A final issue concerning cost analysis is that of joint costs, as fractions of local per capita income-a method that assumes that is, the situation where several interventions are essentially essentially no health sector inputs to be internationally trad- made available with a (partially) common set of inputs. The able. The chapters "Cancer," by Barnum and Greenberg chapters in this collection handle this in part by defining (chapter 21, this collection), and "Tuberculosis," by Murray, interventions in terms of natural packages; the chapter "Polio- Styblo, and Rouillon (chapter 11, this collection), attempt to myelitis," by Jamison and others (chapter 6, this collection), divide costs into those for traded goods and those for non- considers the preventive intervention for polio to be diphthe- tradables. Their assessments do suggest that local costs will ria-pertussis-tetanus vaccine plus polio immunization, and as- often be important and that those who attempt to assess the sesses the cost-effectiveness of that package, because polio cost-effectiveness of intervention in a country-specific context immunization would (almost always) be given with the vac- should pay close attention to this issue unless there is a com- cine. In many cases, however, such packaging would get too Table 1-4. Factors Influencing Variation in Cost-Effectiveness Influencingfactor Examples Epidemiological environment Prevalence of condition Screening and referral programs for leprosy; for cervical and breast cancer Incidence of condition BCG immunization for tuberculosis; preventive measures for many injuries Case-fatality rate Measles immunization; oral rehydration therapy for diarrhea Transmission dynamics of infectious conditions Treatment of sexually transmitted diseases in core vs noncore groups; vector control for malaria, dengue Existence of competing risks of synergisms Measles vaccination: amplification of cost-effectiveness by strengthening individuals in a general way. Among the very young or elderly, competing risks reduce the cost-effectiveness of some targeted interventions. Indivzdual characteristics Age Cancer treatment: more cost-effective for younger patients Tendency to compliance Tuberculosis chemotherapy; antihypertensive medication Tendency to self-refer Sexually transmitted diseases control Levels of risk factors Hypertension and hyperlipidemia Individual variation in values Attitude toward disability relative to risk of death; can lead to individual differences in intervention effectiveness System characteristics Local costs of non-traded inputs to health care system Real costs of care-intensive interventions (such as hospitalization to ensure compliance with tuberculosis chemotherapy) are low where wages are low, because most health care personnel are relatively immobile Generalized systemic competence Case management of dengue hemorrhagic fever: high cost and low effectiveness in unsophisticated systems. Cost-effectiveness, at the margin, of some interventions in a system with high level of professionalism and capacity may be much less than in less well developed systems Discount rate Hepatitis B immunization: where discount rates are high, interventions with payoffs well into the future become relatively less attractive, and age of the patient becomes a less significant determinant of cost-effectiveness Source: Author. 10 Dean T. Jamison Figure 1-2. Increasing Costs per Disability-Adjusted therapy for sputum positives, by contrast, although costly, Life-Year Associated with More Complete Control will virtually never be applied when unneeded; it is highly of Dengue specific. Targeting BCC or other interventions to populations at highest risk, at least until full coverage can be afforded, will Cost per year (millions of U.S. dollars) maximize cost-effectiveness while simultaneously advancing 3.5 equity objectives (Mosley and Jolly 1987). As an illustration, figure 1-3 (from Murray, Styblo, and Rouillon, chapter 11, this 3.0 - C = Policy of improved case management EC collection) shows how BCG cost-effectiveness improves with VC = Above, plus chemical vector control 2.5 - EC = Above, plus environmental vector control rising risk of infection (and, hence, rising intervention speci- 2.0 - ficity), whereas chemotherapy remains of essentially constant attractiveness. 1.5 - In addition, targeting costs and compliance costs can dilute 1.0 / cost-effectiveness. Treatment can be very cost-effective for 0.5 - self-referred compliant patients ("Tendency to compliance" . e and "Tendency to self-refer" in table 1-4); as compliance 0.0 becomes more problematic, or targeting more costly, cost- 200 400 600 800 effectiveness decreases. For example, oral rehydration therapy DALY s saved per year (ORT) in the hospital or clinic setting is highly cost-effective; (in population of 1 million) it will only be used for severe cases of diarrhea, and it is likely to be applied effectively by qualified medical personnel. When Source: Chapter 14. ORT is taken to the community, however, cost-effectiveness declines substantially, both because of a decrease in interven- tion specificity (mild cases will be treated unnecessarily) and bulky and the chapter authors have simply been asked to do because home treatment will be applied less effectively than the best they could while noting where joint costs would need hospital treatment in severe cases. Similarly, targeting costs to be considered in country-specific applications. can be decreased if an immunization program to prevent neo- In this collection, then, for comparison across interventions natal tetanus shifts from trying to reach pregnant women to we use the common denominator of dollar cost per DALY immunizing all childbearing women, although there will be a gained, with the understanding that intervention costs and loss of specificity (at least with respect to preventing neonatal cost-effectiveness will likely vary across locales (even after tetanus but not, presumably, with respect to adult tetanus). controlling for intervention quality) because of differences in individuals, in epidemiological conditions, in delivery system characteristics, and in the degree of penetration of the inter- Figure 1-3. Cost-Effectiveness of BCG Immunization vention into the population. Table 1-4 lists many important and Tuberculosis Case Treatment as Function of factors that lead to variation in cost-effectiveness, and, to the Annual Risk of Infection extent that interventions are first applied where their cost-ef- fectiveness is highest, these factors collectively will lead to rising costs per DALY with increased application of an interven- Cost per death averted (U.S. dollars) tion.": Figure 1-2 illustrates this for control of dengue; up to a 2,500 point, improved case management is most cost-effective, but beyond that point, if a higher level of control for dengue is to 2 be sought, chemical and then environmental strategies of 2,000 vector control must be introduced. This phenomenon of rising costs per DALY comes up im- 1,500 - plicitly in many of the chapters; the cause of the phenomenon BCG is, frequently, the lack of intervention specificity and, also 0 \ frequently, costly targeting and compliance problems. Inter- ,0 vention specificity refers to what fraction of intervention recip- ients would benefit assuming that the intervention is applied 500 - exactly to the individuals to whom it should be applied (the factors "Prevalence of condition," "Incidence of condition," Case treatment and "Levels of risk factors" in table 1-4). Take BCG (bacille 0.5 1 1.5 2 2.5 3 3.5 Calmette-Guerin) vaccination as an example; it should be applied to all newboms, but it is a benefit, ex post, only to that tiny fraction of children who would have died in childhood from miliary tuberculosis (TB) without it. Tuberculosis chemo- Source: Chapter 1. Disease Control Prionties in Developing Countries: An Overview 11 When an intervention requires large fixed costs, total pro- Figure 1-4. Reaching Optimum Intervention Levels gram costs need to be weighed against total effects; simple assessment of marginal cost and effectiveness fails to suffice. Cost per DHLY (U.S. Dollars) The fixed costs involved in, to take several examples, investing 100 in major facilities, mounting a media-based health education program, or devising regulations and procedures can be sub- 80 - stantial. Fixed costs need not be financial; managerial or / political attention to a problem may have an important fixed 60 I cost element. When fixed cost may be important, understand- 40 ing the total burden of disease is necessary for estimating 40 potential total intervention effects (Mooney and Creese, ap- 20 I pendix C, this collection). Intervention A These points are relatively obvious, but there is often an 0 optimistic bias toward assessing cost-effectiveness under as- 1,000 2,000 3,000 4,000 5,000 sumptions of favorable targeting and compliance costs and of Annual expenditure on each intervention favorable intervention specificity. One might expect, as pre- viously noted, rising marginal costs and decreasing marginal effectiveness as interventions are extended through popula- Source: Author. tions; these combine to dilute cost-effectiveness. Thus favor- able case cost-effectiveness estimates can be real, but their cation, then, for interventions, depends on both economic and margin of applicability may be limited. In principle, it is epidemiological information because of the strong effect of desirable to acquire some sense of the responsiveness of inter- epidemiology on the rate of increase in marginal costs. vention cost-effectiveness to a range of parameters, particu- (Resource allocation to research, by contrast, should be driven larly the extent of application of the intervention. In practice, much more by the epidemiological significance of diseases and sensitivity analysis is sometimes possible but often difficult- researchers instincts about where advances might be realized.) and comparisons are then made for "representative" estimates of marginal cost-effectiveness to provide general guidance to The Findings: Public Health Interventions decisionmakers. When there are great differences in the mar- ginal cost-effectiveness of different interventions-as this In this section I summarize the findings of the twenty-five chapter concludes there to be-this "general guidance" can condition-specific chapters of this collection. In the first sec- suggest important redirections of policy.} ' Figure 1-4 illustrates tion I noted our conclusion that dividing interventions into how differences in marginal costs per DALY across interventions two broad categories-public health and clinical-was most can lead to inefficiency. In that example, intervention B is conducive to discussing policy tradeoffs, and these summariz- assumed to have a constant cost per DALY of $60 for the ing remarks are so divided. (Table 1-2 defined what is included range of expenditure levels considered (up to $4,000); inter- in each of these categories.) This section deals with public vention A starts with a lower cost per DALY ($20) but one that health interventions, and the following section deals with is rising to $100 per DALY at an expenditure of $4,000. The clinical ones. point here is that when, at prevailing levels of expenditure on each intervention, cost per DALY is lower for one of them (as Five Strategies of Public Health Intervention it would be for A at an expenditure of $1,000 on each of them), reallocation of money will increase output without Public health interventions are organized into five separate increasing cost. Hence, the previous allocation would have strategies in this collection-those designed to change per- been inefficient. sonal behavior, to control environmental hazards, to immun- Finally, it is worth reiterating that the assessments to be ize, to provide mass chemoprophylaxis, and to establish summarized in this chapter provide only an ordering of inter- mechanisms for screening and referral. Appendix table IA-1 ventions based on estimates of current marginal cost-effective- lists various components of each of these strategies and pro- ness; epidemiological information is required to assess how vides examples of the condition(s) they might effectively much of each intervention needs to be acquired in light of address; this summary of the review's findings on public health rising costs per DALY gained (Prost and jancloes, appendix D, intervention will, then, explicitly indicate which of these this collection). For example, one of the most cost-effective strategies is being assessed. I feel that in reviewing health interventions (for adults) is screening hospital blood supplies policies, or intervention alternatives, it will often be useful to for human immunodeficiency virus (HIV) seropositivity; as do so within each of these five broad strategies because of cost-effective as this may be, relatively few deaths can be commonalities of logistics, policy instruments, and approaches averted by it. Chemotherapy for patients with tuberculosis, within each. (This is true despite the frequently great diversity conversely, although somewhat less cost-effective, could be of conditions to be addressed within any one intervention expected to save hundreds of times more lives. Resource allo- strategy.) 12 Dean T. Jamison Before tuming to the condition-by-condition summary of tive of intervention-primary prevention, secondary preven- findings, I should touch on the issue of joint costs (and multiple tion, cure, rehabilitation, or palliation (as defined in table 1-2). outcomes) of interventions in light of conclusions from the The first column states findings on cost-effectiveness and, individual chapters. The analysis upon which this collection where applicable, notes the most important factors (from is based is structured by diseases (or adverse health conditions table 1-4) likely to lead to variability in cost-effectiveness more generally), and the issues addressed in the individual across circumstances. chapters thus concern the nature, cost, and effectiveness of the Although the entries in appendix table IA-3 are self- interventions available for dealing with each condition. In explanatory, a few general observations on each primary cate- many cases, of course, any given intervention will address gory of intervention may be worth making. multiple conditions and, indeed, may well have important effects outside the health sector altogether. Perhaps the clear- PERSONAL BEHAVIOR CHANGE. Some personal behavior est examples are control of smoking, breastfeeding, and envi- changes that are favorable for health outcomes tend to occur ronmental improvements. Limitation of smoking markedly naturally as incomes rise; these include, at least for many reduces risk for lung cancer, ischemic heart disease, and cultures, improved hygienic behaviors, increased energy in- chronic obstructive pulmonary disease; outside the health take and quality in the diet, and decreased crowding. Im- sector it reduces (at least to some extent) property damage from provements in these behaviors are typically important for the fire and frees productive resources for alternative use. pre-epidemiological transition diseases and, as entries in ap- Breastfeeding, likewise, has multiple health effects; it en- pendix table 1A-3 indicate, can often be affected by educa- hances child immunity, reduces exposure to infection, pro- tional interventions even though the main force driving vides balanced nutrition and, by suppressiig ovulation, improvements-income increases-is beyond the domain of postpones the next pregnancy (Anderson 1990). The cost of health policy. breastfeeding, however, includes, as do many health-promo- Other behaviors are likely either to be less dependent on ting interventions, substantial amounts of mothers' time- income levels (for example, breastfeeding behavior, sexual which is not easily valued in terms, say, of wages forgone (Leslie practices) or to be adversely influenced by income increases, 1992). Finally, whereas environmental interventions have at least for a period of time (for example, dietary excess, beneficial health consequences, their main objectives often lie sedentary lifestyles, smoking, alcohol consumption). Most of outside the health sector; World Bank (1992) provides a recent these are risk behaviors for posttransition conditions. Al- comprehensive discussion. though the natural course of development is unlikely to im- Appendix table IA-2 lists a number of public health inter- prove these behaviors, there is more of a scope for affordable ventions that have a range of outcomes; in country-specific government policy to influence them. Regulatory policies and, applications, assessment of the cost-effectiveness of these in- particularly, taxation policies for tobacco, alcohol, and fatty terventions should, ideally, quantitatively aggregate interven- meats show great promise for inducing behavioral change and, tion effects along these multiple dimensions of outcome. currently, are very much underused. Education of elites and Likewise for clinical intervention there will frequently be joint the public are complementary instruments, not least because costs (associated, for one example, with the availability of they generate the political will and popular support for regula- diagnostic facilities in a district hospital); again, in country- tion and taxation. The extremely high cost-effectiveness of specific application, these matters need to be assessed more smoking control makes it, perhaps, the top priority for govem- quantitatively than they can be in a general overview. mental action (Stanley, appendix A, this collection); although Authors of the disease-specific chapters of this collection less well documented, the probably high cost-effectiveness of have generally noted where interventions for the condition alcohol control makes it another priority. they were addressing had multiple outcomes, and in this chap- ter, I too note the most important cases. It is clear, looking ENVIRONMENTAL HAZARDS CONTROL. Rising incomes help across findings of the individual chapters in this collection, with improving water supply and sanitation and that is likely that multiple effect and joint cost problems do complicate the to be important in prevention of a broad range of infectious task of assessing cost-effectiveness in many important in- and parasitic diseases. Vector control is at least marginally stances; that said, it is more generally true that these problems cost-effective for a number of conditions (malaria, onchocer- are relatively minor or can be dealt with by reasonable approx- ciasis, dengue) in some environments. Industrialization intro- imations and simplifications in the analysis. duces new hazards into the environment (lead, mercury, and the like) that can produce severe lifetime disability if not Findings effectively controlled. Improvements in household ventila- tion, indoor fireplaces, and cookstoves can substantially reduce Appendix table IA-3 presents, in very summary form, the risks for chronic obstructive pulmonary disease (COPD); and findings from the chapters in this collection concerning public occupational and transport safety measures are important in health interventions. For each disease category the first col- many specific instances. In principle, protective measures can umn of appendix table 1A-3 indicates relevant intervention be delivered through environmental intervention, and water strategies, and the second indicates, to be explicit, the objec- fluoridation for prevention of caries is one example. Not Disease Control Priorities in Developing Countnes: An Overtiew 1 3 explicitly mentioned in appendix table IA-3 is the problem of These comments only touch on some of the findings sum- lead toxicity resulting from excess use of lead-based paints and marized in appendix table 1A-3 to give a sense of their range combustion of gasoline with high lead content. Recent re- and diversity. Next is a discussion of clinical interventions. search-reviewed in Pollitt (1990)-indicates that lead tox- icity may be far more important than previously thought as The Findings: Clinical Interventions a determinant of slow development and impaired mental functioning. Facilities to provide clinical intervention vary continuously in IMMUNIZATION, MASS CHEMOPROPHYLAXIS. AND SCREENING. size, in the degree of complexity (and range) of the conditions ThMMintervAntionslis apASSdix tHEMOP LaXble 1AN3 unertheNG that they address, in the sophistication of their facilities and The interventions listed on appendix table 1A-3 under the eqimn,adnthtrnngndsllotersaf.2Te headings Immunization, Mass Chemoprophylaxis, and equipment, and in the training and skill of their staff.'2 The Screening all share certain common characteristics (a) they authors have found it useful, nonetheless, to use generally involve the direct administration or application of a specific accepted terminology in categorizing facilities into three technical intervention to individuals on a one-by-one basis groups-clinic-level, districthospitals, and referralhospitals- techica intrvetio to ndiiduls o a ne-y-on bais, while recognizing that that categorization involves much sim- (b) they are directed to certain target populations, and (c) the wlerCgiigta htctgrztO novsmc i coverage of the target population is important to produce the plification and that the appropriate classification structure will covesiragedfetT of these targeertion isstmportanttoproducate- vary substantially from country to country. Appendix desired effect. Technically, each of these intervention strate- table IA-4 indicates (in a very general way), for each of these gies is highly efficacious when correctly applied to a compliant three levels of facility, examples of the kinds of interven- subject, but their actual effectiveness in developing-country th ey m it exand of ty s faity settings is strongly conditioned by the local administrative, tions they might address and what capacity such a facility managerial, and logistical capabilities, as well as by traditional might have for primary modes of diagnostic and therapeutic cultural constraints. intervention. cultural constraints. ~ ~ ~ ~ ~ - Each chapter of this collection addresses the desirability not Although it is no surprise, the all-pervasive potential of publchalth intertionalsoeofeclinicalrintertyn- immunization programs is dramatically underlined in the find- only of public health intervention but also of clinical interven- ings set forth in appendix table IA-3.Most immunization in- tions that might be mounted at various levels of the reterral tervenrions are highly cost-effective; and many of them address system. One lesson that emerges from these chapters is that highly prevalent conditions. Measles and tetanus vaccination currently such analyses are severely constrained by the paucity highl prealen conitios. Masle andtetaus vccintion of data relating to the effect and cost of clinical interventions. appear particularly cost-effective and worthy of relatively In the ofsch analyse it isprap nturalnfor greater attention within immunization programs. Far more In the absence of such analyses, it is perhaps natural for could be efficiently spent on immunization than is now being developing countries to import, to the extent that resources spent; and, even though costs of delivery tend to rise as more permit, the methods of case management used or being deel- marginal populations are reached, extending immunization oped in high-income countries. The key phrase here is, of programs to virtually universal coverage is likely to prove both course, "to the extent that resources permit." Available re- cost-effective and a practical way of significantly improving sources permit importation of high-cost interventions for only the health of the poor. a tiny proportion of a developing country's population." In One particularly promising application of mass chemopro- order to extend access to services for the rapidly emerging phylaxis lies in the administration of anthelmintic medication epidemic of acquired immunodeficiency syndrome (AIDS) as and micronutrient supplements to school-age children. Here well as for the impending epidemic of noncommunicable dis- cost-effectiveness appears quite high for conditions that, al- ease, radically lower cost methods of case management will though of extremely high prevalence, have only recently been need to be developed from the rich range of technologies and seen to be of substantial importance for intellectual and phys- procedures that now exist, or that are coming into being. see tobe f sbstntil iporanc fo inellctul adpys- Appendix table IA-5 summarizes data from the individual ical development. A program of chemoprophylaxis for school- Aptendix the IA-effectizes o f cl in dividual age chldrencould,like he Exanded rograme on chapters on the cost-effectiveness of clinical intervention, by age children could, like the Expanded Programme on disease, subject to the caveat that potentially remediable data Immunization (EPI) for younger children, be expected to serve deficiencies have frequently left a high margin of uncertainty. as the starting point for an ultimately much expanded capacity The table includes discussion, for each condition, of the strat- to deal with the health needs of this age group; the Rockefeller e (le faclitv di mode f intervention) or treastrat- Foundation and the UNNDP are jointly initiating such a program. egy (level of faclity and mode of itervention) for treating it Perhaps the only significant cancers for which treatment is indicates the objective of intervention, and summarizes find- cost-effective (breast, cervical) are ones for which early screen- ings on cost-efectiveness. ing and referral are important; so, as NCDs begin to emerge, this Several observations stand out from appendix table 1A-5: strategy will become increasingly relevant. The emerging strat- * Curative care for tuberculosis and sexually transmitted egies for treatment of acute respiratory infections in children diseases appears extremely cost-effective; further, such care all rely heavily on community-based programs for early detec- is not now being provided to anything like the extent it tion and quick referral; with increased experience, im- should be, given the high burden of morbidity and mortality provements in capacity for cost-effective screening and referral resulting from these conditions. Surgical treatment of cata- programs can be expected to develop. ract is also highly cost-effective. 14 Dean T. Jamison * The extremely diverse range of clinical interventions systematically to assess intervention cost-effectiveness in the of moderate cost-effectiveness (medical management of health sector. The effort required is substantial, but results that angina or diabetes are examples as is surgical management allow broad intrasectoral assessment of intervention priorities of cervical cancer) suggests that country-specific analyses of can he obtained. these conditions are required and that facilities capable One substantive conclusion is that the available evidence of competently handling diverse conditions will need to be points to great variation, across interventions, in marginal developed. cost-effectiveness.'4 Appendix table IA-6 summarizes this ev- * The cost is sufficiently high for some clinical interven- idence by grouping interventions into ranges of marginal cost tions to imply that even if they are effective (as is the case per DALY in hypothetical (but realistic) environments. The with coronary artery bypass grafting to deal with angina), challenge ahead is that of designing and implementing instru- their cost-effectiveness is so poor that their use should be ments of govemment policy that will greatly expand use of the actively discouraged until other, more cost-effective inter- interventions in the first two sections of appendix table I A-6 ventions can be delivered to their appropriate potential. while decreasing use of interventions, like those in the last * Control of pain from terminal cancer could benefit per- section of the table, that provide very little value for money. haps 1.5 million individuals annually at acceptable costs; Garber and Phelps (1992) observe, though, that under a current legislation and standard practices greatly limit what reasonable range of assumptions it will make economic sense is done in relation to what potentially could be done. to pay for DALYs up to a cost of about twice the level of income; Rehabilitation (in particular from leprsy,poliomyelitis, this leads to a second substantive conclusion from appendix and injury) shows promise of being extremely cost-effective; table IA-6, which is that, in many countries, quite a broad bteyid habilitaion array of interventions is likely to prove attractive by any but very little attention has been accorded reasonable economic standard. and little is known about how best to provide services on a My third substantive conclusion concerns the extent to population basis or what might be expected in terms of which public health as opposed to clinical strategies tend to be effectiveness and cost. more cost-effective and the extent to which seeking primary Again, as with the discussion of public health interventions, preventive objectives will tend to be more cost-effective than onethemethatemergesfromthisreviewofclinicalcost-effec- seeking other objectives. Appendix tables IA-7 and 1A-8 tiveness is that of complexity and diversity. Many interven- summarize the materialfromappendix table IA-6 in ways that tions are clearly not cost-effective, and public policy should allow these questions to be addressed. Although there are some make every effort to discourage their use. But the available patterns (in particular, primary prevention by way of im- evidence does suggest that a broad range of interventions, munization accounts for many highly cost-effective interven- addressing a similarly broad range of conditions, will prove tions), in general I conclude that there is no especially strong cost-effective. Many of these interventions are not now being general tendency for primary prevention or public health used to anything like the extent that they should be. Likewise, interventions to have superior cost-effectiveness. much of what is currently undertaken by the clinical system is The fourth substantive conclusion is that virtually no cost- misdirected (toward interventions of low cost-effectiveness) or effective interventions require more specialized facilities than simplv inefficiently used. Redirection of substantial resources those available at district hospitals. Thus, even though one from interventions of low cost-effectiveness toward those with cannot argue in general in favor of prevention over cure or very high cost-effectiveness is clearly possible; a central task of public health over clinical intervention, one can conclude that health policy must be to design implementation strategies and district hospitals and lower level facilities potentially offer government policy instruments that can promote these poten- almost all attractive interventions. tial efficiency gains. Conclusions Appendix 1A: Clinical and Public Health Interventions I draw five very broad conclusions-one methodological and the other four substantive-from this collection reporting on The eight tables in this appendix examine clinical and public the World Bank's Health Sector Priorities Review. The meth- health interventions and relate the characteristics of the dif- odological conclusion is that it is feasible, on a broad scale, ferent interventions to their cost-effectiveness. Disease Control Priorities in Developing Countries: An Overview 15 Table IA-I. Public Health Interventions Intervention strategy Adverse health outcomes avoided or treated Promoting healthy behavior Dietary practices Macernal Low birth weight Infant/child Growth-stunting infection, micronutrient deficiency Adult Obesity, diabetes, cancer, ischemic heart disease, hypertensive disease Prevention against infection Personal hygiene Diarrheal diseases, intestinal parasites, skin infections Use of soap Food handling Defecation practices Household ventilation/crowding Respiratory diseases Sexual behaviors Sexually transmitted diseases, premature fertility Personal health practices Exercise Obesity, heart disease Stress control Mental disease Carelessness Injuries, poisoning Tobacco use Cancer, heart disease, chronic obstructive pulmonary disease Controlling substance abuse Alcohol Hypertensive disease, cirrhosis, injuries Drugs Addiction, injuries, mental disease Preventing intentional mutilation Injuries Reproductive practices Contraceptive practice Unwanted pregnancy Pregnancy care Pregnancy outcome Childbirth practice Perinatal/maternal mortality Breastfeeding practice Malnutrition, diarrhea Abortion Matemal mortality, unwanted pregnancy Home care Use of first aid Injury Use of oral rehydration therapy Diarrheal disease, acute diabetic conditions Clinic use Treatment of simple conditions Entry into referral structure Control of environmental hazards Household Water availability/quality Water-related diseases Waste disposal Infectious diseases, toxic exposures Food hygiene Diarrhea, parasites Air quality Respiratory disease Vector control Malaria, yellow fever, river blindness Community Housing quality Injury, poisoning Motor vehicle/road safety Accidents, injuries Occupational hazard control Injury, toxic exposure Public health services Immunization At birth Hepatitis B, tuberculosis First year of life Polio, measles, diphtheria, tetanus, perrussis School age Tetanus Adulthood Tetanus, yellow fever, influenza Mass chernoprophylaxis Food fortification Iodine and other deficiencies Micronutrient supplenientation Vitamin A and other deficiencies Water fluoridation Caries School-based provisi oni of anthelmintics Schistosomniasis, intestinal helminths Mass administration of antibiotics Sexually transmitted diseases Screening and referral Screening prOgrams in clinics, schools, Selected infectious diseases (such as tuberculosis); cancers (such as cervix, breast); cardiovascular work sites, and so on disease risk factors (such as hypertension, hyperlipidemia); high-nsk pregnancy Source: Author. 16 Dean T. Jamison Table IA-2. Selected Interventions with Multiple Outcomes Outcome Intervention Main health outcome Secondary health outcorne Nonhealth outcomes Provision of water supplies Control of diarrheal diseases Control of skin, respiratory, and Saving of household time; and sanitation helminthic infections welfare improvements Provision of soap Control of diarrheal diseases Control of skin, respiratory, and helminthic infections Reduction of vehicle speed Reduced severity and incidence Reduction in property damage limits of crash-related injuries from vehicle crashes; energy conservation; time costs Control of smoking Reduced incidence of lung Reduced incidence of minor Welfare loss for current addicts, cancer, heart disease, and cancers; reduction in bum welfare gain for nonsmokers; chronic obstructive pulmonary injuries freeing of land and labor for uses disease other than tobacco production Vector control Reduced incidence of vector- Improved welfare when vectors, bome diseases such as mosquitoes, are nuisances Female education Reduced child mortality rates Improved child growth; reduced Higher levels of female adult health productivity and eamings; improved congruence between actual and desired fertility levels Breastfeeding Improved child growth through Protection of child against Savings in costs of infant improved nutrient availability infectious disease; postponement formula and bottles; time costs and protection against diarrhea of next pregnancy; possible long- for mother term cognitive benefits to child Family planning services Reduced child mortality Reduced matemal morbidity and Economic and welfare gains mortality from improved control of level and timing of fertility Source: Author. Table 1A-3. Public Health Interventions: Cost and Effectiveness Condition Intervention strategv Objecatve Cost and effecatveness Comments Acute respiratory Screeninig and referral: households need to Cure Costs per disability-adjusted life-year Has variable efficacy in all age groups. infections (ARI) be educated to identify signs of (DALY) $20 in high-mortality See ARI case management chapter for (see chap. 4) pneumonia in children and bring them to environments to $50 in low-mortality details of program clinics for treatment environments Behavior change: breastfeeding promotion Secondary prevention (via strengthening Approximately $50 per DALY (for ARI Multiple benefits, including averted via education programs of child to reduce effects of infection) consequences of breastfeeding only) infant mortality Reduced protein-energy malnutrition via Secondary prevention (via strengtheninlg Estimated cost per DALY about $65 Depending upon food availability supplementation programs of child to reduce effects of infection) Immunization: for pertLssis, see discussion of poliomyelitis; Pneumococcal vaccine Primary prevention For appropriate age groups (> 18 mo.), Has variable efficacy for all age groups For measles, see that entry cost per DALY saved about $70 Hemophilus influenzae vaccine Primary prevention No estimates available for cost- Efficacy trials in early stages effectiveness; vaccine costs high, $2 to $14 per dose. Moderate efficacy in children over two years Diarrheal diseases Possible immunizations: using effective Primaryprevention Approximately $10 per DALY, assuming Effective vaccine still not available (see chap. 5) rotavirus vaccine 80 percent vaccine efficacy Using effective cholera vaccine Primary prevention Approximately $75 per DALY, assuming Current vaccines have low efficacies 70 percent vaccine efficacy Immunizatio7l: measles vaccine Primary prevention Approximately $10 per DALY, assuming Rotavirus and measles-associated diarrhea 85 percent vaccine efficacy is much more common than cholera, explaining much of the cost-effectiveness differences Control of environment: improving water Primary prevention Cost-effectiveness of diarrhea averted Estimated to reduce diarrhea morbidity supply and sanitation by upgrading of is unknown, but there are additional and mortality by about 30 percent infrastructure benefits Behavior change: improved domestic and Primary prevention Approximately $170 per DALY, depending personal hygiene via education on case-fatality reductions, incidence rates, and wage levels Breastfeeding promotion by various Primary prevention Approximately $30 per DALY, assuming a See above methods, including changes in hospital reduction in non-breastfeeding of 40 routine and mass-media education percent ( < 2 months), 30 percent (3-5 months) and 10 percent (6-11 months) and assuming a judicious selection of interventions used Improving weaning practices by education Secondary prevention Approximately $30 per DALY, in children See above < 75 percent weight-for-age, age six months to five years (Table continues Oni the following page.) Table IA-3 (continued) Condition Intervention strategy Objective Cost and effectiveness Comments Poliomyelitis (see Immunization: oral or injectable polio Primary prevention; established a global For polio plus DPIT, cost per DALY If injectable used, there could he a chap. 6) vaccine in three or more doses to children objective: eradication of disease from wild approximately $20 in high-mortality reduced number of needed contact raising under one year of age; given polio virus by the year 2000 environments and $40 in low-mortality immunizationi rates and costs per fully simultaneously with diphtheria, pertussis, environments; cost per DALY of DPT vaccinated child in some environments and tetanus DFr immunization withouit polio vaccine somewhat higher and cost of polio without DPIT imimiunization many times higher Helininthic Targeted mass chemotherapy: school-based Secondary prevention Ranging from $6 to $33 per DALY; low Cost-effectiveness figures are infection (see delivery of anthelmintics (praziquantel for estimate assumes either 0.02 or 0.2 DALYs hypothetical, using range of program costs chap. 7) schistosomes and albendazole for saved yearly and high estimate assumes and empirical estimates of efficacy intestinal helmintics in high-endemicity either 0.05 or 0.5 DALYs saved yearly. areas without individual screening) Approximately 80 percent effectiveness and heavily reliant on intensity of infection Targeted screening and treatment: Secondary prevention Depending upon intensity of infection Generally two-thirds as effective at same praziquantel and albendazole given after cost as mass or targeted chemoprophylaxis screening in targeted groups Control of enVironment: onchocerciasis Primaryprevention $100-$200 per DALY saved, and closely Variable efficacy in trials vector control with chemical pesticides tied with breeding pattems of vector Cloth filters for crustacean intermiediate Primary prevention Probable low cost Only enclosed water supplies will of Dracunculus (guinea worm) eradicate disease Measles (see Immunization: four possible scenarios: Primary prevention via individual With use of the current measles antigen Vaccination should begin earlier in life in chap. 8) * the current antigen at nine months protection; with high vaccination at the recommended nine months of age, areas with high measles mortality. This of age coverage rates, some protection resulting base cost per DALY gained $2 to $15, change in vaccination schedule would * the above plus coverage at all from intemiption of transmission depending on case-fatality rates and cost alter cost-effective figures "opportunities" of the measles portion of the program. * the E-Z antigen at six months Epidemiologic model to assess * the E-Z antigen at six months effectiveness of altemative antigens and and nine months two-dose schedules demonstrates that significantly more deaths could be averted over the base cost at costs per incremental DALY of twice the cost per DALY of the base case-still relatively attractive Tetanus (see Immunization: routine vaccination Primary prevention Approximately $2-10 per DALY gained, Choice of strategy and target group varies chap. 9) targeting womeni of childbearing age, and predominantly from averted neonatal according to incidence, available pregnant women, schoolgirls, infants, and tetanus mortality. Depends on incidence resources, health service organization, males rates, which in turm are tied with health channels for contact, immediacy of infrastructure, especially birth practices desired impact, and so on Behavior change: training birth attendants Primary prevention Suspected to be higher than costs for Training, supervision, and support costs routine vaccination of pregnant women may be too high for much of the developing world Rheumatic heart Immunization: to prevent incidence of Primary prevention Possible low cost Vaccine under development; unlikely to disease (see rheumatic fever/rheumatic heart disease he available for many years chap. 10) (RF/RHD) precursor, group A streptococcus Screening and referral: for children with Primary prevention of RF Cost about $300 per DALY; high because May encounter resistant strains, pharyngitis refer to clinic for antibiotic proportion of pharyngitis cases that necessitating expensive antibiotics prophylaxis against RF develop into RF is low Tuberculosis (see Immunization: BCG added to DPT program Primary prevention $7 per DALY; cost-effectiveness drops Reported effectiveness of vaccine is chap. II) substantially when annual risk of widely variable. Costs of BCG program infection < I percent alone are four times those of adding it to preexisting EPI Targetedchemoprophylaxis: selective Primary and secondary prcvention Suspected to be reasonable Mass prophylaxis has high costs with screening in high-risk populations (AIDS limited effectiveness patients and family members of tuberculosis patients) and treatment of smear-positive patients Leprosy (see Immunization: tise of fC'G vaccine to Primary prevention Probably reasonable, as costs be shared Vaccine reported to be 30-80 percent chap. 12) prevent leprosy with tuberculosis immunization program. effective against leprosy, depending on Will depend on incidence rates age at vaccination and regions studied Targeted screeni11g: passive screening via Primary prevention $0.50 per DALY, sensitive to percentage of Does not include treatment costs (see clinical exam cases which are multibacillary or appendix table IA-5) paticibacillary (harder to diagnose clinically) and incidence rates Malaria (see Control of environment: chemical vector Primary prevention $5-$250 per DALY, depending on type of If population widespread, the marginal chap. 13) control via intradomicillary spraying of mosqtuito which is the primary costs of eradication are high. Cost- insecticide to kill adult mosquitoes determinant of case fatality. Also linked effectiveness will decrease if spraying is with incidence rates and geographical carried out in nonunifomi manner distribution of the human and mosquito population Environmental vector control via Primary prevention Suspected to he high Role limited to urban areas. Chemically drainage and land management impregnated bed netting with promising techniques results Dengue (see Control of environment: spraying of Primary prevention $2,200 per DALY tied to how quickly Has to be repeated several times per year chap. 14) chemical insecticide for Aedes mosquito Acdes replaces itself Direct expenditures and educationi for Primary preventioni $3,500 per DALY needs integrated Closely related to labor costs. Has climinating breeding sites of Aedes program to be effective potential of sustained long-term control mosquito or eradication Possible immunization: two-dose Primary prevention $1,600 per DALY heavily dependernt on Vaccine in planning stages vaccination will probably be needed incidence rates (Table continues on the following page.) Table IA-3 (continued) Condition Intervention strategy Objecive Cost and effectiveness Comments Hepatitis B (see Immunization: adding three dose vaccine Primary prevention $25-$50 per DALY depending on Discounting of benefits is importanit, chap. 15) to preexisting immunization program prevalence of carrier state and ability of since vaccine is given at birth and averted immunization program to provide mortality occLirs late in life. Cost of adequate coverage using a three-dose treating significant morbidity important scheduile (especially cirrhosis). Vaccine costs arc variable but have declined significantly for public sector programs. Maximum cost- effectiveness may he achieved only through integration into routine infant immunization programs Excess fertility Behavior change: increasing the use of Primary prevention $15-$75 per DALY will vary depending Only includes benefits derived from (see chap. 16) condoms via education and subsidization on number of births which exceed the averted mortality by increasing birth resources of a family or society and the interval and limiting teenage pregnancies. mortality associated with this excess. The Assumes theoretical 100 percent effectiveness of the program is an effectiveness essential component which will rely on personal acceptance of condoms Use of information, education, and Primary prevention Suspected to be of low to moderate cost Benefits from averted AR] and diarrhea communication model to lengthen birth and is dependent upon cultural biases cases in breastfed child are significant intervals by encouraging breastfeeding Matemal and See entry in appendix table IA-5 perinatal health (see chap. 17) Protein-energy Targeted mass chemoprophylaxis: food Secondary prevention $70 per DALY, based on relationship Depends on percentage of target malnutrition supplementation for preschool children between food supplementation and child population that are severely or (see chap. 18) resulting in a 100,000-calorie transfer growth for particular weight-for-age child moderately malnourished, since standard and 0.5-kg average weight gain food transfer will affect these groups differently Food supplementationi for pregnant Secondary prevention $25 DALY, based on relationship between See above women resulting in a 100,000-calorie food supplementation and fetal growth as transfer and a 300-gram increase in birth it affects infant mortality weight Micronutrient Targeted mass chemoprophylaxis: daily Secondary prevention $13 per DALY, dependinig on prevalence In regions of severe iron deficiency anemia deficiency self-administered oral iron rates and severity of iron deficiency disorders (see supplementation for duration anemia chap. 19) of pregnancy Use of injected or oral iodinated oil once Secondary prevention $20 per DALY varies with prevalence rates Clean syringes need to be used every two to five years in women of and severity of iodine dleficiency reproductive age Semiannual mass dose of vitamin A for Secondary prevenitioni $9 per DALY varies with prevalence rates Benefits include averted measles, children age zero to five and severity of vitamin A deficiency respiratory and diarrheal mortality Mass chemoprophylaxis: fortification of salt Secondary prevention $5 per DALY varies with prevalence rates As with all food fortification programs, or sugar supplies with iron-containing and severity of iron-deficiency anemia the carrier must be available and accepted compounds by the groups at risk lodization of salt or water Secondary prevention $8 per DALY, depending on prevalence See above rates and severity of iodine deficiency Fortification of sugar with vitamin A Secondary prevention $5 per DALY, depcnding on prevalence See above compounds rates and severity of vitamin A deficiency HIV infection and Screening and referral: blood screening for Primary prevention $1-$250 per DALY linked to the Cost of blood test relies on sophistication other sexually HIV among blood donors using rapid lab prevalence rate of HIV infection among of health care system transmitted tests blood donors and sexual activity group of diseases (see proposed blood recipient chap. 20) Behavior change: infoirmation, education, Primary prevention $1-$150 per DALY depends on sexual Social barriers to condom usage are often and communication program to: (a) activity and prevalence of each type of important. Practical problems include the decrease frequency of sexual partner sexually transmitted disease in target identification of the "access group" by changc, and (b) increase proportion of population (i.e., HIV transmission rates in which high-risk groups can be identified sex acts protected by condoms population with syphilis greater than in population with chancroid) Female education and employment: Primary prevention Likely to be highly effective in the long Cost-effectiveness difficult to compute reduce supply of female sex workers by run and may yield many benefits other raising their opportunity cost and reduce than disease control demand for their services by attracting more women to urban areas Cancers (see Screeningand referral: use of PAP smear at Secondary prevention $100 per DALY, based on prevalence rates Assumes that there will be an appropriate chap. 21) five-year intervals to screen for cervical and health care setting in which program referral system for further treatment if cancer is carried out indicated by PAP smear Annual breast examinations to screen Secondary prevention $50 per DALY based on prevalence rates Same as above and usc of annual for breast cancer and health care setting in whicih program mammography at one-year intervals for is carried out women fifty-nine and above reduces cost- effectiveness tenfold Behavior change:smoking cessation classes Primary prevention $20 per DALY tied to percentage of Less expensive to prevent onset of antismoking effort aimed at preventing smoking than to have smokers quit "new starters" or those already smoking Smoking reduction via tobacco tax Primary prevention See COPD entry in this table See CoPD entry in this table Diabetes (see Behavior change: health education to Primary prevention of NIDDM Costs might be $0.02 to $0.50 per capita NIDDM patients often have limited chap. 22) improve dietary and exercise habits may per year; effectiveness unknown but modification program rates hold potential for prevention of non- probably low. Cost-effectiveness insulin-dependent diabetes (NIDDM) enhanced by similarity of the behavior changes for reducing cardiovascular and some cancer risks Screening and referral: screening for Primary and secondary prevention Unknown; cost-effectiveness depends on High costs associated with complication glucose intolerance in high-risk groups of NIDDM prevalence in the screened group and cost- (stroke, coronary artery disease, (such as obese, pregnant women) may effectiveness of referred interventions ketoacidosis, and coma) allow for more precise targeting of health education and perhaps medication (Table continues on thl Jullowing page.) Table IA-3 (continued) Condition Intervention strategy Objective Cost and effectiveness Comments Cardiovascular Behavior change: through a public Primary and secondary prevention Costs perhaps less than $1 per capita per Effectiveness will depend on depth of disease (see prevention package (mass education and year in targeted population, about $150 impact. chap. 23) individual counseling); screening and per DALY referral services to those at high risk Chronic Behavior change: smoking reduction via Primary prevention 20 percent tax on tobacco might reduce Heavy smokers (those with greatest risk) obstnictive tobacco tax and education/cessation overall tobacco consumption by about 20 may be unresponsive to program pulmonary programs percent and avert perhaps 40 deaths per disease (see year in a typical population of 200,000 chap. 24) smokers. Adding educational programs, $20 per DALY can be achieved Injury (see Alcohol taxation to discourage use Primary prevention Low to moderate cost per DALY depends If public is strongly against an alcohol tax, chap. 25) resulting in 30 percent decrease in fall, on case-fatality rates and percentage of illicitly produced alcohol use may transportation, and bumn injuries injuries that are alcohol attributable increase, resulting in excess mortality Education programs to reduce Primary prevention Moderate cost per DALY depends on depth Injury education programs have limited transportation, bum, and poisoning of impact and case-fatality rates effectiveness without safety laws. Alcohol injuries by 40-50 percent awareness component often essential Control of environment: manufacture, Primary prevention High cost per DALY depends on case- Safer product will still have to be modification, and use of products such as fatality rates and percentage of injuries inexpensive and some improvements will seat belts, safer stoves, and childproof which are preventable by environmental require enormous fixed costs caps to cause a 50 percent reduction in improvements transportation, fall, bum, and poisoning injuries Cataract (see chap. Behavior change: ocular protection from Primary prevention Inexpensive depending upon extent to Probably beneficial, but to what extent is 26) solar radiation (e.g., hats and sunglasses) which cataract progression is retarded by unknown eye protection Oral health (see See appendix table IA-5 chap. 27) Schizophrenia and See appendix table IA-5 manic-depressivc illness (see chap. 28) Source: See chapters on individual diseases in this collection. Disease Control Priorities in Developing Countnes: An Overview 23 Table 1A-4. Clinical Intervention: Level of Facility and Mode of Intervention Intervention mode Therapeutic Typical conditions Physical or Level of clinical facility addressed Diagnostic Medical Surgical psychological therapy Clinic (private, Minor trauma; simple Clinical Short list of Sutures Important potential community, and injections; support of essential drugs role for supervisng school- and work- population-based (about 20) physical therapy based) interventions; uncomplicated childbirth; family planning District hospital Complicated childbirth Clinical; basic Long list of Capacity for dealing Design and fractures and bums; laboratory; basic essential drugs with abdominal management of more complicated infections; radiologic facilities (about 200) surgery, many complex regimens of cataract; hemia; fractures, cesarean physical and appendectomy; diabetes, sections, some psychological therapy hypertension, and rehabilitative surgery similarly complex condition Referral hospital More complicated More advanced As above, but As above but also Support capacity for medical and surgical laboratory and also specialized capacity for more district hospitals conditions radiologic facilities drugs, chemo- complicated surgery therapy, and of head and chest radiotherapy Source: Author. Table I A-5. Clinical Interventions: Cost and Effectiveness Condition Intervention strategp9 Objective Cost and effectiveness Comments Acute respiratory Clinic level: antibiotic treatment of Cure Costs per disability-adjusted life-year Resistance to "first line" antibiotics will infections (see pneumonia in young children (for whom (DALY) saved is $20 in high-mortality necessitate use of expensive antibiotics. chap. 4) case-fatality rates high). To succeed, to $50 in low-mortality environments Costs of allergic reaction to medication control programs must educate families to could be substantial. bring children with cough or difficulty breathing to a facility quickly Diarrheal diseases Clinic level: education and distribution Secondary prevention Ranging from $35-$350 per DALY, Assuming cost per diarrhea of $1.00-$5.00 (see chap. 5) of ORT sugar-salt solution depending on the case-fatality rate of and 0.05%-0.5% of deaths prevented per target population and the cost of labor case treated Use of antibiotics and antimotility agents Curative Very poor cost-effectiveness for most May be indicated for dysentery or cholera diarrhea cases for which they are either harmful or useless Poliomyelitis Clinic level: physiotherapy, psychotherapy, Rehabilitation Potential high cost per DALY Needs constant source of skilled labor (see chap. 6) provision of simple prostheses to enhance physical function and promote social integration Dist-ict and referral hospital levels: surgery Rehabilitation Probable moderate to low cost per DALY Developed health care system and of varying degrees of complexity expensive specialists when needed Helminthic See entry in appendix table 1A-3 infection (see chap. 7) Measles Clinic level: therapeutic doses of vitamin A Secondary prevention Limited evidence that case-fatality rates Proportion of measles morbidity and for children with severe measles; other can be reduced by 50 percent or more from mortality which is preventable via therapy (including use of antibiotics) to their initial levels of 0.01-0.05. If such supplementation is still being investigated further reduce adverse consequences therapy costs $10, cost per DALY of therapy would be $20-$80 Tetanus (see Referral hospital level: case management Cure $100 per DALY and is tied to case-fatality Needs moderately sophisticated health care chap. 9) care including neurorespiratory reduction system, wide range in protocol costs. Few resuscitation; antispasmodic therapy; neonatal tetanus patients are brought for antitoxin drugs, wound care, and intensive medical care. nursing Rheumatic heart Clinical level: regular administration of Secondary prevention $100-$200 per DALY in compliant Compliance is an issue disease (see antibiotics at three- to four-week intervals patients chap. 10) prevents recurrence in patients who have had rheumatic fever (RF), later bouts of RF, hence emergence of rheumatic heart disease Referral hospital level: open-heart surgery in Secondary prevention and rehabilitation $1,000-$2,000 per DALY, depending on Needs advanced health care system higher-level referral hospitals to permit age of patient and local surgical costs restoration of function in mitral or other valves (valvuloplasty) Tuberculosis (see Clinic level: short-course chemotherapy Cure $3 per DALY linked to hospitalization costs. Resistant strains could become a significant chap. II) with two-month hospitalization Primary prevention Standard chemotherapy has lower drug problem (especially among AIDS patients) costs, but a lesser core rate compared with causing an increase in costs short-course chemotherapy (with similar hospitalization periods) Leprosy (see Clinic level: multidnig therapy with Cure $7 per DALY will depend on percentage of Compliance with daily oral medication chap. 12) monthly visits to health center and daily cases which are multi- or paucibacillary and dnig resistance needs to he considered oral medication since the latter has much higher treatment in detail. Does not include screening costs costs. Does not include benefits of (see appendix table IA-3) or capital costs decreased transmission ($3 per DALY) District hospital level: treatment of Rehabilitation $190 per DALY, depending on level of Sensitive to labor and hospital costs complications, including reconstructive services provided surgery, ulcer therapy, and alternate medication Malaria (see District level: treatment of passively Cure $200-$500 per DALY tied to case-fatality See appendix table IA-3 for chemical chap. 13) detected malarial patients in regions Primary prevention rates and levels of endemicity vector control data. Active case searches of moderate-to-high endemicity and are expensive and drug resistance may be a chemical vector control problem if antimalarials are used haphazardly Dengue (see District or referred hospital levels: chap. 14) Improved case management with better Ctire $630 per DALY, tied to status of health Difficult to define shared costs. Possible education of physicians, lab facilities, and Palliation care system only in countries with system as defined pharmacies Improved case management and possible Cure $1,250 per DALY saved See appendix table IA-3 imimunization Palliation Primary prevention Improved case management and chemical Cure $1,200 per DALE saved See appendix table IA-3 for chemical vector control Palliation vector control data Primary prevention Improved case management and Cure $3,400 per DALY saved See appendix table I A-3 for environmental vector control Palliation environmental vector control Primary prevention Hepatitis B (see See cancer entry in this table chap. 15) Excess fertility Clinic level: insertion of intra-uterine Primary prevention $30-$150 per DALY saved While the initial cost of an IUD will be (see chap. 16) (lUDs) and oral contraceptives (Octs) greater than the initial (cP oLitlay, the disbursement amortized IUD cost (over its lifetime) is lower Matemal and C linic and district hospital level: improving Primary prevention Approximately $ 30-$2 50 per DALY linked Calculations are tleoretical. Family perinatal health the community-based outreach system to level of services provided, reductions in planning may be added in areas of low (see chap. 17) which provides prenatal and birth maternal/perinatal death rates and number contraceptive prevalence attendant care. Upgrading building of low-birth-weight babies preventedl via facilities to ensure safe deliveries the intervention (including surgical capabilities) (Table continues (n thc fvolk)owing page.) Table 1A-5 (continued) Condition Intervention strategy Objective Cost and Effectiveness CComments Protein-energy Dismct and hospital level: treatment with Rehabilitation Approximately $150-$250 per DALY tied Cost-effectiveness figures are still malnutrition feeding for child, education of mothers, to case-fatality reduction and level of theoretical (see chap. 18) and medication for infections services provided Micronutrient Clinic level: blood transfusion of severely Cure Moderate-to-high cost per DALY, All blood has to be tested for HIV and deficiency anemic patients (especially pregnant depending on case-fatality reduction from hepatitis, which will raise costs disorders (see women prior to delivery) transfusion chap. 19) HIV infection and Clinic level: use of ophthalmic antibiotic Primary prevention $5-$125 per DALY with lower values at Ointment easily applied and usually other sexually ointment at birth to prevent gonococcal higher prevalence rates requires one dose transmitted ophthalmia neonatorum diseases (see Treatment of sexually transmitted diseases Cure $1-$55 per DALY. Most cost-effective Cost-effectiveness linked to health care chap. 20) with antibiotics Primary prevention interventions are those targeted at most setting sexually active group in an HIV epidemic Clinic and district hospital level: treatment Palliation $80-$1,250 per DALY, depending on level Unreliable decreases in morbidity with use of AIDS with medical and surgical of services provided, with lower costs for of antivirals interventions home care and higher costs for antivirals Cancers (see Referral hospital level: case management Cure Cost per DALY as follows: leukemia- $150 per DALY when program is confined chap. 21) treatment of various cancers via surgery, Palliation $10,000; cervix-$2,600; breast-$3,100; to pain relief. With the exceptions of early supportive care, and chemotherapy lung-$12,000; liver-$11,000; colon and stage oral, cervix, breast, and rectum, rectum-$5,000; stomach-$10,500; technically advanced treatment with high esophagus-$10,600; mouth and foreign exchange content is needed for pharynx-$3,700; depends on most cures improvements in case-fatality rates Diabetes (see Clinic level: oral hypoglycemics to stabilize Secondary prevention Cost of outpatient provision of oral Limited compliance rates reported with chap. 22) non- insulin-dependent diabetes mellitus hypoglycemics about $25 per patient per long-term use of daily medication (NIDDM); concomitant health education year; can be quite effective in forestalling complications, including insulin- dependent diabetes (IDDM) Injected insulin and health education for Secondary prevention Cost of insulin therapy (life-saving) about The misuse of insulin could lead to IDDM $210 per year; estimated cost per [DALY excessive morbidity and mortality about $240 Cardiovascular Prinmary care level: medical management Secondary prevention Cost per DALY gained about $2,000 for Medication costs often expensive disease (see of hypertension typical case, linked to mortality reduction chap. 2.3) Medical management of Secondary prevention Cost per DALY gained about $4,000 in See above hypercholesterolemia typical case, linked to mortality Medical management of stable angina Secondary prevention Cost per DALY gained $100-$200 Rehabilitation Management after stroke or myocardial Secondary prevention Cost per DALY gained $150-$200, Depends on level of service provided and infarction (Ml) by behavioral change and tied to disability and mortality reductions costs of medication appropriate medication District hospital level: low-cost management Secondary prevention Cost per DALY gained approximately With all cardiovascular interventions of unstable angina acute Ml Rehabilitation $150-$350 High-cost management of unstable angina Secondary prevention Perhaps $30,000 per DALY saved Needs advanced health care system or acute Ml Rehabilitation Referral hospital level: angioplasty or bypass Secondary prevention Over $5,000 per DALY gained See above graft surgery Rehabilitation Chronic District hospital level: treatment of Palliation Approximately $200-$300 per day for Needs advanced health care system obstructive exacerbation, including mechanical treatment in hospital with a minimal effect pulmonary ventilatory assistance, steroids, and fluids on mortality rates disease (see chap. 24) Injury (see District or referral hospital level: treatment Cure Probably expensive, since based on types Very difficult to estimate cost- chap. 25) of injuries including medical and surgical Rehabilitation of injuries treated effectiveness, as it depends on levels of interventions treatment Cataract (see Clinic level: use of fixed surgical facilities, Cure $20-$40 per DALY. Some variation Includes costs for glasses every five years chap. 26) mobile surgical teams, or eye camps to expected, depending on societal and provide unilateral or bilateral cataract personal perception of disability from extraction blindness Oral health (see Clinic level: plaque and calculus removal, Secondary prevention Cost per treatment episode could range Pain and inability to chew as a result of chap. 27) fissure sealants, and topical fluoride. Cure between $5 and $20. Cost per DALY saved advanced untreated caries causes Extraction of teeth with advanced caries will depend on number of cases treated to significant disability. Societal demand for prevent one advanced case of caries and is dental procedures often high. Fluoridation suspected to be moderately high, since of water supply requires advanced supply there is no mortality burden. Costs also system. Identification of high-risk groups very sensitive to who is performing may be effective in reducing costs intervention; preventive care may also be done with lower costs Schizophrenia Clinic level: use of antipsychotic Rehabilitation Cost per DALY gained is $250-$300 for Clinical training of health center staff, out- and manic- medication to treat schizophrenic either the schizophrenia or manic- reach, and compliance components of case depressive patients with additional use of lithium depressive treatment program. Highly are important. illness (see to treat manic-depressive patients. sensitive to clinical/societal perception of chap. 28) disability Source: See chapters on individual diseases in this collection. Table IA-6. Intervention Characteristics and Cost-Effectiveness Potential intervention Strategy Objective application Potental groupa Age $25 per DALY b Breastfeeding promotion Public health: Behavior change Secondary prevention Moderate Childhood Diphtheria-pertussis-tetanus plus polio Public health: Immunization Primary prevention Substantial Childhood immunization Measles immunization Public health: Immunization Primary prevention Substantial Childhood Tuberculosis immunization Public health: Immunization Primary prevention Moderate Childhood lodization of salt Public health: Mass chemoprophylaxis Secondary prevention SLibstantial All ages Fortification of sugar with vitamin A Public health: Mass chemoprophylaxis Secondary prevention Substantial Childhood Semiannual mass dose of vitamin A Public health: Mass chemoprophylaxis Secondary prevention Substantial Childhood Rotavirus immunization Public health: Immunization Primary prevention Limited Childhood Hepatitis B immunization Public health: Immunization Primary prevention Substantial Childhood Medical treatment of measles with vitamin A Clinical: Primary care Cure Limited Childhood Medical treatment of acute respiratory infections Clinical: Primary care Cure Moderate Childhood with antibiotics Use of ophthalmic ointment at birth to prevent Clinical: Primary care Primary prevention Substantial Childhood gonococcal infection Targeted mass anthelmintics Public health: Mass chemoprophylaxis Secondary prevention Substantial Schoolage Antituberculosis chemotherapy with short- Clinical: District hospital Cure Substantial All ages course hospitalization Smoking prevention or cessation programs Public health: Behavior change Primary prevention plus secondary prevention Substantial Adults Use of condoms to prevent excess births and Public health: Behavior change Primary prevention Moderate Adults sexually transmitted diseases Blood screening for HIV Clinical: District hospital, Referral hospital Primary prevention Limited Adults Iodine injections for pregnant women Public health: Mass chemoprophylaxis Secondary prevention Substantial Adults Daily oral iron for pregnant women Public health: Mass chemoprophylaxis Secondary prevention Limited Adults Cataract removal Clinical: District hospital Cure Substantial Elderly Medical treanieint of leprosy Clinical: Primary care Cure Moderate Adults Malaria control with chemical pesticides Public health: Environmental Primary prevention Moderate All ages $25-$75 per DALY Pneumococcal immunization Public health: Immunization Primary prevention Moderate All ages Use of oral rehydration solutions Public health: Behavior change Secondary prevention Substantial School age Improved weaning practices Public health: Behavior change Secondary prevention Moderate Childhood Food supplements for children Public health: Mass chemoprophylaxis Secondary prevention Limited Schoolage Food supplements for pregnant women Public health: Mass chemoprophylaxis Secondary prevention Limited Adults Improved antenatal care by upgrading facilities Clinical: Primary care, District hospital, Referral Primary prevention Limited Adults and providing family planning hospital $75-$250 per DALY Medical treatment of tetanus Clinical: District hospital Cure Limited Childhood Cholera immunization Public health: Immunization Primary prevention Limited Childhood Malaria control with passive case finding and Clinical: Primary care, Public health, Primary prevention plus cure Moderate All ages chemical pesticides with treatment Environmental Medical and surgical treatment of leprosy Clinical: Primary care, District hospital Rehabilitation plus palliation Limited All ages complications Antibiotic prophylaxis for children with history Clinical: Primary care Secondary prevention Limited Childhood of rheumatic fever Public preventive package for most Public health: Behavior change Primary prevention plus secondary prevention Moderate Adults cardiovascular risk factors Insulin therapy for non-insulin-dependent Clinical: Primary care Secondary prevention Limited Adults, elderly diabetic individuals Management of stable angina with medication Clinical: Primary care Rehabilitation plus secondary prevention Limited Adults, Elderly Management of post-myocardial infarction or Clinical: Primary care, Public health, Behavior Secondary prevention Moderate Adults, Elderly post-stroke patients change Low-cost medical management of unstable or Clinical: District hospital Rehabilitation plus secondary prevention Limited Adults, Elderly inyocardial infarction Cancer pain management Clinical: Primary care Palliation Substantial All ages Onchocerciasis control with chemical pesticides Public health: Environmental Primary prevention Moderate All ages Schizophrenia or manic-depressive illness Clinical: Primary care Rehabilitation Moderate Adults treatment with medication $250-$1 000 per DALY Referral of pharyngitis cases for antibiotic Public health: Screening and referral Primary prevention Limited Childhood prophylaxis to prevent rheumatic fever and rheumatic heart disease Improved dengue case management via Clinical: Behavior change Primary prevention Limited All ages education of health care providers > $1,000 per DALY Medical and surgical management of chronic Clinical: Referral hospital Rehabilitation plus palliation Limited Adults, Elderly obstructive pulmonary disease Surgery for rheumatic heart disease Clinical: Referral hospital Rehabilitation plus secondary prevention Limited Adults Management of moderate hypertension with Clinical: Primary care Secondary prevention Moderate Adults, Elderly medication Management of hypercholesterolemia with Clinical: Primary care Secondary prevention Limited Adults, Elderly medication High-cost management of Ml or unstable angina Clinical: District hospital Secondary prevention Limited Adults, Elderly Management of coronary artery disease with Clinical: Referral hospital Rehabilitation plus secondary prevention Limited AdLilts, Elderly surgery Medical and surgical management of cancers Clinical: Referral hospital Cure plus palliation Limited All ages Dengue control with chemical pesticides with or Public health: Environmental Primary prevention Moderate All ages without improved case management Dengue control via drainage and land Public health: Environmental Primary prevention Limited All ages management with or without improved case management a. Age groups are defined as follows: Childhood = age 0 to 4; School age = age 5 to 14; Adults age 15 to 59; Elderly = age 60 plus. Most interventions will be useful for a range of age groups; the principal age group to whom the intervention would be addressed is indicated. h. DALY = disability-adjusted life-years. Source: Tables IA-3 and IA-5. 30 Dean T. Jamison Table IA-7. Intervention Cost-Effectiveness by Objective Primary Secondary Cost per DALY Numbera prevention prevention Cure Rehabilitation Palliation < $25 22 10 8 5 0 0 $25-$75 6 2 4 0 0 0 $75-$250 13 4 6 0 2 1 $250-$1,000 2 2 0 0 0 0 > $1000 9 2 5 1 3 2 Total 52 20 23 6 5 3 a. The total number of interventions does not equal the number of objectives, as some interventions have multiple objectives. Source: Appendix table IA-6. Table 1A-8. Intervention Cost-Effectiveness by Public Health and Clinical Strategy Public health Clinical Environ- Mass chemo- Screening Behavior Referral Cost per DALY mental prophylaxis Immunization and referral change Primary care District hospitai hospital < $25 1 6 5 0 3 4 3 1 $25-$75 0 1 1 0 2 1 1 1 $75-$250 2 0 1 0 2 8 3 0 $250-$1,000 0 0 0 1 1 0 0 0 > $1,000 0 0 0 0 0 2 1 4 Total 3 7 7 1 8 15 8 6 Source: Appendix table IA-6. Appendix 1 B: Countries and Territories as Grouped mies grouping was previously labeled Industrialized nonmarket in this Collection econornies. Analyses in the collection sometimes further aggregate The table on the facing page lists the country and territorial countries into indusrrialized economies included in the first and groupings used for aggregating country into regional data second groups in this table and developing economies as shown throughout this collection. The Industrialized transition econo- in the third, fourth, fifth, and sixth groups. Table 1B-I. Regional Groupings of Countries and Territories Industrial market economies Industrialized transition economies Latin America and the Caribbean Sub-Saharan Africa Middle East and North Africa Asia and the Pacific Austrialia Albania Antigua and Barbuda Angola Afghanistan Bangladesh Austria Bulgaria Argentina Benin Algeria Bhutan Belgium Czechoslovakia Bahamas Botswana Bahrain Brunei Canada Former German Dem. Rep. Barbados Burkina Faso Egypt, Arab Rep. of Cambodia Channel Islands Hungary Belize Burundi Gaza Strip China (excluding Taiwan) Cyprus Poland Bolivia Cameroon Iran, Islamic Rep. of Fiji Finland Romania Brazil Cape Verde Iraq French Polynesia France U.S.S.R. Chile Central African Republic Israel Guam Germany, Fed. Rep. of Yugoslavia Colombia Chad Jordan Hong Kong Greece Costa Rica Comoros Kuwait India Iceland Cuba Congo, People's Rep. of the Lebanon Indonesia Ireland Dominica Cote d'lvoire Libya Kiribati Italy Dominican Republic Djibouti Morocco Korea, Dem. People's Rep. of Japan Ecuador Equatorial Guinea Oman Korea, Republic of Luxembourg El Salvador Ethiopia Pakistan Lao People's Dem. Rep. Malta Grenada Gabon Qatar Macao Netherlands Guadeloupe Gambia, The Saudi Arabia Malaysia New Zealand Guatemala Ghana Syrian Arab Republic Maldives Norway Guyana Guinea Tunisia Mongolia Portugal Haiti Guinea-Bissau Turkey Myanmar Spain Honduras Kenya United Arab Emirates Nepal Sweden Jamaica Lesotho West Bank New Calednoia Switzerland Martinique Liberia Yemen, People's Dem. Rep. of Pacific Islands United Kingdom Mexico Madagascar Yemen Arab Republic Papua New Guinea United States Montserrat Malawi Other North Africa Philippines Other Europe Netherlands Antilles Mali Singapore Other North America Nicaragua Mauritania Solomon Islands Panama Mauritius Sri Lanka Paraguay Mozambique Taiwan Peru Namibia Thailand Puerto Rico Niger Tonga St. Kitts and Nevis Nigeria Vanuatu St. Lucia Reunion Viet Nam St. Vincent and the Grenadines Rwanda Western Samoa Suriname Sati Tome and Principe OtherMicronesia Trinidad and Tobago Senegal Other Polynesia Uruguay Seychelles Venezuela Sierra Leone Virgin Islands (U.S.) Somalia Other Latin America South Africa Sudan Swaziland Tanzania Togo Uganda Zaire Zambia Zimbabwe Other West Africa 32 Dean T. Jamison Notes countries must now. in addition, face the burden of ageing and chronic and degenerative diseases" (World Health Organization 1988, p. 102). l am deeply indebted to manvof my colleagues forcomments and discussions 8. In many ways this review is very much in the spirit of Walsh and Warren's conceming earlier drafts of parts of this material; they include Jacques (1979. 1986) assessment of priorities for control of communicable childhood Baudouy, Robert Black, nohn Briscoe, J. Richard Bugamer Donald Bundy, diseases in developing countries; the current effort involves more extensive Guy Carrin, Lincoln Chen, E. Chigan, Andrew Creese, Joseph Davis, Nich- use of economic analysis and covers a much broader range of conditions. In olas Drager, William Foege, Davidson Gwatkin, Jean-Pierre Habicht, Ann subsequent work for the United Nations Development Programme (UNDP), Hamilton. Alaya Hammad, Jeffrey Hammer, D. A. Henderson, Ralph Hen- Walsh (1988) has extended her earlier work with Warren. Amler and Dull derson, Kenneth Hill, MichelJancloes. Jeffrey Koplan, Jean-Louis Lamboray, (1987) and the Department of Health and Human Senrices (1991) have Joanne Leslie, Bemhard Liese,judith McGuire, Richard Morrow, Christopher reviewed a broad range of preventive intervention policies for the United Murray, Philip Musgrove, Mead Over, Thomas Pearson, Richard Peto, Mar- States, and, more for clinical preventive services, the U.S. Preventive Services garet Phillhps, Nancy Pielemeier, Andre Prost, William Reinke, Julia Rushby, Task Force (1989) has reviewed the effectiveness of 169 interventions. The Robert Steinglass, Eleuther Tarimo, Carl Taylor, Anne Tinker, Kenneth state of Oregon in the United States has ordered over 700 interventions, using Warren, and David Wemer. cost-effectiveness and other criteria, for the purpose of rationing limited public DavidBell,Jos6-LuisBobadilla,RichardFeachem,AnthonyMeasham,and resources to provide health care for the poor; a recent edited collection W.HenryMosleyprovidedmewithparticularlyusefulinsightsandcomments. (Strosberg and others 1992) discusses many facets of the Oregon plan. Patel Peter Cowley provided valuable assistance with preparation of the tables in (1989) has reviewed estimates of cost and effectiveness of a range of health the chapter and helpful comments on the chapter as a whole. interventions for UNICEF, and Udvarhelyi and others (1992) provide a com- The opportunity to give the Heath Clark Lecture for 1989-90 at the London prehensive revLew of medical cost-effectiveness and cost-benefit studies from School of Hygiene and Tropical Medicine provided me both with valuable the perspective of their methodological adequacy. All these approaches to the feedback on portions of this chapter and with the chance to work on it for analytic evaluation of health practices fall within the general area of what is several months in a highly stimulating environment. Much of this work was increasingly known as "health technology assessment"; Garber and Fuchs done while I was at the U niversity of California, Los Angeles, which has been (1991) provide a valuable general overview of the field. donevery whileIwas supprtie ofUthiversity ofCaliffo niartLos.Angeles,whichhas 9. If one is simply assessing the relative attractiveness of alternative means I1. Issues associated wit thse health transiion and its implications for polic' for achieving a single, specific health objective-for example, reducing infant are increasingly widely discussed; for example, see Bell 1992; Bicknell and mortality-this measurement problem disappears, and one can judge interven- Parks 1989; Bobadilla and others, chapter3, ths collection; Chen and others tion cost-effectiveness simply in terms of, say, cost per infant death averted. 1992; Chesnais 1990; Evans, Hall, and Warford 1981; Foege and Henderson 10. There is at least anecdotal evidence to suggest that in immunization 1986; Harlan, Harlan, and Oii 1984; Jamison and Mosley 1991; and Mosley programs immunizations are often first provided where the cost per child and Cowley 1991.The work of Ju1o Frenk and his colleagues in Mexico-for contact is lowest; if, as is likely, these children have relatively low incidence example, Frenk and others 1989-has provided a particularly influential rates (for example, of tuberculosis) or case-fatality rates (for example, of impetus for work in this area. measles), then the Expanded Programme on Immunization may not be starting 2. General economic conditions and behavioral pattems in society influ- with the most cost-effective population subgroups. ence health outcomes (Bell and Reich 1988; Behrman 1990; Berman, Kendall, 11. In an early application of cost-effectiveness analysis within the health and Bhattacharyya 1989; Cochrane, Leslie, and O'Hara 1982; DaVanzo and sector, Bamum and others (1980) go beyond comparing marginal cost-effec- Gertler 1990). Thi collecion deals with these wider issues, for both adults and tiveness to attempting an analysis of maximization of total outcome for children, only insofar as they can be addressed by health-related intervention. different levels of expenditure on child surv ival; Forgy (1991) uses data from 3. A comprehensive analysis of health problems of adults in the developing th1s collection on child survival to undertake a similar analysis. world that calls for more explicit policy and programmatic attention in 12. It is important to recognize that some facilties address only a narrow addressing those problems has recently been completed for the World Bank range of conditions-for example, there are cancer and TB hospitals. Over and (Feachem and others 1992). Piot (chapter 20) discuss the usefulness of clinics for sexually transmitted 4. A recent assessment of intervention options for mental disorders by the diseases, and javitt (chapter 26) discusses use of mobile surgical camps (dis- World Health Organization concludes-although treatment costs are not trict-hospital level, In some senseo to deal with catarac . explicitly considered-that relativelv simple interventions could be much 13. Health care expenditures of approximately $460 bilhon in 1986 for the more widely used to address widespread mental disorders (World Health 242millionpeopleoftheUnitedStateswellexceededtheGNPofChina($320 Organization 1991). billion), with a population of 1.05 billion; it was close to triple the combined 5. The very different character of health interventions from other "goods" GNPs of all the World Bank member countries of Sub-Saharan Africa, which ,,, , , . , , , \ ~~~~~~~had a total population of about 425 million and a combined GN P of about $ 175 typicallv "chosen" by market forces-in particular consumer (and provider) billion. ignorance about links between interventions and health improvement-gen- 14. erates a need for specialist assessment of intervention choice and for serious r14. A separate Ite of evidence, albeit only suggestive, tor inefficgency consideration of mechanisms that deal with market failure to decide on the resulting from variation in margdal cost-effectiveness is the vers high degree level and composition of interventions to be provided. For a valuable review, of observed varation In procedure requence in somewhat similar environ see Barr 1992. A comprehensive approach to dealing with market failure ments (Sanders, Coulter, and McPherson 1989). through "managed competition" is provided by Enthoven 1988. Implications ofthis literature from the perspective ofdeveloping countries have been drawn in a major recent publication of the World Bank (World Bank 1993). References 6. These elements need consideration, obviously, independentlyof whether the government, the private sector, or nongovemmental organizations are Akin, J., Nancy Birdsall, and D. de Ferranti. 1987. FinancingHealth Services in responsible for delivering the relevant service. Akin, Birdsall, and de Ferranri Developing Countries. Washington, D.C.: World Bank. (1987) discuss these issues from the perspective of financing health systems; Amler R. W., and H. B. Dull, eds. 1987. Closing the Gap: The Burden of Birdsall (1989 ) further discusses the govemment role in the health sector, and Stiglitz (1989) provides a valuable general overview of the economic role of ;nnecessary Illness. New York: Oxford University Press. the state. Anderson, M. A. 1990. "Nature and Magnitude of the Problem of Suboptimal 7. At about the same time, the World Health Organization was also BreastfeedingPracrices."PaperpresentedatthelntemationalPolicymakers beginning to address these issues; Dr. Hiroshi Nakajima, the current director- Conference on Breastfeeding, Florence, Italy, July 30-August 1. general. observed: "Even before we win our battle against the communicable Barnum, Howard. 1987. "Evaluating Healthy Davs ofLife Gained from Health diseases, which has engaged us since our earliest days, many developing Projects." SocialScienceandMedicine24(10):833-41. Disease Control Priorities in Developing Countries: An Overview 33 Barnum, Howard, R Barlow, L. Fajardo, and A. Pradilla. 1980. A Resource Feachem, R. G. A., T. Kjellsrrom, C. J. L. Murray, Mead Over, and M. A. AllocatonModelforChildSurvival. Cambridge, Mass.: Oeldeschlager, Gunn Phillips. 1992. The Health of Adults in the Developing World. New York: and Hain. Oxford University Press. Barr. Nicholas. 1992. "Economic Theory and the Welfare State: A Survey and Foege, W. H., and D. A. Henderson. 1986. "Management Pnorities in Primary Interpretation."Journal of Economuic Literature 30:741-803. Health Care." In J. A. Walsh and K. S. Warren, eds., Strategies for Prnmars Basta, S. S., D. K. Soekirman, and N. S. Scrimshaw. 1979. "Iron Deficiency Health Care. Chicago: University of Chicago Press. Anemia and the Productivity of Adult Males in Indonesia." American Forgy, Lawrence. 1991. "Cost-Effectiveness in Child Health: A Minimum Journal of Clinical Nunmnon 32:916-2 5. Information Approach to Planning and Forecasting." Paper presented at the Behrman,1. R. 1990. "A Survey on Socioeconomic Development, Structural seminar on Child Survival Interventions: Effectiveness and Efficacy, at the Adjustment and Child Health and Mortality in Developing Countries." In Johns Hopkins University School of Hygiene and Public Health, Baltimore, K. Hill, ed., Child Survival Programs: Issues for the 1990s. Baltimore: Johns June 20-22. Hopkins Universitv School of Hvgiene and Public Health, Institute for Frenk, J., J.-L. Bobadilla, Jaime Sepulveda, and M. Lopez Cervantes. 1989. Intemnational Programs. "Health Transition in Middle-income Countries: New Challenges for Bell, D. E. 1992. "Some Implications of the Health Transition for Policy Health Care." Health Policy and Planning 4:29-39. and Research. " In L. Chen, A. Kleinman, J. C. Caldwell, J. E. Potter, and Garber, Alan M. and Victor R. Fuchs. 1991. "The Expanding Role of Tech- N. Ware, eds., Health and Social Change. Westport, Conn.: Aubum nology Assessment in Hiealth Policv." Stanford Law arid Pohcy Review House. 3:203-9. Bell, D. E., and M. R. Reich, eds. 1988. Health, Numtion, and Economic Cnsis. Garber, Alan XI., and Phelps, Charles E. 1992. "Economic Foundations of Dover, Mass.: Aubum House for the Harvard School of Public Health. Cost-Effectiveness Analysis." Working Paper 4164. National Bureau of Berman, P., C. Kendall, and K. Bhattacharyya. 1989. "The Household Pro- Economic Research, Cambridge, Mass. duction of Health: Putting People at the Center of Health Improvement." Ghana Health Assessment Project Team. 1981. "Quantitative Method of [n Towards More Efficacy in Child Survival Strategies: Understanding the Social Assessing the Health Impact of Different Diseases in Less Developed and Pnrvate Constraints and Responsibilities. Baltimore: Johns Hopkins Uni- Countries." lntenianionalJournal of Epidemiology I0(l):73-80. versity School of Hygiene and Public Health. Harlan, W. R., L. C. Harlan, and W. L.Oii. 1984. "Changing Disease Patterns Bicknell, W. I., and C. L. Parks. 1989. "As Children Survive: Dilemmas of in Developing Countries: The Case of Malaysia." In P. Leverton and L. Aging in the Developing World." Social Science and Medicine 28(1):59-67. Massi, eds., Health Informianon Systems. New York: Praeger Scientific. Birdsall, Nancy. 1989. "Thoughts on Good Health and Gocd Government. Institute of Medicine. 1986. Neii: Vaccine Development Establshing Priorities. Daedalus 118:23. Vols. I and 2. Washington, D.C.: National Academy Press. Briscoe, J. 1989. "Adult Health in Brazil: Adjusting to New Challenges." Jansison, D. T.,J. R. Evans, T. King, 1. Porter, N. Prescott, and A. Prost. 1984. Report 7807-BR, World Bank, Washington, D.C. Chmna: The Health Sector. A World Bank Country Study. Washington, D.C. Briscoe, J., and D. de Ferranti. 1988. Water for Rural Communities. Washing- Jamison, D. T., and W. H. Mosley. 199 1. "Selecting Disease Control Priorities ton, D.C. in Developing Countries: Health Policy Responses to Epidemiological Bumgamer, J. R. 1992. China: Long-term Issues and Optons in the Health Change."AmericanJournialofPubihcHealth81:15-22. Transition. A World Bank Country Study. Washington, D.C. World Bank. Johannesson, Magnus. 1992. "On the Discounting of Gained Life-Years in Caims, J., and K. Johnston. 1991. "Condition-Specific Outcome Measures as Cost-Effectiveness Analysis." InternationalJournal of Technology Assessment an Alternative to Across-Programme QALYs." Health Economics Unit, iHealthCare8:35964. University of Aberdeen. Last, J. M., ed. 1988. A Dictionary of Epidemiology. 2d ed. New York: Oxford Chen, L., A. Kleinman, J. C. Caldwell, J. E. Potter, and N. Ware, eds. 1992. University Press for the Intemational Epidemiological Association. Health and Social Change. Westport, Conn.: Aubum House. Leslie. Joanne. 1989. "Women's Time: A Factor in the Use of Child Survival Chesnais, J-C. 1990. "Demographic Transition Pattems and Their Impact on Technologies?" Health Policy and Planning4(l):1-16. the Age Strucrure." Population and Developmenr Review 16l2):327-36. . 1992. "Women's Time and the Use of Health Services." IDS Bulletn Cochrane, S., J. Leslie, and D. O'Hara. 1982. "Parental Education and Child 23:4-7. Health: Intracountry Evidence "Health Policy and Education 2:213-50. MM'BWR (Morbidity and Mortality Weekly Report). 1992. "Years of Potential Cropper, Maureen L., Sensa K. Aydede, and Paul R. Portney. 1992. "Rates of Life Lost before Ages 65 and 85--United States, 1989-1990." MMWBR Time Preference for Saving Lives." American Economic Review 82:469-72. 41(18):313-15. DaVanzo, J., and P. Gertler. 1990. "Household Production of Health: A Mosley, W. H., and P. Cowley. 1991. "The Challenge of World Health." Microecononsic Perspective on Health Transitions." Rand N-3014-RC. Population Bulletin 46(4):1-39. Rand Corporation, Santa Monica, Calif. Mosley, W. H., and R. Jolly. 1987. "Health Policy and Program Options: Drummond, MN. F., G. L. Stoddart, and G. W. Torrance. 1987. Methods for the Compensating for the Negative Effects of Economic Adjustment." In G. A. Economic Etaluation of Health Care Programs. Oxford: Oxford University Comia, R. Jolly, and F. Stewart, eds., Adjustment with a Human Face. Oxford: Press. Clarendon Press. Enthoven, Alain C. 1988. Theory and Practice of Managed Compention i Health Murray, C. J. 1990. "Rational Approaches to Priority Setting in Intemational lCealth." Journal of Tropical Medicine and Hygiene 93(5):303-11. CGire Finance. Amisterdam: North Ho(lland. EuroQol Group. 1990. "EuroQol-A New Facility for the Measurement of Musgrove, Philip. 1991. "The Burden of Death at Different Ages: Assump- Health-Related Quality of Life." Health PolzcN 16:199-208. tions, Parameters and Values." Occasional Paper 12. Latin America and the Healt-Relaed Qulity f Lif." Helth Plicy 6.199208.Caribbean Regional Office, Human Resources Division, Technical Depart- Evans, J. R., K. L. Hall. and J. Warford. 1981. "Health Care in the Developing men W o nasIngton, D.c. Wol:Prozblems of Scarcity and Choice." New EniglandJournal of Medicine ment, World Bank, Washington, D.C. World: 1117 . Nord, Erik. 1991. "The Relevance of QALYs in Prioritizing between Different Patients." Paper presented at the 12th Nordic HESG meeting, Copenhagen, Fallowfield, Leslie. 1990. The Qualt-y of Life. London: Souvenir Press. August. Feachem, R. G. A., W. J. Graham, and 1. M. Tinsaeus. 1989. "Identifying Over, Mead. 1988. "Cost-Effective Integration of Immunization and Basic Health Problems and Health Research Priorities in Developing Countries." Health Services in Developing Countries: The Problem of Joint Costs." Journil of Tropical Medicme and Hygiene 92(3): 133-91. Working Paper 23. World Bank. Washington, D.C. 34 Dean T. Jamison Patel, Mahesh S. 1989. "Eliminating Social Distance between North and Udvarhelyi, l. Steven, Graham A. Colvitz, Arti Rai, and Amold M. Epstein. South: Cost-Effective Goals for the 1990s." StaffWorking Paper 5. UNICEF, 1992. "Cost-Effectiveness and Cost-Benefit Analyses in the Medical Liter- New York. ature." Annals of Internal Medicine 116:238-44. Phelps, Charles E., and Alvin 1. Mushlin. 1991. "On the (Near) Equivalence USDHHS (U.S. Department of Health and Human Services). 1991. Healthy of Cost-Effectiveness and Cost-Benefit Analysis." International Journal of People 2000: National Hecath Promotion and Disease Prevention Objectives. Technology Assessment in Health Care 7:12-21. Washington, D.C.: U.S. Govemment Printing Office. Pollitt, E. 1990. Malnutriton and Infection in the Classroom. Paris: UNESCO. U.S. Preventive ServicesTask Force (R. S. Lawrence, Chairman). 1989. Guide Preston, S. H. 1991. "Health Indexes and Health Sector Planning." Paper to Clinical Preventive Services. Baltimore: Williams and Wilkins. presented at the Workshop on the Policy and Planning Implications of the Walsh, J. A. 1988. Estabishing Health Priorities in the Developing World. Boston: Epidemiological Transition in Developing Countries, at the National Re- Adams Publishing Group for the United Nations Development Programme. search Council, Washington, D.C., November 20-22. Walsh, J. A., and K. S. Warren. 1979. "Selective Primary Health Care-An Rosser, R. M., and P. Kind. 1978. "A Scale of Valuations of States of Illness: Interim Strategy for Disease Control in Developing Countries." New En- Is There a Social Consensus?" International Journal of Epidemiology 7(4): gland ournal ofkfedicine 301:967- 74. 347-58. - , eds. 1986. Strategies for Pnmsary Health Care: Technologies Appropnate Sanders, D., A. Coulter, and K. McPherson. 1989. "Variation in Hospital for the Control of Disease in the Developing World. Chicago: University of Admission Rates: A Review of the Literature." King's College Fund, Paper Chicago Press. 79. London. Weinstein, N. D. 1989. "Optimistic Biases about Personal Risks." Science Squire, Lyn. 1989. "Project Evaluation in Theory and Practice." In B. Hollis 246:1232-33. Chenery and T. N. Srinivasan, eds., Handbook of Development Economics, WHO (World Health Organization). 1988. From Alsa-Ata to the Year 2000: Vol. 2. Amsterdam: North Holland. Reflectons at the Midpoint. Geneva. Stiglitz, J. 1989. "On the Economic Role of the State." In A. Heertje, ed., The . 1991. Evaluaton of Methods for the Treatment of Mental Disorders. WHO Economic Role of the State. Cambridge, Mass.: Basil Blackwell in association Technical Report 812, Geneva. with Bank Insinger de Beauford NV. World Bank. 1980. Health Sector Policy Paper. Washington, D.C. Strosberg, Martin A., Joshua M. Weiner, and Robert Baker, with 1. Alan Fein, . 1992. World Development Report 1992: Development and the Environ- (ed.) 1992. Ranoning America's Medical Care: The Oregon Plan and Beyond. ment. Washington, D.C. Washington, D.C.: The Brookings Institution. . 1993. World Development Report 1993. Investing in Health. Washing- Tan-Torres, Teresa. 1990. "Comparison of Different Methods of Eliciting ton, D.C. Utilities for Outcome States in Leprosy." University of the Philippines, Zeckhauser, R., and D. Shepard. 1976. "Where Now for Saving Lives?" Law Department of Medicine, Clinical Epidemiology Unit. and Contemporary Problems 40:5-45. 2 Causes of Death in Industrial and Developing Countries: Estimates for 1985-1990 Alan D. Lopez Information about the cause structure of mortality hasfor many predict the level of mortality for a given cause (among twelve years served as the cornerstone for monitoring health progress broad cause-of-death groups) from information about the and for the determination of health priorities. Over the years, crude death rate. Perhaps the greatest limitation of this method information systems on routine causes of death have been is that it implicitly assumes that the cause-level relationship is established in many countries and, throughout the industrial invariant over time. This is unlikely to be the case in contem- world at least, these have evolved to the point at which there porary developing countries, where modemization, social and is now virtually complete coverage of deaths and medical economic change, and the infusion of medical technology certification. A handful of developing countries, primarily in have undoubtedly altered the nature and severity of disease Latin America and East Asia have achieved comparable stan- epidemics. Adjustment of the estimates generated by the model dards of reliability of their cause-of-death statistics. In several on the basis of a critical appraisal of available epidemiological other countries, considerable progress has been achieved to- information, as has been done in Indonesia, for example (see ward obtaining reasonably reliable mortality information for Hull, Rohde, and Lopez 1981), increases the reliability of the at least the urban populations, although calculation of mortal- estimates. This type of complementary analysis is clearly quite ity rates is often impeded by lack of information about the complicated at a global, or even regional, level, however, and population at risk. For the majority of developing countries, hence the indirect estimates of cause-of-death pattems should however, the reliability of cause-of-death data emanating from be viewed only as a first approximation to the underlying vital registration systems-where they exist-is sufficiently epidemiological environment. The method followed in this poor essentially to preclude their use for assessment of the chapter of estimating the global and regional mortality situation national health situation. in 1985 is one of progressively assembling nationally represen- In the absence of reliable information on medically certified tative mortality pattems according to the degree of confidence causes of death, countries are increasingly adopting lay report- in their reliability.' Thus the first section of the chapter deals ing schemes to obtain at least a broad overview of health exclusively with the industrialized market and nonmarket conditions in populations in which deaths are not routinely countries. The market economies include the following: recorded or medically certified. This technique has undoubt- Australia Greece New Zealand edly generated useful information for several populations, but Austria Iceland Norway it has not been nearly as widely exploited as, for example, the Belgium Ireland Portugal indirect techniques for assessing levels of mortality on the basis Canada Italy Spain of information from women about children ever bom and , T children surviving. As a result, global and regional pattems of Denmark Japan Sweden mortality by age and sex can be, and have been, established France Malta United Kingdom with some confidence, whereas the cause structure of mortal- Franc e Mala United States ity, for at least half of the world's population, is, at best, Germany, Fed. Rep. Netherlands United States uncertain. The countries listed below are considered industrial non- One method of estimating causes of death in countries market economies: without reliable information is to develop a model of the epidemiological transition based on the experience of the Albania German Dem. Rep. Romania industrial countries. Hakulinen and others (1986), for exam- Bulgaria Hungary U.S.S.R. ple, estimated a series of cause-specific regression equations to Czechoslovakia Poland Yugoslavia 35 36 Alan D. Lopez These thirty-three countries in 1985 accounted for a popu- Table 2-1. Population of the Developing World lation of about 1.2 billion people (that is, 1,200,000,000) or Population Percentage of total approximately 25 percent of the world total.2 Some 1 I million Region (millions) population deaths occur each year in these countries (out of a global total of roughly 50 million), virtually all of which are medically Sub SaharanAfrica 456 12.4 certified.' There remain, however, important differences In iddletEa/N arica 376 10 diffrencs in Latin America/Caribbean 402 11.0 diagnostic and coding practices among these countries due to India 765 20.9 differences in medical training, availability of diagnostic aids, China 1,065 29.0 and the like. Percy and Muir (1989), for example, in their study Other Asia/Pacific 604 16.5 of the international comparability of cancer mortality data in Total 3,668 100.00 seven industrialized countries, found that cancer deaths were typically overreported by 3 to 4 percent compared with the Source: Author. United States. The largest difference was estimated for France, where coding procedures resulted in a 10 percent higher death cians at district primary health care centers. Obviously, reli- rate for cancer than would have been the case if the procedures able information about the mortality pattem for these two in use in the United States had applied. The implications of countries will have a significant effect on the estimation pro- artifacts such as these for comparative mortality analyses could cedure for all developing countries. Indeed, in 1985, one-half be quite substantial. For example, the age-adjusted cancer of the population of the developing world were living in China death rate for France in 1984 was 139.5 per 100,000, or about and India as shown in table 2-1. 5 percent more than what was observed in the United States (132.4 per 100,000). After allowing for coding differences, the Causes of Death French cancer death rate was recalculated by Percy and Muir at 127.5 per 100,000, almost 5 percent less than the rate for The causes of death selected for mortality estimation are the United States. Although these artifacts are likely to be less shown in table 2-2, along with the corresponding codes of the of a problem for aggregate analyses, such as those reported in International Classification of Diseases (ICD). For the ninth this study, it must be kept in mind that the resulting cause-of- revision of the ICD (ICD-9), the codes refer to items in the Basic death pattem, even for industrial countries, is not exact. None- Tabulation List, whereas for ICD-8, the numbers refer to dis- theless, for broad cause-of-death groups at least, the structure eases and injuries included in list "A." Both of these are of mortality can be reasonably well established from the data summary or "short" lists and are used by WHO to collect and available. store mortality information. In some cases, the items available The estimation of cause-specific mortality for the develop- in the summary lists do not correspond exactly to the compo- ing regions of the world is even less precise. Global estimates sition of cause-of-death categories defined on the basis of the of disease-specific morbidity and mortality have been prepared detailed rubrics of the ICD. The discrepancies are generally by several technical programs in the World Health Organiza- minor, however, and in any case will have much less effect on tion. These estimates are frequently based on studies carried the estimates than the artifacts and uncertainties mentioned out at the community level in various developing countries above. and then extrapolated to yield regional and global figures. This Eight broad categories of causes of death are considered: is clearly a very imprecise method, but in the absence of vital registration there is little altemative but to "evaluate and * Infectious and parasitic diseases extrapolate." In addition, cause-of-death data are available for * Neoplasms a number of developing countries, particularly in Latin Amer- * Diseases of the circulatory system and other selected ica, and provided the coverage and reliability of the informa- degenerative diseases4 tion is known, even approximately, these data can be exploited * Chronic obstructive lung diseases (principally chronic to estimate national cause-of-death pattems. Regional and bronchitis and emphysema) global estimates can then be obtained, although this aggrega- * Complications of pregnancy tion introduces an additional degree of uncertainty into the estimates because of missing data for some countries. For some * Perinatal conditions large populations in the developing world, cause-of-death in- * Injury and poisoning (all extemal causes) formation is now becoming available which is of sufficient * All other causes quality and representativeness to permit reasonable estimates of the epidemiological situation for the country as a whole. For the industrial countries at least, the remainder category China now has routine mortality data, almost all of which is has been further disaggregated into "other specific causes" and medically certified for over 100 million people in rural and "symptoms and ill-defined conditions." For some categories, urban areas in the eastem half of the country, where the bulk namely, infectious and parasitic diseases, diseases of the circu- of the population resides. India also has implemented a rural latory system, and the remainder category, estimates of mor- survey of causes of death. It is based on lay reporting of the tality from more specific conditions listed in table 2-2 have also cause of death, with verification on a sample basis by physi- been attempted. This was done because disease-specific esti- Causes of Death in Industrial and Developing Countries Estinmates for 1985-1990 37 Table 2-2. Causes of Death and Corresponding Categories in the Intenational Classification of Diseases, Injuries, and Causes of Death (ICD) Cause of death ICD-8 List A icn-9 Basic Tabulation List Main Categories Infectious and parasitic diseases A1-44, A90-92, A99 01-07, 320-322 Neoplasms A45-61 08-17 Circulatory system and certain degenerative diseases A64, A80-88, A98, A102, A105-106 25-30, 181, 341, 347, 350 Chronic obstructive pulmonary (lung) disease A93, A96 323-326 Complications of pregnancy Al 12-118 38-41 Perinatal conditions A131-135 45 Injury and poisoning A138-150 E47-E56 Ill-defined causes A136, A137 46 Other causes Other codes All other codes Infectious diseases Diarrhea Al-5, A99 01 Tuberculosis A6-10 02 Acute respiratory infection A15-17, A89-92 033-035, 320-322 Measles A25 042 Polio A22-23 040, 078 Yellow fever, dengue, and encephalitis A26-27 044-045 Malaria A31 052 Schistosomiasis and filariasis A39, A41 072, 074 Intestinal parasites A42-43 075-076 Circulatory and degenerative diseases Ischemic heart disease A83 27 Cerebrovascular disease A85 29 Other cardiovascular diseases A80-82, A84, A86-88 25, 26, 28,30 Diabetes A64 181 Certain degenerative diseases (nephritis, cirrhosis, ulcers) A98, A102, A105, A106 341, 347, 350 Other disorders Mental disorders A69 210-212 Oral health diseases A97 330 Micronutrient deficiency A62-63, A67 180, 193, 200 Malnutrition A65 190-192 Source: Intemational Classification of Diseases, Injuries, and Causes of Death. matesareundoubtedlyofmuchgreaterrelevancefordetermin- million (90 percent) of the 11 million deaths were at age ing health priorities than an aggregate of conditions which forty-five and over. In contrast, there were 355,000 infant and may require substantially different strategies for prevention child deaths below age five (3.2 percent of the total), 275,000 and control. of which occurred among infants. The vast majority of these infant deaths in turn occurred very early in life (typically Causes of Death in Industrial Countries in 1985 within the first week) and were due to various perinatal and congenital conditions which are difficult to eliminate. None- Tables 2-3, 2-4, and 2-5 summarize the cause-of-death struc- theless, further progress in reducing this toll of over 350,000 ture for the industrialized world as a whole (table 2-3) as well young-child deaths each year can be expected through the as for the groups of industrial market economies (table 2-4) and reduction of inequalities in access to, and use of, prenatal care the industrial nonmarket economies (table 2-5) separately. and infant and child health services among different sectors of the population. Indeed, reduction of such inequalities is cen- All Industrial Countries tral to the health-for-all-strategies in these countries and ap- plies not only during infancy and childhood but at later ages The structure of mortality shown in table 2-3 for the industri- as well. alized countries as a whole is very much what one would expect Infectious and parasitic diseases (including acute respiratory for a population with an average life expectancy of seventy- diseases) claimed just over half a million deaths in the indus- four years. Of the 11.05 million deaths reported for these trialized countries about 1985, two-thirds of which were among countries during 1985, 7.63 million, or approximately 70 per- the elderly (sixty-five years and over). Even so, 110,000 of cent, occurred at age sixty-five and over. Another 2.3 million these deaths were among children below the age of five, with were at age forty-five through sixty-four. That is, almost 10 all but about 6,000 of these occurring in the nonmarket coun- 38 Alan D. Lopez Table 2-3. Causes of Death in Industrial Countries, 1985 Number (thousands) Percentage Cause of death Males Females Total Males Females Total Infectious and parasitic diseases 266 240 506 4.7 4.4 4.6 Acute respiratory infections 184 184 368 3.3 3.4 3.3 Tuberculosis 30 10 40 0.5 0.2 0.4 Neoplasms 1,263 1,030 2,293 22.5 18.9 20.8 Circulatory and certain degenerative diseases 2,720 3,210 5,930 48.6 59.0 53.7 lschemic heart disease 1,199 1,193 2,392 21.4 21.9 21.7 Cerebrovascular disease 590 914 1,504 10.5 16.8 13.6 Diabetes 59 94 153 1.1 1.7 1.4 Complications of pregnancy 0 4 4 0.0 0.1 0.0 Perinaral conditions 60 40 100 1.1 0.7 0.9 Chronic obstructive lung diseases 245 140 385 4.4 2.6 3.5 Injury and poisoning 536 236 772 9.6 4.3 7.0 Ill-defined causes 115 132 247 2.1 2.4 2.2 All other causes 397 410 807 7.1 7.5 7.3 Total 5.601 5,444 11,045 100.0 100.0 100.0 a. Estimated at 6,000 due to under-reporting. Source: Calculated from WHO mortality database. tries. It should be emphasized here that this category does not principal causes of death from infectious diseases in industrial include deaths coded to acquired immunodeficiency disease countries, accounting for four-fifths of the half-million deaths (AIDS) because it was not possible to distinguish these deaths still due to infections in the industrialized world. from other causes in the mortality data reported to WHO. Still, In these countries, cancer, primarily malignant neoplasms, the number of new AIDS cases reported for the industrialized claims the lives of 2.3 million persons each year, 55 percent of countries about 1985 was less than 30,000, and hence the which are males. Cancer too has a relatively high average age mortality from AIDS would not have altered dramatically the at death, with 1.4 of the 2.3 million deaths occurring beyond overall total of deaths from infectious and parasitic diseases. age sixty-five. Nonetheless, there is still very substantial scope Rather, acute respiratory infections, primarily pneumonia for preventing premature death from neoplastic diseases. Of among the elderly, account for about two-thirds of the deaths the remaining 900,000 cancer deaths, all but a handful from infections in industrial countries. Another 40,000 deaths (15,000) occurred between the ages of fifteen and sixty-four. are attributable to tuberculosis, primarily at age forty-five and The causes of cancer are still very much a matter of investi- over. Acute respiratory infections and tuberculosis are thus the gation, but in countries such as the United States, where Table 2-4. Causes of Death in Industrialized Market Countries, 1985 Number (thousands) Percentage Cause of death Males Females Total Males Females Total Infectious and parasitic diseases 152 156 308 4.3 4.8 4.5 Acute respiratory infections 115 126 241 3.3 3.8 3.5 Tuberculosis 9 4 13 0.3 0.1 0.2 Neoplasms 888 719 1,607 25.1 21.9 23.6 Circulatory and certain degenerative diseases 1,682 1,773 3,455 47.6 54.1 50.7 Ischemic heart disease 730 595 1,325 20.6 18.1 19.4 Cerebrovascular disease 334 467 801 9.5 14.2 11.8 Diabetes 48 73 121 1.3 2.2 1.8 Complications of pregnancy 0 1 1 0.0 0.0 0.0 Perinatal conditions 24 18 42 0.7 0.5 0.6 Chronic obstructive lung diseases 168 86 254 4.7 2.6 3.7 Injury and poisoning 288 143 431 8.1 4.4 6.3 Ill-defined causes 83 99 182 2.3 3.0 2.7 Allothercauses 251 284 535 7.2 8.7 7.9 Total 3,536 3,279 6,815 100.0 100.0 100.0 Source: Calculated from WHO mortality database. Causes of Death in Industrial and Developing Counmes: Esnmates for 1985-1990 39 Table 2-5. Causes of Death in Industrialized Nonmarket Countries, 1985 Number (thousands) Percentage Cause of death Males Females Total Males Fenmales Tota" Infectious and parasitic diseases 114 84 198 5.5 3.9 4.7 Acute respiratory infections 68 58 126 3.3 2.7 3.0 Tuberculosis 21 6 27 1.0 0.3 0.6 Neoplasms 375 312 687 18.2 14.4 16.2 Circulatory and certain degenerative diseases 1,038 1,437 2,475 50.3 66.4 58.5 Ischemic heart disease 469 598 1,067 22.7 27.5 25.2 Cerebrovasculardisease 255 448 703 12.4 20.7 16.6 Diabetes 12 21 33 0.6 1.0 0.8 Complications of pregnancy 0 3 3 0.0 0.2 0.1 Perinatal conditions 35 23 58 1.7 1.1 1.4 Chronic obstructive lung diseases 77 54 131 3.7 2.5 3.1 Injury and poisoning 248 93 341 12.0 4.3 8.1 Ill-defined causes 32 33 65 1.6 1.5 1.5 All other causes 146 126 272 7.1 5.8 6.4 Total 2,065 2,165 4,230 100.0 100.0 100.0 Source: Calculared from WHO mortality database. cigarette smoking has been prevalent for several decades, it has tory diseases and to certain degenerative diseases. Circulatory been estimated that roughly one-third of cancer deaths can be diseases alone claim 5.45 million lives each year, or almost directly attributed to cigarette smoking (Doll and Peto 1981). exactly 50 percent of the total. More females than males die Recent estimates (WHO 1991a) have attributed 42 percent of from circulatory and certain degenerative diseases (3.21 mil- all male cancer deaths and 8 percent of female cancer deaths lion, compared with 2.72 million), although among what in the industrial countries to cigarette smoking. Dietary factors might be termed "premature deaths," males predominate, with are also thought to account for a similar proportion (one-third) roughly 785,000 males succumbing each year to these diseases of cancer deaths. Other behavioral factors have been causally before the age of sixty-five, compared with 390,000 females. associated with certain sites of the disease, including excessive Of the circulatory diseases, the principal cause of death is alcohol consumption (esophagus, pharynx), reproductive and ischemic heart disease, which each year claims 2.4 million sexual behavior, occupation, and pollution. lives, 1.9 million (or roughly 80 percent) of which occur among By far the leading type of cancer causing death in the those age sixty-five and over. The numbers of deaths are industrialized world today is lung cancer. Almost 500,000 lung roughly evenly divided between males and females, although cancer deaths were diagnosed in industrial countries in 1985, premature death from the disease (that is, before age sixty-five) three-quarters of them among males. Other leading types of is much (in fact, three times) more common among males. cancer include stomach cancer (380,000 deaths), breast cancer There has been considerable research into the causes of is- (165,000 female deaths), and prostatic cancer (95,000 male chemic heart disease and a number of risk factors have been deaths). Overall, since 1950, there has been relatively little identified, the principal ones being hypertension, cigarette change in nonlung cancer mortality-death rates have risen smoking, and elevated serum cholesterol. In countries such as slightlyformalesanddeclinedslightlyforwomen.Lungcancer the United States, Australia, and Canada, where health pro- mortality, almost all of which can be attributed to cigarette motion campaigns to reduce the prevalence and severity of risk smoking (USDHHS 1989), has risen dramatically in industrial factors in the population have been in operation for several countries during the last fortv years or so (see figure 2-1), years, marked declines in death rates, of the order of 30 to 50 although there are signs that the epidemic, at least among percent, have been observed since the late 1960s. males, has stabilized in several countries, including the United Another leading cause of death from circulatory disease is States, Australia, Switzerland, and the former Federal Repub- stroke (cerebrovascular disease), which each year claims the lic of Germany (Lopez 1989). Male lung cancer death rates in lives of 1.5 million persons in industrial countries, 60 percent these countries may soon begin to decline, as they have already of whom are women. Most of these deaths occur at the ad- begun to do in England, Wales, and Finland. Among women, vanced ages with slightly less than 240,000 deaths occurring by contrast, death rates from lung cancer are rising virtually before age sixty-five. The sex differential in premature mortal- throughout the industrialized world as a result of the wide- ity from stroke is less marked than for ischemic heart disease, spread adoption ofcigarette smoking among women during the the proportion of male deaths being only marginally higher 1950s and 1960s. than female deaths (55 percent compared with 45 percent). More than one-half of all deaths in industrial countries (5.93 The principal other cardiovascular disease causing death is million, or 54 percent) are attributed each year to the circula- nonischemic heart disease, including pulmonary and hyper- 40 Alan D. Lopez Figure 2-1. Relative Mortality Trends from Selected Causes of Death, by Gender, in Industrial Countries, 1950-86 Mortality i(330) 280 280- 260 Males 260 Females 240 / 240- 220 - 220 _ / ..I 200 200 180 180 / _ ~~/ , i/'". 160 / 160 / 140 /140-/ 120 120 100 ,- - 100 - 80 Lung Cancer 80 | Motor vehicle accidents Suicide 60 All other cancers 60 | _" ----- Ischemic Heart diseases - ----- Stroke O I I I I II I I*O I I I I I I I I 1950 55 60 65 70 75 80 85 1950 55 60 65 70 75 80 85 Year Year Note: Mortality in the base period, 1950-54, is set at 100, and mortality in other years is determined in relation to this base. Source: Lopez 1990b. tensive heart disease. This category of causes accounted for the percent) of the deaths assigned to this category occurred be- death of an additional 1.5 million persons in 1985, almost yond age sixty-five. 900,000 of whom were females. These diseases were also char- The remainder of this broad category of circulatory and acterized by a high average age at death-1.3 million (87 degenerative diseases consists of four specific conditions; cir- Causes of Death in Industrial and Developing Counmes: Estimates for 1985-1990 41 rhosis of the liver, diabetes mellitus, ulcer of the stomach and The quality of cause-of-death statistics for industrial coun- duodenum, and nephritis and nephrosis. Roughly 480,000 tries as a whole is reflected in the relatively low proportion of deaths are coded each year to these conditions, of which deaths for which a specific diagnosis could not be offered.5 In diabetes and cirrhosis of the liver each claim about 150,000. 1985, the number of deaths coded to the category of signs, Interestingly, the sex ratio of mortality is reversed for the two symptoms, and ill-defined conditions was approximately diseases. Twice as many females die of diabetes than males, 250,000, of which 180,000 were deaths among the elderly. whereas twice as many males die of cirrhosis of the liver. This Certainly the ascertainment of a single underlying cause of outcome is certainly consistent with observations on the alco- death in the presence of multiple pathologies, as is often the hol consumption patterns of men and women (Capocaccia and case for death at advanced ages, is difficult, and the verdict of Farchi 1988). senility is sometimes applied. Ill-defined conditions account The substantial improvements in public health in industrial for only about 2.3 percent of deaths overall in the industrial countries during the course of the twentieth century have world, and even among the elderly, where they are most virtually eliminated pregnancy-related deaths among women. common, the percentage is essentially the same. Slightly more In 1985 only about 4,500 such deaths were reported in the female than male deaths are coded to ill-defined conditions, industrial countries, the vast majority (80 percent) in the primarily because of the higher average age at death for nonmarket group. Still, misdiagnosis of maternal deaths is women. estimated to be of the order of 50 percent in industrial coun- tries (Royston and Lopez 1987), and thus the true mortality Industrial Market and Nonmarket Mortality Compared from maternal causes is probably closer to 6,000 deaths per year. Virtually all these deaths occur between the ages of Tables 2-3 through 2-5 are obviously not appropriate for mak- fifteen and forty-four and are largely preventable. ing mortality comparisons between the industrialized market Perinatal conditions claimed the lives of 100,000 babies in and nonmarket countries in view of differences in population the industrialized world in 1985, 60 percent of whom were size and structure between the two groups. Of the 1.17 billion boys. In almost all cases these deaths occur during the neonatal population living in industrial countries, roughly two-thirds period (up to twenty-eight days after birth) and to a large (760 million) live in the market economies and the remaining extent are due to congenital anomalies, birth trauma, and one-third in the nonmarket countries. In order to control for other circumstances of birth. Given the constitutional (rather differences in population size and composition, age-standard- than environmental) nature of these deaths, it is difficult to ized death rates have been computed for various causes of see how this mortality can be reduced much further, although death, using the "European" population age structure as the the much more common occurrence of such deaths in the standard. The results are shown in table 2-6. nonmarket countries suggests that there is still scope for a For both males and females, total mortality rates are about reduction in death rates in this group. 40 percent higher in the nonmarket countries than in the A further 1.17 million deaths in 1985 were ascribed to other market economies. Death rates from infectious and parasitic specific diseases, of which the chronic obstructive pulmonary diseases are low (of the order of 30 to 60 per 100,000 popula- diseases (primarily chronic bronchitis and emphysema) ac- tion) in both groups of industrial countries but are nonetheless counted for almost 400,000. Of these, roughly 250,000 were still higher in nonmarket countries, especially for males. These male deaths. This group of conditions, for which the principal higher rates are largely due to the higher tuberculosis mortality risk factor is cigarette smoking (USDHHS 1989), also tends to among the elderly in the nonmarket group. Interestingly, claim most lives at the higher ages, with 80 percent of deaths overall death rates from neoplasms are virtually identical for occurring among people age sixty-five and over. males in the two country groups but are higher for females in The other remaining broad category of causes of deaths, the market economies. This no doubt reflects the rapid rise in namely, injury and poisoning (external causes), accounted for female lung cancer mortality in several countries such as the about 770,000 deaths in the industrialized world in 1985. Male United States, Australia, Denmark, and the United Kingdom, deaths from violent causes (536,000) were more than twice as a phenomenon which is much less apparent in Eastern Europe. common as female deaths from such causes (236,000). More- Circulatory and certain degenerative diseases contribute over, roughly one-half (260,000) of the male deaths occurred most to the higher overall death rates observed in the non- between the ages of fifteen and forty-four, and roughly half that market countries. The age-standardized death rate for males in number again (147,000) between forty-five and sixty-four. these countries is almost 850 per 100,000, about 65 percent Only about 17 percent (92,000) of all male deaths due to higher than the level observed in the market economies. violence occurred at age sixty-five and over. There is thus Among women, the differential is even greater, with death considerable scope for additional preventive measures to rates from these diseases in Eastem Europe being 80 percent counter premature male mortality from violence (principally higher overall than in other industrial countries. Moreover, motor vehicle accidents) which each year accounts for almost mortality rates in the nonmarket countries are uniformly 10 percent of all male deaths in industrial countries. Mortality higher for all major component diseases of the circulatory from violent deaths among females is much lower (4.3 per- system, in particular ischemic heart disease and stroke, and for cent), with almost 50 percent of these deaths occurring among chronic obstructive lung diseases. By contrast, diabetes death women age sixty-five and over. Accidental falls are a leading rates are higher in the market economies, although the net cause of death from violence among elderly women. effect of this differential on comparative mortality for the 42 Alan D. Lopez Table 2-6. Age-Standardized Death Rates (per 100,000) for Selected Causes, Industrial Country Groups, 1985 Market economzes Nonrarke economies Total Cause of death Males Females Males Fenales Males Fernales Infectious and parasinc diseases 48.3 28.5 62.0 36.1 57.5 33.9 Tuberculosis 2.7 0.8 12.7 2.6 6.1 1.5 Acute respiratory infections 37.3 22.2 37.6 24.8 41.0 25.2 Neoplasms 264.6 154.4 266.3 136.1 268.7 149.2 Circulatory and certain degenerative diseases 516.7 323.3 847.5 582.2 605.6 405.3 Ischemic heart disease 222.4 108.8 382.1 241.0 265.1 149.5 Cerebrovascu lar disease 102.7 83.1 210.7 180.0 131.9 114.4 Other cardiovascular diseases 138.1 98.7 203.4 134.5 155.1 110.4 Diabetes 14.2 14.0 8.5 8.6 12.8 12.3 Otherdegenerativediseases 39.2 18.6 42.8 18.1 40.7 18.7 Complications of pregnancy n.a. 0.2 n.a. 1.6 n.a. 0.7 Perinatal conditions 7.5 5.8 14.3 9.7 10.5 7.5 Injury and poisoning 79.5 31.8 137.0 41.9 99.2 35.7 Ill-defined causes 26.8 18.7 26.2 14.2 26.5 17.3 All other causes 127.5 72.7 153.4 71.9 136.6 73.6 Total 1,070.9 635.3 1513.6 897.2 1,206.5 724.3 n.a. Not applicable Note: Data are standardized onco the WHO "European" population structure. Souirce: Calculated from WHO mortalitv database. category as a whole is negligible in view of the relatively low Central-Eastem European pattem to a structure more typical mortality from diabetes. The remaining degenerative diseases of Asian countries. At best, therefore, the broad mortality (which include cirrhosis of the liver) exhibit virtually identical comparisons presented here provide a summary perspective on mortality levels in the two groups of countries. health conditions for roughly one-quarter of the world's pop- The only other significant category of causes of death for ulation. which the mortality differential is quite substantial is injuries and poisonings, especially among males. In the nonmarket Mortality in Industrial Countries: Update for 1990 countries, the male death rate from violence of 137 per 100,000 is more than 70 percent higher than the average in the other More recent data for the late 1980s and, for several countries, industrial countries. Among females, mortality from violence 1990, are now available at WHO. Not surprisingly, the cause has traditionally been much lower than for males, and hence pattern ofmortality in 1990 ismuch thesame asfor 1985,given the relatively large differential in favor of women in the market the relative inertia of a mortality structure dominated by the economies (approximately 25 percent) has comparatively lit- chronic diseases. tle effect on the overall mortality differential among females in the two groups of countries. CANCER. In 1990 there were 2.42 million deaths-1.35 Certainly, a more detailed investigation of specific causes of million males, 1.07 million females. The leading type was lung death would help to shed more light on the underlying factors cancer (400,000 males deaths, 120,000 female deaths), fol- which contribute to the mortality differentials observed be- lowed by colon-rectum cancer (276,000 deaths, both sexes tween the market and nonmarket countries. Differences in combined), stomach cancer (244,000), breast cancer (175,000 individual lifestyle, including cigarette consumption, diet, and women), and prostatic cancer (105,000 men). general health consciousness, no doubt account for a substan- tial proportion of the differences in mortality. The mortality CARDIOVASCULAR DISEASES. During 1990, 5.43 million rates are aggravated by more pervasive environmental factors, deaths were coded to this category-2.46 million males and such as pollution and occupational hazards for certain cancers 2.97 million females. By far the largest category was ischemic and respiratory diseases. Yet any interpretation of aggregate- heart disease (2.33 million), followed by cerebrovascular dis- level mortality differences such as those outlined above must eases (stroke) (1.48 million). Other (nonischemic) heart dis- take into account the substantial heterogeneity of the popula- eases claimed 1.06 million lives in 1990. tions being compared. This is particularly true for the non- market couLntries. The former U.S.S.R., which alone CERTAIN DEGENERATIVE DISEASES. Diabetes mellitus caused accounted for two-thirds of the population of the industrialized 170,000 deaths in 1990; ulcer of stomach and duodenum, nonmarket economies, was itself a very heterogeneous country 50,000; cirrhosis of the liver, 175,000; nephritis, nephrotic with mortality profiles for subpopulations which vary from the syndrome, and nephrosis, 125,000. Causes of Death in Industrial and Developing Countries: Estirnates for 1985-1990 43 CHRONIC OBSTRUCTIVE PULMONARY DISEASES. In 1990 there occur within a complex epidemiological environment. Chil- were 388,000 deaths, 240,000 among men. dren are often afflicted with multiple infections, which in turn are aggravated by malnutrition and poverty. The estimation of INJURIESANDPOISONINGs.Duringl990,some865,000deaths mortality attributable to a single underlying cause is thus occurred in industrial countries from extemal causes, the ma- extremely difficult in developing countries, where infectious jority (615,000) among males. Motor vehicle accidents diseases are still common. One must be prepared to accept a claimed the lives of 215,000 persons, of whom 160,000 were considerable degree of overlap between estimates for specific males. The other leading cause of violent death was suicide, diseases (there is a parallel in the industrial countries but at with 140,000 male deaths and 50,000 female deaths in 1990. the other extreme of life, that is, at the advanced ages, when there are often several pathologies present at or about the time Estimated Cause-of-Death Patterns in 1985 of death). As a consequence, the estimates of mortality from leading The estimates for specific causes of death in developing coun- causes of death in developing countries are presented here tries are summarized in table 2-7 for children and adults sepa- along with these disease interactions. This is clearly a depar- rately and are discussed in more detail in this section according ture from the convention of specifying a single underlying cause to their etiology. Table 2-8 provides an estimated distribution of death, but it is no doubt much closer to the reality which of mortality by broad cause groups within each of four geo- prevails in many parts of the developing world. vMeasles is a graphic regions, which together encompass the entire devel- case in point. The most common complications of measles oping world. The method of estimation and the sources used include pneumonia, diarrhea, and malnutrition. Studies in are also given in this section. It is immediately apparent that Latin America (Puffer and Serrano 1973) and Africa (Ofosu- there is a substantial degree of uncertainty in the mortality Amaah 1983) have shown that many measles deaths were in estimates for specific causes, and hence the estimates must be fact attributed to complications of the disease, resulting in viewed with considerable caution, particularly for individual considerable underascertainment. The quantification of mea- diseases. sles mortality shown in table 2-7 explicitly recognizes these Quite apart from these more methodological considerations, relationships between underlying, immediate, and associated the vast majority of childhood deaths in developing countries causes of death-the global estimate of 2 million measles Table 2-7. Estimated Causes of Death in Developing Countries, by Age, 1985 (thousands) Age Cause of death Under five Five and over All ages Infectious and parasitic diseases 10,500 6,500 17,000 Diarrheal diseases 4,000 1,000 5,000 Tuberculosis 300 2,700 3,000 Acute respiratory diseases 4,300 2,000 6,300 Measles, whooping cough, and diphtheria 1,500 - 1.500 Other acute respiratory diseases 2,800 2,000 4,800 Other measles and whooping cough' 700 - 700 Malaria 750 250 1,000 Schistosomiasis - 200 200 Other Infectious and parasitic diseases 450 350 800 Complications of pregnancy -500 500 Perinatal conditions 3,200c - 3,200 Neoplasms - 2,500 2,500 Chronic obstructive lung diseases - 2,300 2,300 Circulatory diseases and certain degenerative diseases - 6,500 6,500 External causes 200 2,200 2,400 Other and unknown causes 700 2,800 3,500 Total 14,600 23.300Q' 37,900 -Negligihle a. Does not include 400,000 measles-related deaths included under diarrheal diseases. bh Some of these dieaths actually may be attributable to malaria. The global estinmate of mortality from the disease is between I and 2 million annually. c. Includes an estimated 775,000 deaths from neonatal tetanus. d. Of these, an estimated 1.6 million deaths occur at ages 5 to 14 years. Diarrheal diseases and acute respiratory disease are each estimated to account for abouit 300.000 deaths at these ages; another 1 50,000 or so are dLue to malaria and tuberculosis. Accidents and violence are a leading cause od death at these ages and niay well claim 200,000 to 300,000 lives each year in this age group in developing countries. Source: Aujthor's estimates. 44 Alan D. Lopez Table 2-8. Estimated Causes of Death in Developing Countries, by Region, 1985 (thousa"nds) Latin America Sub-Saharan Middle East! Catse of death and the Caribbean Africa North Africa Asia Total Infectious and parasitic diseases 900 4,500 2.400 9,200 1,700 Neoplasms 300 250 200 1,750 2,500 Circulatory diseases and certain degenerative diseases 900 650 550 4,400 6,500 Complications of pregnancy 35 125 80 260 500 Perinatal conditions 300 680 420 1,800 3,200 Chronic obstructive pulmonary 90 60 50 2,100 2,300 diseases Extemal causes 250 350 200 1,600 2,400 Other and unknown causes 425 585 400 2,090 3,500 Total 3,200 7,200 4,300 23,200 37,900 Source: Author's estimates. deaths in 1 985 has been disaggregated into estimates of the hood deaths, although the actual underlying cause of death components of the disease-complication intera-tion. Many of may have been some other condition. Childhood mortality in these deaths would in tum be associated with malnutrition. many cases is the result of multiple infections, often aggravated Several other examples could be cited, including the com- by malnutrition, and the attribution of death to a single under- plexity of malarial infection. Severe anemia is often a conse- lying cause is often extremely difficult in these circumstances. quence of repeated attacks of malaria but can also result from In the absence of more precise diagnostic information about hookworm infection and nutritional deficiencies. Although the underlying cause of death, however, the estimate of 4 this epidemiological complexitv makes the estimation of mor- million childhood deaths will be taken as indicative of the tality from specific diseases uncertain, there are clear im- volume of mortality due to diarrheal diseases, with about 10 plications for health interventions. Hlealth care programs percent of these (that is, 400,000) arising as a complication of primarily designed to control the spread of infection from a measles. singlediseasecanbeexpected toexertadisproportionateeffect Snyder and Merson do not provide estimates of diarrheal on child survival by simultaneously reducing mortality from disease mortality in the population age five and over. Still, the associated causes. studies which they reviewed suggest an annual death rate from diarrheal diseases in the adult population of about 1.4 per 1,000 Infectious and Parasitic Diseases in 1980. Applying this estimate to the estimated adult popu- lation in developing countries (excluding China) yields a total The basis for the estimates of mortality from specific infectious of about 3.1 million adult deaths. By contrast, age-specific data diseases is given below. Although joint estimates are provided, on morbidity from diarrheal diseases reported by Snyder and the very poor quality of the data and information available to Merson suggest about 0.2 episodes per adult per year. Assuming make them suggests that they be viewed extremely cautiously. the same case-fatality ratio as estimated for children (0.3 Confidence intervals of the order of 50 percent around each percent), this yields an estimate of about 1.3 million adult estimate would seem reasonable. deaths. A third way to estimate diarrlheal disease mortality is to use figures on the age-specific death rates from these diseases DIARRHEAL DISEASES. In their review of morbidity and mor- observed for the industrial countries at levels of life expectancy tality from diarrheal diseases based on twenty-four studies in comparable to that of contemporary developing countries, the developing world, Snyder and Merson (1982) estimated excluding China (Preston 1976, p. 93). This method suggests that there were roughly 4.6 million childhood deaths (below a total of about 700,000 adult deaths and, interestingly, a total age five) each year associated with these diseases. Allowing for of about 3.9 million deaths below age five, or roughly the same the effect of oral rehydration treatment in the meantime, and estimate obtained earlier. In view of this proportionality be- for population growth, the annual number of deaths is cur- tween child and adult deaths, and the more recent evidence rently thought to be about 4.4 million. However, evidence on used in the method of case-fatality rate per episode, a figure of the case-fatality rate from the disease and the number of about I million adult deaths from diarrheal diseases would episodes (1,300 million per year) suggests just under 4 million seem plausible, with about 300,000 of these occurring at age childhood deaths. five through fourteen, based on age-specific fatality rates re- The distinction between "association" and "cause" is em- ported in community studies (Kirkwood 1990a). phasized by the authors. Acute diarrheal diseases are thus These estimates are unlikely to be drastically altered by the estimated to be associated with approximately 4 million child- addition of mortality in China. On the basis of the information Causes of Death in Industnal and Developing Countnes: Estunates for 1985-1990 45 for reporting areas, the number of diarrheal disease deaths the 2 million measles deaths, are due to AR]. Similarly, about among adults is unlikely to exceed 50,000 per year, which is two-thirds (or 400,000) of the pertussis deaths in 1985 were well within the margin of uncertainty of the global estimates also attributable to AR!. Evidence from the industrial countries derived above. at an earlier stage of the epidemiological transition suggests a mortality ratio of about 0.7 between acute respiratory diseases TUBERCULOSIS. The experience of the WHO Tuberculosis (influenza, pneumonia, bronchitis) and diarrheal diseases. On Control Programme suggests that the most reliable estimates this basis, the volume of mortality from the remaining respira- of mortality in developing countries are obtained from the tory diseases should be about 2.8 million deaths of children case-fatality ratio among detected cases. This is predicated on under age five. This yields a total of about 4.3 million for the knowledge that without appropriate chemotherapy, tuber- ARi-related deaths under age five. culosis infection is highly fatal. A review of studies on the Acute respiratory infections, particularly pneumonia, also proportion of cases detected that are also treated suggests that, claim a substantial number of lives at older ages. A comparison overall, the case-fatality rate of the disease in developing of age-specific death rates among adults from influenza, pneu- countries is probably on the order of 15 percent (Murray, monia,andbronchitiswiththosefromdiarrhealdiseases(Pres- Styblo, and Rouillon, chapter 11, this collection). Applying ton 1976) suggests that death rates from the respiratory this to incidence data yields an estimate of about 3 million category are roughly two to four times higher. This would imply tuberculosis deaths each year in developing countries. Approx- an annual toll of between 2 million and 4 million adult deaths imately 200,000 of these deaths occur in China, leaving a total from acute respiratory diseases in developing countries each for other developing countries of 2.6 million deaths. Murray, year.6 Data from China, India, and Latin America, however, Styblo, and Rouillon (1989) estimate that about 15 percent of suggest that the lower limit of this range is more reasonable, these deaths (or 450,000) occur below age fifteen and of these, and hence an estimate of 2 million deaths beyond age five from about two-thirds (or 10 percent of the total) are deaths of these diseases is proposed. According to survey data for Sub- children under five. Thus roughly 300,000 childhood deaths Saharan Africa reported by Kirkwood (1990b), about 300,000 (birth through age four) which occur each year in developing of these deaths occur from age five through age fourteen. countries are estimated to be due to tuberculosis. VACCINE-PREVENTABLE DISEASES. Separate estimates for three ACUTE RESPIRATORY INFECTIONS (ARI). A review of informa- of these diseases (poliomyelitis, tuberculosis, and neonatal tion available to WHO on causes of mortality in young children tetanus) are presented elsewhere in this section. The over- suggests that between 25 and 30 percent of deaths among the whelming majority of deaths from the remainder are from under-fives are attributable to these diseases (Leowski 1986). measles and whooping cough. As mentioned earlier, measles This estimate is supported by results obtained from longitudi- probably claimed about 2 million lives below the age of five in nal mortality surveys of communities conducted in Nepal, 1985. Of these deaths, 1.1 million were also associated with Pakistan, the Philippines, and Tanzania, using verbal autop- ARI and another 400,000 with diarrhea. Whooping cough is sies. Despite the caveats associated with this type of estimation estimated to have killed about 600,000 children in 1985, and procedure, it is probable that acute respiratory infections ac- 400,000 of these deaths were likewise associated with ARI. count for about 4.3 million child deaths (that is, from birth Taken together, the vaccine-preventable diseases either were through age four) each year in developing countries. The the cause of, or were closely associated with, 3.7 million deaths rationale underlying this estimate is outlined below. of young children in 1985, of which 2 million were from Essentially, the group of acute respiratory infections can be measles, 600,000 from whooping cough, 300,000 from tuber- classified into two broad subcategories, namely, certain vac- culosis, about 800,000 from neonatal tetanus, and 25,000 from cine-preventable diseases (measles, tuberculosis, whooping poliomyelitis. cough. and diphtheria) and other respiratory diseases (primar- ily pneumonia, influenza, acute bronchitis, and bronchiolitis). POLIOMYELITIS. The immunization coverage rate for this Separate estimates of mortality are available for the compo- disease is about 70 percent globally (WHo 1989). It is estimated nent diseases of the first category. According to estimates by WHO that about 70 percent of the world's population lives prepared by the WHO Expanded Programme on Immunization, in polio-endemic areas and that the annual incidence of the there were approximately 2 million measles deaths and some disease is about 250,000 cases. Overall, 25,000 deaths per year 600,000 deaths from whooping cough in 1985 among children are estimated to occur from the disease. in developing countries. These estimates were prepared on the basis of assumptions about vaccine effectiveness (95 percent YELLOW FEVER, DENGUE, AND ENCEPHALITIS. The estimation of for measles, 80 percent for whooping cough), susceptibility of annual mortality from these diseases is particularly difficult the unexposed population (100 percent for measles, 80 percent because of their epidemic nature, which results in considerable for whooping cough), estimated coverage by immunization, fluctuations from year to year. The estimated total mortality and case-fatality rates. from these diseases in 1985 was about 15,000, but it should be Informationfromcommunity-levelstudiesavailabletowHo noted that yellow fever in Nigeria was comparatively low in indicates that about 1.1 million, or slightly more that half of that year. Since the number of yellow-fever deaths in Nigeria 46 Alan D. Lopez during epidemic years can easily reach 10,000, the figure of and parasitic diseases, including amebiasis, hookworm, AIDS, 15,000 annual deaths from these diseases has been increased and hepatitis B, undoubtedly claim several hundred thousand to 20,000 in an attempt to allow for epidemic variations. lives each year (neonatal tetanus is considered with the group of perinatal conditions). It is estimated on the basis of regional MALARIA. Estimates of malaria mortality in Africa made estimates for the major component diseases of this category some thirty years ago suggested that there were about 1 million that of these 17 million, slightly more than one-half occur in deaths each year from the disease. Recent studies, based on Asia and about one-quarter in Sub-Saharan Africa. The lowest active surveillance and intervention projects in Africa, suggest mortality is estimated for Latin America, where about 900,000 that the global total number of deaths is probably in the range deaths are estimated to have occurred in 1985 from infectious of 1 to 2 million deaths per year. There is much more confi- and parasitic diseases. dence in this range than in a point estimate, but for the purposes of this study, and without any additional guidance as Complications of Pregnancy, Childbirth and the to what part of the range is more probable, the lower limit has Puerpenrium, and Perinatal Conditions been chosen, yielding an estimate of 1 million deaths per year. Of these, about 500,000 are estimated to occur among children In this section, separate estimates have been provided for two in Africa. The widespread and relatively indiscriminate use of broad categories of cause of death since they are restricted to chloroquine as practiced in Africa, keeps mortality down but specific population groups, namely pregnant women (maternal also favors the selection of chloroquine-resistant parasites. As causes) and newborn infants (perinatal conditions). chloroquine resistance increases in geographic extension, fre- quency, and intensity, there is a serious threat of rising mortal- MATERNAL MORTALITY. It is well known that registered data ity, because there is no alternative drug that is equally safe and on maternal deaths generally underestimate the extent of cheap. matemal mortality, even in industrial countries (see Ziskin and Data from lay reporting of causes of death in rural India others 1979; Smith and others 1984; Rubin and others 1981). suggest that about I percent of deaths in India, or about Mortality models based on these data, such as models of the 150,000, are due to malaria. Half of these deaths occur among epidemiological transition, will therefore tend systematically infants and young children. On the basis of age-specific mor- to underestimate deaths due to complications during preg- tality data for certain endemic countries, malaria is also esti- nancy or the birth process, irrespective of the overall level of mated to account for about 150,000 of the 1.6 million female mortality. Typically, the higher the level of female childhood deaths each year at age five through fourteen. mortality, the greater the underestimation of deaths from such causes. As an alternative procedure, selected community-wide SCHISTOSOMIASIS. The prevalence of schistosomiasis is esti- studies have been evaluated (see, for example, Fortney and mated at about 200 million people and is endemic in seventy- others 1986; Royston and Lopez 1987) in order to estimate the six countries. Information about the severity and age-specific relation between the overall level of female mortality and prevalence of the disease, however, indicates that the upper mortality from pregnancy complications or the birth process limit of the estimated number of persons with severe infection in various sociocultural settings. Applying these community- is of the order of 13 million. Assuming a 0.1 percent case-fa- based estimates to estimates of the number of births in major tality rate (limited to cases of severe infection), this implies a regions leads to an overall estimate of approximately 500,000 global total of about 13,000 deaths per year. Walsh (1988, p. maternal deaths in developing countries in 1985. Of these, 15) has estimated an annual mortality of between 250,000 and an estimated 35,000 occurred in Latin America and the 500,000, apparently based on the assumption of a case-fatality Caribbean, 125,000 in Sub-Saharan Africa, 80,000 in the rate for all cases (nor only severe manifestations) of between Middle East and North Africa, and about 260,000 in Asia and 0.1 and 0.25 percent. Thus the range of mortality estimates Oceania. varies from 13,000 to 500,000. An approximate mid-point of the range (200,000 deaths) may be taken as a rough guide to PERINATAL CONDITIONS. On the basis of community-level the annual toll of mortality but, as with many other of the data, wiio has estimated the annual number ofperinatal deaths diseases under consideration here, the degree of uncertainty is each year to be approximately 7.3 million, of which only about substantial. Information on the likely range of estimates is 300,000 occur in industrial countries (wHO 1989). Of the remain- probably of greater relevance for establishing health priorities ing 7 million perinaral deaths, the proportion which are early than these attempts at providing more precise figures. neonatal deaths (birth to six days) appears to vary between 40 and 50 percent in developing countries, according to statistics SUMMARY FOR INFECTIOUS AND PARASITIC DISEASES. Summing published in the United Natons Demographic Yearbook. This up the estimates for the diseases listed above yields an annual percentage implies an estimate of about 3.2 million early mortality of about 17 million deaths (see table 2-6), about 10.5 neonatal deaths, almost all of which can probably be attributed million of which are estimated to occur among infants and to one of the perinatal conditions. Neonatal tetanus alone children less than five years of age. These are undoubtedly the would account for about one-quarter (or roughly 800,000) of most significant communicable diseases, but other infectious these deaths. Out of a total of 3.2 million deaths, almost 60 Causes of Death in Industrial and Developing Countries: Estimates for 1985-1990 47 percent (1.8 million) are estimated to have occurred in Asia, incomplete, yield an estimate of about 300,000deaths per year, 680,000 in Sub-Saharan Africa, 300,000 in Latin America and which is just under 10 percent of all deaths. The information the Caribbean, and about 420,000 in the Middle East-North on overall life expectancy in Sub-Saharan Africa and the Africa region (see table 2-8). Middle East-North Africa region, and the regional incidence estimates reported by Parkin, Laara, and Muir (1988), suggest Chronic Diseases and Violent Death levels of cancer mortality comparable to what was observed for India. Such levels yield another 450,000 cancer deaths (neo- Specific mortality estimates for the major non-communicable plasms) in these two regions (table 2-8). diseases and violence are discussed below. Although the etiol- On the basis of the incidence levels for cancer reported by ogy of the constituent diseases (e.g. different types of cancer) Parkin, Laara, and Muir (1988), it may be estimated that there can vary considerably, arguing for more specific estimates, only were some 300,000 to 400,000 deaths from cancer for the broad categories of causes are discussed here in view of the remainder of Asia in 1985. The addition of these deaths yields uncertain diagnostic accuracy of the cause of death informa- an overall total of 2.5 million cancer deaths each year in the tion upon which they are based. developing world. This process of aggregation thus leads to an estimate of cancer mortality which is identical to that esti- NEOPLASMS. On the basis of incidence data reported to the mated from the mortality-to-incidence ratio method. The fact International Agency for Research on Cancer, Parkin, Laara, that the two estimates do not differ should not be seen as and Muir (1988) have estimated that there were a little over necessarily a verification of either method. The evidence, 3.2 million new cases of cancer in the developing world in however, would seem to suggest that the annual number of 1980. This corresponds to an incidence rate of 94.5 per cancer deaths in the developing world is at least 2.5 million. 100,000. Applyingthisfiguretotheestimated 1985 population On the basis of incidence data, the principal sites of cancer yields a total of about 3.6 million new cases in 1985. This may mortality in the developing world are stomach, mouth-pharynx, well be an underestimate, however, as population aging in the esophagus, and lung among males, and cervix, breast, stomach, developing world will certainly imply an increased burden of and mouth-pharynx (particularly in India) among females. illness from cancer, even if the relative levels of prevalence of risk factors were to remain unchanged. This is certainly not CHRONIC OBSTRUCTIVE LUNG DISEASE (COLD). Global and re- the case, as is evident from the dramatic increase in cigarette gional estimates for this category of diseases are particularly consumption in developing countries in recent years (WHO difficult because of the lack of reliable data for the majority of 1985). Nonetheless, the figure of 3.6 million new cases in 1985 developing countries. The mortality information available provides a reasonable, if conservative, benclhmark from which from China suggests that as much as 15 percent of all deaths to derive estimates of mortality. . are due to these diseases, which would imply about 1 million A very crude first approximation can be obtained from the deaths. In China, at least, there is some basis for expecting high observed relation between incidence and mortality in the COLD mortality in view of past smoking patterns, particularly industrial countries. In 1985 there were an estimated 3.25 among males, and the very high levels of indoor air pollution million new cases of cancer in the industrialized world and 2.3 emanating from the cooking and heating fuels used (World million deaths, yielding a mortality-to-incidence ratio of 0.7. Bank 1989). In India, data from the lay reporting system in If this ratio were to apply in developing countries, the esti- rural areas suggest a COLO) mortality figure of about 6 to 8 mated number of deaths wouLld he on the order of 2.5 million. percent, a proportion which is at least consistent with preva- Still, health services for cancer patients are undoubtedly more lence studies in specific communities (see, for example, Malik widely available in industrial countries and are probably more and Wahi 1978). This would imply an additional 700,000 to effective in treating the disease. One would therefore expect 800,000 deaths in India alone. Death rates of the order of 40 that not only is the average age at death from cancer lower in to 60 per 100,000 have been reported to wlio for other parts of developing countries but also the mortality-to-incidence ratio Asia. From these estimates, an additional 350,000 deaths from is probably higher than in industrial countries. Thus the an- COLD in the remainder of Asia and the Pacific are estimated for nual toll of cancer deaths is no doubt higher than the 2.5 1985. Data forLatin America suggest that at least90,000 adults million suggested by this method of estimation, but how much succumbed to these diseases in 1985. On the basis of compar- higher is rather uncertaini. ative life expectancy, the annual mortality in Sub-Saharan In China alone, there are about 1 million cancerdeaths each Africa and the Middle East-North Africa region is estimated year, according to the mortality data from reporting areas. Lay at 60,000 and 50,000, respectively, yielding a total for all reporting of the cause of death in rural India suggests that about developing countries of 2.3 million deaths. There is, however, 4 percent of all deaths in India are due to cancer, and this considerable uncertainty associated with this estimate. proportion is confirmed by studies carried out in communities in Andhra Pradesh; Goa, Daman, and Diu; and Maharashtra CIRCUJLATORY DISEASES AND SELECTED DEGENERATIVE D]SEASES. states. This would suggest an annual mortality from cancer in As for the chronic obstructive lung diseases, perhaps the best India of about 400,000 deaths, or 50 per 100,000 population. way to proceed in estimating global mortality for this group of National data for Latin America and the Caribbean, although diseases is by considering the situation in major regions sepa- 48 Alan D. Lope: rately. In China, available data suggested that there are at least were about 250,000 violent deaths in the region in 1985, this 1 million deaths from stroke alone each year and another being about 8 percent of all deaths. Evidence from other million deaths from all forms of heart disease. The estimated developing regions suggests that violence is less important as total for the four specific degenerative diseases is about a cause of death and a proportionate mortality of about 5 400,000, suggesting anoverall total for this category ofdiseases percent is probably not unreasonable. This assumption yields in China of about 2.5 million deaths per year. In rural areas of an estimate of 350,000 deaths in Sub-Saharan Africa, 200,000 India, on the basis of data from the Rural Cause of Death in the Middle East-North Africa region, and 350,000 in the Survey (Indian Office of the Registrar General 1989), diseases remainder of Asia. The estimated total number of violent of the circulatory system (primarily heart diseases and stroke) deaths is thus 2.4 million deaths per year for the developing account for about 10 percent of deaths. The remaining degen- world, of which about 8 to 9 percent (that is, 200,000) are erarive diseases make up an additional 2 percent or so of all estimated to occur before age five. deaths, suggesting an annual mortality from this category of about 1.1 million. Matermal and Childhood Mortality: Update for 1990 The group of countries that make up the remainder of Asia (excluding India, China, and westem Asia) has an average life The number and relative importance of deaths from chronic expectancy of about fifty-seven years, which, if one uses the diseases among adults in developing countries are likely to indirect mortality estimation techniques described by have altered very little between 1985 and 1990. But given the Hakulinen and others (1986), would imply a relative mortality natural history of these diseases, one might reasonably expect of about 12 percent due to the cardiovascular and other to see a significant change brought about by child health chronic degenerative diseases. This estimate yields a further interventions in this short period of time. According to WHCO 800,000 deaths in the remainder of Asia and the Pacific, or a (1992), this has indeed been the case. The number of infant grand total for Asia of 4.4 million deaths each year. and child deaths declined to about 12.9 million in 1990. In Data from countries of Latin America, when aggregated and large part, this decline has been attributed to the rapid im- adjusted for underreporting, yield another 800,000 to 900,000 provement in immunization coverage since the mid-1980s deathsfror these causes. Very little information on the extent (WHO 1992). In 1990, measles was estimated to have caused of mortality from cardiovascular disease is available for the 880,000 deaths of children under age five in developing coun- African and Middle East regions, but with life expectancy in tries, down from 2 million in 1985. Relatively few of these these areas typically on the order of fifty to fifty-five years, it is deaths were attributable to measles alone (220,000), the ma- difficult to see how less than 10 percent or so of deaths could jority arising from interactions with acute respiratory infec- be ascribed to these diseases. As a very rough estimate, accom- tions (480,000) and diarrhea (180,000). Similarly, whooping panied by a substantial degree of uncertainty, an additional 1.2 cough in 1990 was estimated to have caused 360,000 child million deaths have been attributed to these causes in the two deaths, 260,000 of which were estimated to have occurred in regions combined as shown in table 2-8. The global total for conjunction with acute respiratory infections. There has also this category is thus estimated at about 6.5 million deaths per been a dramatic decline in neonatal tetanus deaths, from year, of which the four specific degenerative diseases probably 800,000 in 1985 to 560,000 in 1990. claim between 600,000 and I million lives each year. By contrast, relatively little change has occurred in the other principal causes of child death. Allowing for the association INJURY AND POISONING. In almost all cases, the attribution of with measles, diarrheal diseases are estimated to have claimed death to a violent cause as opposed to a disease is relatively about 3.2 million children in 1990. Acute respiratory infec- straightforward. This is not to deny that the disaggregation of tions, primarily pneumonia, were the cause of 3.5 to 3.6 million violent deaths into accidents, suicides, homicides, and other child deaths in 1990; if one includes the deaths associated with violence is often quite complicated and may well be influenced measles and whooping cough mentioned above, this toll rises by sociocultural or legal factors, for example, mitigating against to about 4.3 million deaths a year. Mortality from other tradi- a verdict of suicide in some cultures. In this review attention tional diseases of childhood, such as malaria, has remained will be confined to the broad category of violent deaths, and essentially unchanged, as has the category of perinatal condi- hence for the most part, these considerations would not apply. tions (including neonatal tetanus), which, in 1990, are esti- Moreover, because of the relative ease of distinguishing a mated to have caused about 3 million early infant deaths. violent death from other causes, one can have greater confi- Finally, a recent update of the annual number of maternal dence in the data generated by alternative collection schemes, deaths in developing countries shows relatively little change such as lay reporting. compared with 1983-85 (WHO 1991b). In 1988-90 the annual In China, roughly 700,000 deaths from external causes are number of maternal deaths in developing countries was esti- estimated to occur each year, or about 10 percent of the total. mated at just over half a million (505,000), with an approxi- In India, the proportion is less, about 6 percent, or 550,000 mate 5 percent decline in risk being more than compensated deaths, from accidents and violence. Data for Latin America, for by a 7 percent rise in the number of women exposed to risk. after adjusting for incomplete coverage, indicate that there Maternal mortality appears to have increased by about 10 Causes of Death in Industrial and Developing Countries: Esnmates for 1985-1990 49 percent in Africa (170,000 deaths in 1988-90), declined within the group vary from high-mortality countries in Africa slightly in Latin America (25,000 deaths), and remained un- and Asia, where infant mortality rates exceed 200 per 1,000 changed in Asia. live births, to countries such as China, Cuba, Argentina, Chile, and Uruguay, where life expectancy is comparable to Summary and Conclusions that observed in many industrial countries. Not surprisingly, the cause-of-death structure in the former category is domi- The estimation of mortality levels, structure, and trends is nated by the communicable diseases, whereas in the latter fundamental to any assessment of the health situation. For group the chronic diseases are of most concern. Other devel- several countries, including virtually all the industrialized oping countries, with life expectancy in the range of about world as well as a number of countries in Latin America and fifty-five to sixty-five years, are intermediate in their progres- eastern Asia, reasonably reliable and comparable mortality sion through the epidemiological transition. Clearly, in these statistics are available to determine the age, sex, and cause of countries, as with very high-mortality populations where in- mortality. In the group of industrial countries, which together fant and child deaths continue to claim up to 50 to 60 percent made up about one-quarter of the world's population in 1985, of the total, health strategies must continue to focus on the the leading causes of death are those conditions for which prevention and control of infectious diseases, particularly the prevention is largely a matter of personal lifestyle. Diseases diarrheal diseases, respiratory infections, and the vaccine- such as ischemic heart disease, stroke, cancer, and, in particu- preventable diseases. Malaria, too, is a leading cause of child lar, lung cancer, the chronic obstructive lung diseases, diabe- death in some areas. Effective primary health care delivery tes, and cirrhosis of the liver, which dominate the mortality undoubtedly offers the greatest hope for success in rapidly pattern in industrial countries, all have a significant behavioral bringing down mortality levels for these diseases. component. Cigarette smoking, for example, is by far the But what is also clear is that the chronic diseases have principal risk factor for lung cancer, chronic bronchitis, and already emerged as a significant, if not the significant, health emphysema and is also causally associated to varying degrees problem in a sufficient number of developing countries to with ischemic heart disease, stroke, and several other types of warrant increasing attention. There are now at least as many cancer. Indeed, cigarette smoking is probably the leading cause cancer deaths in developing countries as in the industrialized of mortality in the industrial countries, claiming an estimated world and at least another 9 million deaths each year in 2.1 million lives each year (or 20 percent of the total) in these developing countries from other chronic diseases. To some countries. This number can be expected to rise to at least 3 extent, this is an inevitable consequence of progress toward the million by the 2020s as the full effect of the smoking epidemic conquest of infectious diseases. Still, the widespread adoption among women is felt (Peto and others 1992). of cigarette smoking in many parts of the developing world is By and large, the industrial countries, particularly those with rapidly eroding some of the gains in life expectancy which have market economies, have been very successful in deferring been achieved as a result. Although the annual amount of death to higher and higher ages. In these countries, the tobacco-attributable mortality in the developing world is cur- challenge will be to ensure a comparable delay in the onset of rently less than that ofthe industrial countries (about I million chronic disease, thus minimizing the proportion of the years of to 1.5 million deaths per year), this death toll is projected to life gained which are spent in a state of chronic morbidity or rise rapidly during the next three to four decades to reach 6 disability. Further reductions in inequalities in health status million to 8 million in the 2020s (Peto and Lopez 1991). A within national populations will also bring rewards in terms of substantial proportion of this premature mortality is expected gains in life expectancy, particularly for adults. Accidental to occur in China, where the dramatic increase in consump- deaths, suicide, and other violence continue to claim a signif- tion of manufactured cigarettes during the last twenty years or icant proportion (7 to 8 percent) of lives in industrial countries so (China now consumes one-quarter of the world total) can and constitute a very substantial cost to society in terms of be expected to result in 2 million to 3 million smoking-attrib- potential years of life lost. Behavioral factors-in particular, utable deaths in the 2020s, altnost a million of which will be alcohol abuse-similarly underlie many of these deaths. from lung cancer (Peto and Lopez 1991). Mortality rates from most leading causes of death are signif- The rapidly emerging epidemic of smoking-attributable icantly higher in nonmarket industrial countries. Indeed, in mortality is one of the principal public health issues which some cases, most notably Hungary, death rates from major developing countries will face during the next few decades, and chronic diseases among men have been rising for several years. in many cases it will be superimposed on a health system still Information on the prevalence and distribution of risk factors preoccupied with the control of infectious diseases. The inev- is clearly an essential component of health strategies designed itability of death thus implies that, although this transition to to counter these trends. chronic diseases is to be expected, what is important is that Although the health situation in developing countries has these deaths be postponed to as late in life as possible. As for been assessed collectively, it is absolutely imperative to keep the industrial countries, effective preventive strategies have a in mind the heterogeneity of this group when interpreting the much greater likelihood of achieving this than costly attempts estimates of the cause-of-death structure. Mortality levels at cure. 50 Alkn D. Loper Notes Lopez, A. D. 1990a. "The Interrelationship between Lung Cancer and To- bacco Consumption: Evidence from National Statistics-" In Matti Hakama, This chapter was prepared while I was a staff member of WHO's Dision of Valerie Beral, J. W. Cullen, and D- M. Parkin, eds., EvaluaingEffectiveness Global Epidemiological Surveillance and Health Situation and Trend Assess- of Prirnari Prevention for Canter. Scientific Publication 103, 1ARC-. I Von: ment. The views expressed here are mine and do not necessarily reflect the Intemational Agency for Research on Cancer. opinions or policies of the World Health Organization. Nonetheless, my - - 1990b. "Competing Causes of Death: A Review of Recent Trends in research has benefited from the contribition and comments of manv col- Mortality in Industrialized Countries with Special Reference to Cancer." leagues at WHO, particularly those in technical programs concerned with the Annals of the New York Academ-y of Science 609 (November 2):58-76. health problems specifically discussed in the chapter. Their contribution is Malik, S. K., and P. L. WaIst- 1978. "Prevalence of Chronic Bronchitis in a gratefully acknowledged. I shotild also like to express my sincere gratitude to Group Of North Indian Adults." Journal of the Indian Medical Association Dean Jamison for his insistence that this type of assessment shculd and could 70(1):6-8. he done, and for his very helpful remarks on several earlier versions of it. The Ofosu-Amaah, Samuel. 1983. "The Control of Measles in Tropical Africa: A paper has also benefited from a critical review by Tinso Hakulinen and Althea Review of Past and Present Efforts." Review of 1Ifecnious Diseases 5(3):546-53 . Hill. In acknowledging the very considerable assistance which I received in Parkin, D. M., E. Laara, and C. S. Muir. 1988. "Estimates of the Worldwide the preparation of this chapter. I also accept that any errors and omissions are Frequency ofSixteen Major Cancers In 1980." intertonalJouralofCancer asy responsibility. 41e184 97r -. The year 1985 has been chosen in order to present a mid-decade review. Whenever possible, estimates for 1990 have been included as well. Percy, Constance, and CallIm S. Muir. 1989. 'The International Compara- 2. The populations of the fonner Yugoslavia, the former German Demo- bility of Cancer Mortality Data: Results ofan Inremational Death Certifi- cratic Republic (a.DR), and the formerU.S.S.R. are shown in thischapter, since cate StUdy." Ainencan journal of Epidetniology 129(5 ):934-46. they were contnries at the time the data were provided to the World Health Peto, Richard, and A. D. Lopez- 1991. "Worldwide Mortality from Current Organization (WHO). Smoking Pattems." In Betty Durston and Konrad Jamrozik, eds., Tobacco 3. The only industrial country for which cause-of-death data about 1985 ind Health 1990. The Global War. Proceedings of the Seventh World are not available to WHO is Albania. This is unlikely to affect the results for Conference on Tobacco and Health, Perth, Australia. April 1-5, 1990- the group as a whole, since the total annual number of deaths in Albania is of Health Department of Western Australia. the order of 17,000, which is less than 0.2 percent of the total for the Peto.Richard,A.D.Lope-,JillianBoreham,MlichaelThun,andClarkHeath- industrialized world. 1992. "Mortality from Tobacco in Developed Countries: Indirect Estima- 4. Namely, cirrhosis of the liver, ulcers of the stomach and duodenum. tion from National Vital Statistics." Lancet (23 May) 1268-78. nephritis and nephrosis, and diabetes nellitus. Preston, S. H. 1976. Mortality Patterns in National Popzlations. New York: 5. This diagnostic specificity does not, however, guarantee international Academic Press. comparability due to the reasons mentioned earlier. Puffer, R. R., and C \-. Serrano. 1973. Patterns of Mortahltv in Childhood. 6. Some adult deaths from the vaccine-preventable disease. wotild also be Scientific Publication 262, PAHO (Pan-American Health Organization). expected, but these are likely to he comparatively few compared with adult Washington, D.C. merrta[ity from acmtie bronchitis and pneumonia. Rovston, Erica, and A. D. Lopez. 1987. "On the Assessment of Maternal Mortality." W'orld Health Statistics Quarterls 40(3):214-24. Rubin, George, Brian McCarthy, James Shelton, Roger Rochat, and Jules References Terry. 1981. "The Risk of Childbearing Re-evaluated." Ariencan Journal of Public Health 71(7):712-16. Capocaccia,Riccardo, andGinoFarchi. 1988."MortalitvfrotnLiverCmrrhosis Smith, Jack C., J. M. Hughes, P. S. Pekow, and R. W. Rochat. 1984. "An in Italv: Proportion Associated with Consumption of Alcohol." Journal of Assessment of the Incidence of Maternal Mortality in the United State,." Clinical Epideminlogy 41(4):347-57. American Journal ot Public Health 74(8):780-83. Doll, Richard, and Richard Peto. 1981. The Causes ofCanier. Oxford: Oxford Snvder,J. D., and Ni. H-. Merson. 1982. "The Magnitude ofihe Global Problem University Press. of Acute Diarrhoeal Disease: A Review of Active Surveillance Data." Fortney, J. A., Irene Susanti, Saad Gadalla, Saneya Saleh, Susan Rogers, and Bull'tin of the World Health Organization 60(4):605-13. Malcolm Potts. 1986. "Reproductive Mortality in Two Developing Couti- t SDHsS (United States Department of Health and Human Services). 1989. tries-." Anericat jourtal of Public Health 76(2):134--36. Reduciig the Health Consequences IfiSmokinug: 25 Years of Progress. A Report Hakulinen, Timo, Harold Hanslowka, Alan Lopez, and Tadashi L. Nakada. of the Surgeon General. Rockville, Md.: Centers for Disease Coitrosl, Office 1986."Global and Regional Mortality Patterns by CaUseof Death in 1980" on Smoking and Health. IntertationalJountal of Epidemiology 15(2):226-33. Walsh, J. A. 1988. Establishing Health Priorities in the Developing World. Boston: Hull, T. H., J. E. Rohde, and A. U). Lopez. 1981. "A Framework for Estimatring Adams Publishinig Group for the United Nations Development Programnme- Causes of Death in Indonesia. Miajalaj DemograJll Indonesia 15:77-125. World Bank. 1989- "China: Long-term Issues in Options for the Health IndlanOfficeoftheRegistrarGeneral. 1989.Surre). fCau4sesoJfDeath(Rural)- Sector." Annual Report, 1987. Series 3, 20. Delhi. W HO (World Health Organization). 1985. VWorldfHealth Statistics Annual 1955. KirkWood, Bettv R. 1991a. "Diarrhoea." In R. G. A. Feachem and D. T. Geneva. Jam ison, eds., Disease anid MIt talt itn Sub-Saharan .Africa. New Ycrk: Oxf ord - 1989. Expanded Psogramtnre (n Immunization: Update. Mav. Geneva. University Press._ - 1991a. "Smoking as a Cause ofCaancer." Tobacco Alert October:2, . 1991b. "AcuIte Respiratorv Infections." In R. G. A. Feachein and D. . 1991b. "New Estimates of Maternal Mortalitv." Weekly Epidemnuolog- T. Janlison, eds., Disease and Mortality ms Sub-Sahiatvai Africa New York: ical Record. 66:345-48. Oxford University Press. - 1992. Eighth Report onl the WSSorld Health Situation. Geneva. Leowsku, Jerzv. 1986. "Mortalhty from Acute Respiratory Infections in Chil- Ziskin, Leah Z., Margaret Gregorv, and Michael Kreitzer. 1979. "Improved dren iiiider 5 Yeiars of Age: Global E-stiniates." World I lealth Statistics Surveillance of Maternal Deaths." International Jounial of Guarlecolug-' and Quarterly 39:1 38-144. Obstetrics 16:282-86. 3 The Epidemiologic Transition and Health Priorities Jose Luis Bobadilla, Julio Frenk, Rafael Lozano, Tomas Frejka, and Claudio Stern The world approaches the end of a century and a millennium what is achievable by current knowledge and what is actually having achieved, during the last century, the most intensive achieved. To the extent that controlling certain diseases be- and extensive health changes in history. Advances in our comes technically feasible, their continuing existence among knowledge of the causes and effects of diseases, progress in deprived populations becomes morally compelling. The prob- sanitation and nutrition, development of vaccines and drugs, lem of equity has thus emerged as one of the central concerns expansion of a vast network of facilities and personnel, appear- of our times. ance of a whole sector of the economy devoted to health-re- As part of the increasing complexity of the health field, most lated goods and services, predominance of medical over govemments around the world have developed various forms religious or legal considerations in the interpretation of human of organized social response to the problems of their popula- experience: these are but a few of the shifts that have radically tions. Although the role of the state in providing health transformed the health scene in most countries. services has recently decreased in some countries, the overall The results are well known: more people enjoy the benefits trend after World War 11 has been oriented toward an incre- of good health than ever before. Yet, as with all true progress, mental participation. Modem nations, on the whole, now improvement in health has created new problems and consider health an essential element of development, requir- challenges. In a very real sense, it can be said that the health ing integration into national plans. field has been a victim of its own successes. Thus, the dramatic The simultaneous presence of more complex health condi- reduction in the incidenice and severity of infectious diseases tions, on the one hand, and increased commitment to extend- has allowed the survival of a large number of individuals who, ing health services, on the other, has given a strategic character tor that very reason, become increasingly exposed to the risk to the definition of priorities. This is compounded, in many of chro)nic ailments and injuries. Similarly, control of fertility countries, by a situation of stagnant and even shrinking re- is modifx'ing structure of the population, leading to an unprec- sources. In such a context, it has become imperative to develop edented proportion and quantity of elderly people, who pose methods for long-term health planning that are solidly enormous demands on the health care system. grounded on demographic and epidemiologic information. In addition, the general process of industrialization, urban- Needless to say, many economic and political variables outside ization, and modernization-of which the expansion in health the planning process affect the final allocation of resources. At services has been both a consequence and a facilitator-has the very least, health planning allows us to know the way in entailed high costs. The environment has been seriously and which those other variables make resource allocation deviate permanently damaged in many parts of the world. We have from the technically defined optimum. At its best, health leamed that mass consumption of tobacco is directly responsi- planning can constitute a force for mobilizing social resources ble for a large burden of premature deaths and disabilities. The toward the satisfaction of health needs. shift in diet from vegetables and cereals to animal foodstuffs Indeed, our ability to identify new health changes and and some artificial products has been recognized as contribut- opportunely respond to them depends on a clear understanding ing to cardiovascular disease and some cancers of the digestive of the determining factors and on the consequent design of system. innovative models for health promotion, and for disease pre- The net result of these changes has been that the health field vent ion and treatment. In this chapter we attempt to apply the has achieved a level of complexitv never seen before. Such theory of the epidemiologic transition to explain the recent complexity is compounded by the tact that the benefits of transtormations and the future evolution of health needs in progress have been veryi unequally distributed both between populations. We particularly refer to middle-income countries, and within nations. Together with the explosion in science where such transition adopts specific forms and where the gap and technology there has been a widening of the gap between between the changing character of health problems and the 51 52 Jose Luis Bobadilla, Julio Frenk, Rafael Lozano, Tomas Frejka, and Claudio Stern adequacy of the social response is of significant concern. one hand, there are the health conditions or needs of the popu- Among the middle-income countries, Mexico is used to illustr- lation; these are represented mostly by negative deviations ate the basic information and methods required to incorporate from physical and psychological function, although they also demographic and epidemiologic criteria into the process of include nonmorbid conditions like pregnancy and, in the most planning for health services and related programs. comprehensive perspective, positive well-being. On the other The opening section of the chapter presents the conceptual hand, there is the organized social response to the health needs basis for understanding the epidemiologic transition in mid- of the population; this entails any organized collective action dle-income countries and the main components of health aimed at promoting health, or at preventing and treating planning. In it we define demographic variables as important disease. In this context, the health transition refers to the measures of population health needs and therefore as funda- changes over time of both the health conditions and the mental elements for setting priorities. Next, we describe some organized social response. In turn, the health transition com- of the important social and demographic changes which have prises two sets of processes, corresponding to the foregoing occurred in Mexico during the past fifty years. An account of differentiation. The first one is what Omran (1971) described the main aspects of the epidemiologic transition in Mexico as the epidemiologic transition, that is, changes in health condi- follows, in which we propose that it might be a new model not tions. The second is the health care transition, that is, changes hitherto considered by the original theory. The following in the organized social response (Frenk and others 1989b). section presents projections of the most important demo- Empirical evidence shows that these two transitions are depen- graphic and epidemiologic variables from 1980to 2010 and the dent on each other. For example, universal coverage of effec- general trends of social variables in that time period. The last tive health services has contributed to the decline of childhood section includes an analysis of the likely implications that the mortality due to infectious diseases in countries like China, main described changes will have on the health system, and Costa Rica and Sri Lanka. What is less common is the antici- we suggest some reforms that should be introduced to deal with patory adaptation of the health system to ongoing changes in the complex health scenario of the next twenty years. the health needs of the population. In any case, we have a more elaborate theory of the epidemiologic than of the health care Conceptual Framework transition. The theory of the epidemiologic transition as developed to Although some authors may consider it possible to identify a date refers to the changes in the population parameters and in common pattern of epidemiologic transition for the industrial the health and disease patterns, which occur over decades or nations, for the so-called developing countries this is almost centuries. The elements of this theory are: (a) the description impossible without incurring significant mistakes or superficial of the strong links between demographic changes and health generalizations. Actually, not all poor countries are continu- needs of populations, as measured by the mix of causes of death ously in the process of developing; at times of crisis they seem and the age structure of mortality; (b) the coherent classifica- to be static or even underdeveloping. In many respects, the tion of eras that allows for the identification of important differences among poor countries are often greater than those socioeconomic changes that affect survival at different times; between industrial and developing nations. The picture is even and (c) the potential for anticipating changes in disease pat- more complex in the areas of health and health care, since terns and the opportunities offered by such anticipation to many relatively poor countries resemble industrial nations in strategic health planning. these respects (as is true of Costa Rica, Cuba, and Sri Lanka). In his original formulation, Omran (1971) proposed three However, most countries rich and poor present serious inequal- eras of the epidemiologic transition that countries experience ities among different social groups, leading to internal hetero- at different stages of their social and economic development: geneity. Many rich nations are characterized by development The era of pestilence and famine, when life expectancy is low with pockets of underdevelopment; conversely, most of the (20 to 39 years) and the major causes of death are associated developing nations are composed in varying proportions of with malnutrition, infection, and complications of reproduction. underdevelopment with pockets of development. The era of receding pandemics, when the disease pattern is Attempting to describe the health transition for the devel- still dominated by infectious diseases and malnutrition, but oping countries as a whole is thus inappropriate. In the absence major mortality fluctuations, including peaks, are less com- of an optimal classification of countries, we will adopt the one mon. Life expectancy rises to between 30 and 50 years, and used by the World Bank, based on national income per capita. there is a tendency for increasing control of the biological In particular we will refer to "middle-income" countries, which pollution of the environment, as a result of improved sanita- in 1990 had a median gross national product per capita be- tion, with declining rates of infection. tween us$611 and us$7,619 (World Bank 1992). The era of degenerative and man-made diseases, characterized by the rise of cardiovascular diseases, cancer, diabetes, and The Health Transition other degenerative diseases. Life expectancy is over 50 years and fertility becomes a crucial factor of population growth. Our analysis of the dynamics of health in human populations The original description of the theory and some of the later begins with the differentiation of two basic elements. On the updates acknowledge the heterogeneity of social and eco- The Epidemiologic Transition and Health Priorities 53 nomic development among countries. Accordingly, Omran actually the health conditions for a large proportion of adults (1971) suggested that at least three models can be recognized: might not improve or might even worsen (Wildavsky 1977). the classical or Westem model, the accelerated model (such as It could be that the third era represents an advanced stage of that followed by Japan), and the delayed or contemporary a longer transitional era, so that in a fourth era it might be model, the main differences among them being the timing and possible to live in societies with low mortality and morbidity the pace of change. The delayed (or contemporary) model rates of both infectious and noncommunicable diseases and of described the incomplete transition of most developing coun- accidents, violence, and mental diseases. tries. The important decline of mortality started after World The second assumption refers to the sequence of the transi- War 11 and was mainly a result of the adoption of imported tion from one era to the other. It has been proposed that some public health measures and some medical interventions, and modifications to the original theory might be introduced to not so much to improvements in economic and social factors, accommodate experiences from some middle-income coun- as was the case in the classical or Western model. Although tries (Frenk and others 1989b). First, the eras are not necessar- the gains in child survival have been substantial in this model, ily sequential, since two or more may overlap at the same time. the mortality rates are still relatively high. Second, the evolutionary changes in the pattern of morbidity More recently, Frenk and others (1989b) have proposed a and mortality are reversible, so that there can be backward new model, called the "protracted-polarized model" of the movements, as we pointed out earlier. epidemiologic transition. Its formulation is largely based on In addition, the theory presents some limitations that need observations from some large middle-income countries. Its to be highlighted: main features are as follows: * It has limited potential to explain heterogeneity in dis- * The decline of mortality takes place in very short periods ease patterns within countries. As we postulated previously, of time, as compared with the classical model. Western within a given country, it is possible to recognize social European countries that followed the classical model took groups with common infectious diseases, coexisting with more than 130 years to reduce mortality from 35 deaths per social groups affected by noncommunicable diseases. 1000 population to 10, whereas many middle-income coun- * The implicit and operational concept of health is exclu- tries have achieved a similar decline in less than 70 years. sively limited to health losses that can be documented * Theonsetofthemortalitydeclinestartsinthetwentieth through mortality. Several health problems that predomi- century, reaching low levels near the end of the century. nantly produce morbidity, let alone positive aspects of * Despite significant reductions in mortality by infectious health, are by definition ignored. This is extremely import- diseases, these diseases are not brought fully under control, ant, since the advance in therapeutics often postpones or and their incidence rates remain relatively high by the end averts death but does not cure the disease. of the century. This situation, together with the increase of * The capacity to relate social and economic changes to noncommunicable diseases, produces an overlap of eras. health improvements is relevant only to the general aspects * The unequal distribution of wealth and the incomplete of development. Unfortunately, the theory as it stands can- coverage of interventions gives place to a widening of the not explain how social and economic changes are related to gap in health status among social classes and geographical health transformations; it also has a limited value to explain regions. This process has been described as "epidemiologic the role of different forms of social organization on the polarization" (Frenk and others 1989a). Even though this timing, pace, and modality of the epidemiologic transition. process might have occurred under other transition models, in the protracted model polarization possibly goes on for Health Planning: Using the Transition to Set Priorities longer periods. * A review of morbidity data reveals the reemergence of As experience in the developing countries accumulates and as epidemic diseases that had been controlled or eradicated, new analyses are elaborated, our knowledge will expand and which produces a countertransition. the progression of epidemiologic transition theory may help us to understand current health dynamics in a variety of coun- The first two characteristics of this model serve to differen- tries. Such understanding is essential not only to advance our tiate it from others previously described. The other features theories about the nature of change in the health field but also have not been examined for countries presenting the classical to shape future reforms. or the delayed model. It is proposed that thev also might differ With respect to budgetary allocations, there are two main between the protracted-polarized and the other models of methods of planning for health services. The first one, which epidemiologic transition. is the most commonly used, can be called the retrospective There are two implicit assumptions in transition theory that method, because it commonly starts out from a fixed budget need to be critically examined. One is that each era is more and then moves to the allocation of those resources to different desirable than the previous ones because it reflects "progress." health programs, according to previously established priorities. Yet, even when survivorship increases, morbidity and disabil- The alternative method, which can be called prospective, ity do not necessarily decline in the same proportion; and begins with a definition of population health needs. In this 54 Jose Luis BobadilUa, Julio Frenk, Rafael Lozano, Tomas Frejka, and Claudio Stem context, the temi "needs" refers specifically to health condi- (table 3-1). This process has been causally associated with the tions that require care but not to the care itself (Donabedian demographic and epidemiologic transitions. These are dealt 1973). That is tosay, needscan bedefined as health and disease with below in terms of Mexico's social, demographic, and processes, such as death, disease, and disability, as well as health care variables. nonpathological conditions that require care, such as preg- nancy or the monitoring of childhood growth and develop- Socioeconomic Factors ment. The two most important demographic determinants of health needs in a population are the absolute number of From a predominantly rural and agrarian country, Mexico has individuals and the age composition (Jones 1975). Other im- moved a long way toward becoming an urban and industrial portant determinants include changes in the prevalence rates nation. Although economic growth in Mexico during this of disease, injury, and disability, as well as changes in the period was substantial until the early 1980s (when the finan- demand for health services, which are influenced by rising cial crisis and economic stagnation began), various factors expectations and disposable income. have contributed to maintain a highly unequal social and From these needs, the planning process moves on to deter- economic structure. mine the health services that would be required to meet them. Among the factors which have contributed to this high Finally, service production targets are used to estimate the degree of social and economic inequality among the Mexican required resources, including human and material, as well as population, the following are important: mountainous land, their financial expression in a budget. The translation of populated since ancestral times by a large number of different health needs into their service and resource equivalents is ethnic groups; three centuries of colonial rule during which a achieved through a series of mediating factors. Thus, the racially based unequal social structure was institutionalized; a satisfaction of health needs by the use of services is mediated model of capital-intensive industrial development during the by the quality, technological content, and equity in the distri- last fifty years (which is not easily compatible with greater bution of those services. In turn, the amount of services that social equality, since it benefits the urban upper- and middle- are actually produced and used is determined by the availabil- class groups to a much greater extent than the rural population ity, accessibility, and productivity of resources. The planning and the urban working classes); a very fast rate of population process makes use of both normatively and empirically derived growth since the 1940s, demanding large resources for basic standards that provide the quantitative elements required to social and economic infrastructure (also very unequally distrib- translate needs into services and into resources (Donabedian uted, owing to the other reasons given); political centralization 1973; Frenk and others 1988). The retrospective approach to and uninterrupted management for the last seventy years by health planning has the undeniable advantage of establishing, the same political party. from the outset, the budgetary limitations that the planner As a result of some of these factors, territorial, economic, unavoidably has to face. Its major disadvantage, especially for and social "development" has taken place in a highly concen- long-term planning, is that it tends to perpetuate existing trated manner. Industrialization and urbanization are highly conditions. Therefore, if the purpose of planning is to antici- concentrated and coexist with rural dispersion and extended pate the future in order to transfomi it, it must adopt a prospec- traditional agrarian forms of production and subsistence. Dis- tive view that begins by estimating future health needs and tribution of income is among the most unequal in the world. derives from them the resource requirements. In this respect, Differences in living standards are vast. A few have much more the theory of the epidemiologic transition provides an invalu- than they need (the top 10 percent of the population concen- able framework for projecting likely scenarios of health needs. trate about 40 percent of national income), whereas a large Such an anticipatory exercise is the only rational way to proportion of the population subsists at substandard levels in determine health priorities for the long run. every respect, including, of course, health (L6pez-Gallardo For these reasons, the present chapter adopts a prospective 1984). approach in illustrating the nature of the health transition in Thus, Mexico has undoubtedly experienced a process of Mexico. We use the most straightforward measures of health economic growth, and important changes in its economic and needs, namely, population size and mortality. All our calcula- social structure have been taking place. The labor force work- tions and projections to the year 2010 are based on age- and ing in agriculture decreased from 65 to 23 percent between cause-specific mortality rates, in combination with projections 1940 and 1990 (see table 3- 1); the urban population increased of population age groups. From this point of departure, we from 20 to 57 percent between 1940 and 1990; the literate analyze the changes that must be introduced if the health care population increased from 43 to 88 percent between 1940 and system is to respond in an effective manner to an increasingly 1990; houses with running water increased from 17 percent in complex set of population needs. 1950 to 79 percent in 1990; and so forth. But great inequalities not only persist; they have become aggravated as a result of the Mexico: Historic Trends financial and economic crisis prevailing in the 1980s. Income and wealth have become even more unequally concentrated; Mexico has experienced a significant process of social and unemployment and underemployment are on the rise, and economic change during the last fifty years, yielding consider- there are fewer public resources to mitigate poverty and extend able improvements in the living conditions of the population the economic infrastructure and social services to the whole The Epidemiologic Transiton and Health Priorities 55 Table 3-1. Indicators of Social and Economic Developnent in Mexico: 1940-90 Indicators 1940 1950 1960 1970 1980 1990 Labor force in agriculture (percent of economically active population) 65.4 50.2 49.4a 39.2 25.8 22.6 Literate (percent of population aged 15 and over) 43.2 55.9 65.5 74.1 82.7 87.6 Percent ofurbanpopulation 20.0 28.0 36.5 44.9 51.8 57.4 Percent of rural populationc 70.0 56.7 48.3 40.4 33.7 28.7 Percent of houses with running water - 17.0 23.4 61.4 70.2 79.4 Percent of houses with sewer - - 29.7 41.0 51.2 63.6 Percent of houses with one room - 60.4 57.8 39.8 29.8 Occupants per house (average number of persons) d - 4.9 5.5 5.8 5.5 5.0 Gross domestic product per capita (in 1980 U.S. dollars) - 1,408 1,547 2,180 3,096 2,708 -Data not available. a. Estimated from A[timir. 1974. b. Localities with more than I 5,000 inhabitants. c. Localities with less than 2,500 inhabitants. d. Data from G6rnez-de-Leon and Frenk. 1992. Source: Generally, 1940-80: United Nations, 1989a; 1990: Mexico, Instituto Nacional de Estadistica, Geograffa. e Informntica (INEGI), 1992. Exceptions in- dicated above. population (Consejo Consultivo del Programa Nacional de main causes of death, and the infant mortality rate in 1991 is Solidaridad, 1990). estimated to have declined to less than a third of its level in 1950. The increase of deaths due to noncommunicable dis- Historic Trends in Population and Health eases was detected before 1950, but it was not until the late 1960s that substantial incrementsbegan (table 3-2). The sharp During most of the twentieth century Mexico was a typical increase of noncommunicable diseases between 1970 and 1980 example of a country with high levels of fertility and declining is overestimated. A large proportion can be explained by the mortality, a situation that resulted in one of the highest rates improvement in the certification of cause of death during the of population growth in the world. In the decades of the 1950s, decade of the 1970s, and the expansion of health facilities with 1960s, and 1970s, the annual population growth rate for Mex- more accurate diagnostic technologies. icowasover3 percent. Figure 3-1 depicts the trendsofthe birth In the decades of reference, 1950 to 1980, estimates of the and mortality rates during the twentieth century. population size and of the numbers of births were quite suffi- Until 1970 the rate of population growth was one of the cient to anticipate the load of morbidity, injury, and disability most important determinants of health needs. Two out of the to be carried by the health services in any subsequent ten or otherthreeprimarydemographicindicatorstelevantforhealth fifteen years. There was less need for strategic planning, for planning remained fairly constant: the age structure of the several reasons, than at present. The rate of economic growth population during the period 1950 to 1980 showed a pyramidal form, with 43 to 47 percent of the inhabitants under fifteen years of age, and the absolute number of deaths was similar in Figure 3-1. Death and Birth Rates in Mexico, 1950 to that of 1980, about 450,000.The number of births was 1900-2010 the exception: it increased by 74 percent from 1950 to 1970 (table 3-2). During these three decades the rate of economic growth was Rate (per 1,000 inhabitants) higher than that of the population (Wilkie 1978). Similarly, 50 health services expanded both in scope and coverage, at a rate irth rate faster than the population. 40 The intervention of the Mexican state in the provision of 30 Mortality rate health services was institutionalized and strengthened as early as the 1940s. A social security system and the Ministry of 20 Health were created (Frenk, Hernandez-Llamas, and Alvarez- 10 Klein 1980). Health was seen as an essential input to economic development. The social security system, which has become 0 l l l l l l the dominant scheme for the provision of medical care, is 1900 1920 1940 1960 1980 2000 limited to those who are formally employed, especially those Year in the industrial and service sectors. In 1990 it covered about 55 percent of the population of Mexico. During the postwar period in Mexico, health indicators Note: Actual data 1900.80; estimates 1990-201 0. SotIrce: 1900-1970: Aiba 1971; 19801: Mico, JNEGI 1984; 1990-2010; M8ico, tended to improve. Infectious diseases were displaced as the INEGI Y CONAPO 1985; 1990-201 0: Instituto Nacional de Estadistica 1986. 56 Jose Luis Bobadilla, Julio Frenk, Rafael Lozano, Tornas Frejka, and Claudo Stern was faster than that of the population, so that maintaining at Figure 3-2. Age-Specific Fertility Rates in Mexico least a constant level of coverage and quality was taken for during Period of Fertility Decline granted. Table 3-2 presents in the bottom row the growth rate in gross domestic product for three decades, showing a steep Age-specific fertility rates (per 1,000) increase from the 1950s to the 1970s. In sharp contrast, the 350 trend is reversed during the 1980s, when the same indicator I - 'N".1970-72 yields an average of 1 per cent for the decade 1981-90. Al- 300 1977-79 though economic growth rates have recovered somewhat over 2r- -2 the first two 'ears of the 1990s, most projections anticipate 200 -\ " lower growth rates for the period 1993-96, than those reached 150 :;' 1984-86 '. in the 1970s. 100 A significant demographic change of the 1970s was the 50 - onset of a rapid fertility decline. The total fertility rate declined 0 l l l from 6.7 (children per fertile age woman) in 1970 to 5.7 in 10 20 30 40 50 60 1975-76 and to 4.3 by 1981. For 1990 the total fertility rate was estimated at 3.3 children for every woman of child-bearing Age (years) age. This fertility decline is expected to continue, although at a slower pace, for the foreseeable future. Figure 3-2 shows the Source: Aparicio-Jim6nez 1988. age-specific fertility rates (per 1,000 fertile women) in three different periods between 1970 and 1986, when significant epidemiologic transition under the protracted-polarized model decline in fertility occurred. (Frenk and others 1989b) described earlier, characterized by infectious diseases and malnutrition coexistent with noncom- The Epidemiologic Transition in Mexico municable diseases and injury. Mortality in Mexico started to decline early in the twentieth The changes in the pattems of morbidity and mortality, which century. At the beginning of the century, life expectancy at are closely related to the demographic transition, constitute birth was estimated to be about twenty-five years. By 1950, life the main elements of the epidemiologic transition (Omran expectancy had almost doubled. For 1950, estimates of life 1971). It is suggested that Mexico might be undergoing an expectancy range from 46.2 to 49.1 years for men and 49.0 to Table 3-2. Selected Demographic, Epidemiologic, and Economic Measures for Mexico, 1950-2010 Measures 1950 1960 1970 1980 1990 2000 2010 Demographic Population (thousands) 27,376 37,073 51,176 69,655 81,250 103,996 123,158 Annual population growth rate (over previous ten years) - 3.0 3.2 3.1 2.6 1.9 1.7 Age groups (percentage) 0-14 43 46 47 44 38 31 29 15-64 54 51 50 53 57 64 65 65 and over 3 3 3 3 4 5 6 Deaths (thousands) 443 419 499 462 423 535 665 Births (thousands) 1.278 1.663 2.224 2.100 2.352 2.520 2.745 Epidemiologic: cause of death Infectious and parasitic diseases thousands 214 158 196 83 66 50 41 percentage 48.4 37.6 39.2 17.6 15.6 9.3 6.1 Cardiovascular diseases, cancer, diabetes, other selected chronic diseases, and violent deaths thousands 66 77 111 217 229 388 526 percentage 15.0 18.3 22.3 46.9 54.3 72.6 79.4 Economic GDP annual growth rate per capita (over previous ten years) - 3.0 3.7 5.5 1.0 - - - Data not available. Note: Projections for years 2000 and 2010. Source; Deniographic measures 1950-80: United Nations 1986; 1990: Mexico, Instituto Nacional de Estadistica, Geografia e Informatica, 1992. Demo- graphic projecrions: Mexico, Instituto Nacional de Estadistica, Geografia, e Informatlca y Consejo Nacional de Poblaci6n 1985. Epidemiologic measures, 1950-90: Mexico, Direcci6n General de Estadistica, from vital statistics of the Civil Registrar for 1950, 1960, 1980, and 1990 (unpublished); 2000-2010: Ex- pert opinion on linear trends 1950-80; 1960-80: Economic Commission for Latin America and the Caribbean 1986; 1990: World Bank 1992. The Epidemiologic Transition and Health Pnorities 57 52.1 for women (Camposortega 1988). The mortality decline Figure 3-4. Deaths by Cause, According to Life has continued since then, and by 1990, life expectancy was Expectancy at Birth estimated at about seventy years (World Bank 1992). According to the original theory proposed by Omran Percentage (1971), and judged solely by the time of onset and the rapid 100 speed of the mortalitv decline, Mexico would fit in the accel- erated model. The common denominator of the delayed model 80 and its transitional variant (Omran 1983) is the starting point 60 of the pronounced decline in death rates in the decade of the 1950s. Clearly the decline of mortality in Mexico started 40 earlier, in the 1920s. For this reason Mexico does not fit the delayed model nor its transitional variant. 20 One of the universal characteristics of the epidemiologic transition is the shift of the age structure of mortality from the 0 l l l l 65 young ages to the old ages. Figure 3-3 presents the age distri- bution of deaths in Mexico for the period 1940-85. It is Life expectancy at birth interesting that the greatest changes in the age structure of mortality occur between 1970 and 1985. Similar to what Infectious I Chronic Injuries Other Omran (1983) described for the transitional variant of the diseases diseases delayed model, the reductions in infant and child mortality (birth through four years) in Mexico are substantial, but the Source: Mexico; INEGI 1986. levels slacken at relatively high rates. For the period 1981-86 the infant mortality rate was 43 deaths per 1,000 live births trend is as would be expected: as life expectancy rises, the (Bobadilla and Langer 1990), representing about 20 percent of proportion of infectious diseases declines and the proportion the total deaths (figure 3-3). It is likely that the low percentage of noncommunicable diseases rises. Figure 3-4 depicts, at dif- of infant deaths in 1940 is affected by heavy underregistration. ferent levels of life expectancy, the relative distribution of The percentage of total deaths which correspond to the elderly deaths by cause of death. It is interesting that, due to the young (over sixty years) increases from 18 percent in 1940 to almost age structure of the Mexican population, at the highest level 40 percent in 1985. The latest information on mortality shows of life expectancy shown (sixty-eight years) the proportion of that 50 percent of all registereddeaths in 1990 occurred among infectious disease is still high (representing about 35 percent the elderly. of the deaths). As life expectancy at birth increases, the structure of mor- Improvements in living conditions reduced the incidence tality by causes of death changes (Omran 1971). The general rates of diarrheal and other infectious diseases. In addition, the antimalaria and the mass vaccination campaigns contributed significantly to the decline of infections. The introduction of Figure 3-3. Deaths in Mexico by Age, 1940-85 more effective therapy reduced the fatality rates for many infectious diseases and also contributed to the decline in the death rate. Percentage As pointed out earlier, the trends of morbidity are not 100 considered in the original description of the epidemiologic 80 transition. The study of morbidity is particularly important in countries where the greater survival rates have been obtained 60 through reductions of lethality, sometimes in the absence of 40 reductions of incidence rates. In Mexico many infectious dis- 20 eases that have very low rates of mortality still present high o rates of incidence or prevalence. Two groups can be distin- guished according to their trends: first, the diseases prevent- 1940 1950 1960 1970 1980 1985 able through vaccination-such as measles, poliomyelitis, and diphtheria-that are clearly declining; second, diseases that Year have been previously controlled, such as malaria, dengue fever, _ Under 1 year = 1-4 years 5-14 years and cholera, but have recently shown a reemergence (Sober6n and others 1988). Figure 3-5 shows the trends of the absolute number of newly diagnosed malaria cases in Mexico for the period 1942-87. In the early 1940s there were more than 140,000 cases per year. This figure declined to less thus 5,000 Source: MWxico:Secretarlade Salud 1987. in the early 1960s, but then started to increase, so that by 1985 58 Jose Luis Bobadilla, Julio Frenk, Rafael Lozano, Tomas Frejka, and Claudio Stern Figure 3-5. Incidence of Malaria in Mexico, states rises from 1.6 to 3.3, as compared with the wealthier 1942-88 states. The different patterns of epidemiologic change within the Thousands country provide further evidence of the epidemiologic polar- 160 ization in Mexico. In table 3-4 we compare the cause-specific death rates for diarrheal diseases and acute respiratory infec- 140 \ t\ tions with those for cardiovascular diseases and cancer. The 120 death rate for diarrheal diseases and acute respiratory infec- 100 - tions was 1.6 times higher in the southem region in 1940, whereas in 1985 it was almost four times higher in the southern 80 - region than in the northem. Both regions show an impressive 60 - absolute decline, but the greatest occurred in the northern 40 \/n I \ / region, where the corresponding death rate was reduced by 94 percent from 554 (deaths per 100,000 population) in 1940 to 20 \ 31 in 1985. In the southern region the rate was reduced by 86 0 1 1 - percent during the same period, from 878 to 122. 1942 1947 1952 1957 1962 1967 1972 1977 1982 1987 The trend of the death rates due to the selected chronic Year diseases shows very little change for the northern region, with an increase of 25 percent, whereas in the southem region the Source: Mexico: Direcci6n Generalde Epidemiologia 1989. increase was about 85 percent. This discrepancy in the rate of increase of chronic diseases leads to a convergence of the corresponding rates in both regions (table 3-4). It can be there were more than 130,000 cases. This is a typical example concluded from table 3-4 that the inhabitants of the northern of the process called countertransition mentioned earlier. region have been able to control infectious disease and proba- The overlap of eras is closely related to the pace of the bly have entered into the third era of the epidemiologic tran- epidemiologic transition among different social groups. The sition. They have not, however, had the expected increase of universal social inequalities in health have been usually de- cardiovascular diseases and cancer. The death rate for cardio- scribed in quantitative terms, because the burden of disease vascular diseases in the United States in 1986 was 3.5 times as and death is greater among the poor. The mortality decline high as that in the northern region of Mexico. starts among the higher social classes, and eventually the lower The final outcome of this process of regional polarization is classes catch up and close the gap. This is described as a general summarized in the last three rows of table 3-4. The ratio of the pattem of transition (Omran 1983) and is probably applicable selected infectious to chronic diseases changes from 6.1 to 0.3 to all countries regardless of the model of epidemiologic tran- in the northem region, with the crossover occurring in the sition they pass through. Under the protracted model, how- 1970s. In contrast, in the southem region the change is from ever, the lower social classes show very small improvements, 19.8 to 1.5, with no crossover by 1985. The difference in the whichraisesquestions about the time required tocatch up with ratios between the two regions increased from 3.2 to 5.0 in the the upper classes. Thus the period of polarization seen in other period 1940-85. models probably takes longer in the protracted model. Table 3-3 shows the gap between the infant mortality rate in the Future Demographic, Epidemiologic, and Social Changes poorest states of Mexico, those of the southern region, and that of the wealthier states in the northern region of the Mexican In the next twenty years, many middle-income countries will territory. It is shown that the excess of mortality in the poorer continue to show substantial improvements in child health Table 3-3. Infant Mortality Rate (IMR) in Southern and Northern Regions of Mexico, 1962-86 Southern regiona Northern regionb Period Infant deaths IMR Infant deaths IMR Southern/northern 1962-66 39 147 26 92 1.60 1967-71 46 93 55 73 1.28 1972-76 73 112 74 69 1.62 1977-81 71 92 51 45 2.02 1982-86 66 92 32 28 3.26 a. Includes Tabasco, Yucatan, Campeche, Quintana Roo, Qaxaca, Chiapas, Puebla, and Tlaxcala. h. Includes Baja Califoria, Baja California Sur, Sonora, Sinaloa, Nayarit, Nuevo Le6n, and Tamaulipas. C. IMR (Infant mortality rate) refers to deaths of children under one year of age per 1,000 live births. Source: Mexico, Direcci6n General de Planificacion Familiar 1989. The Epidemiologic Transition and Health Pnoities 59 Table 3-4. Cause-Specific Death Rates for Selected Infectious and Chronic Diseases in Southern and Northemn Regions of Mexico, 1940-85 (per 100,000 inhabitants) Cause of death Regiond 1940 1950 1960 1970 1980 1985 Diarrhea and acute respiratory infections Southern 878 507 403 403 178 122 Northem 554 323 227 196 66 31 Southern/northem 1.58 1.57 1.77 2.06 2.70 3.94 Cardiovascular diseases and cancer Southem 44 57 56 54 64 81 Northern 91 121 110 110 112 114 Southern/northem 0.48 0.47 0.51 0.49 0.57 0.71 Ratio (infectious/chronic) Southem 19.8 8.8 7.2 7.5 2.8 1.5 Northem 6.1 2.7 2.0 1.8 0.6 0.3 Southern/northem 3.2 3.3 3.6 4.2 4.7 5.0 a. Southem region includes Tabasco, Yucatan, Campeche, Quintana Roo, Oaxaca, Chiapas, Puebla, and Tlaxcala. Northern regioni includes Baja Califor- nia, Baja California Sur, Sonora, Sinaloa, Nayarit, Ntievo Le6n, and Tamaulipas. Source: 1940: Mexico, Secretarfa de Salubridad y Asistencia 1946; 1950. 1960: Mexico, Instituto Nacional de Estad(stica 1980; 1970, 1980, 1985: Mxico, Direcci6n General de Estadfstica 1973, 1984, 1988. and life expectancy. Most of them will also witness, paradoxi- Despite the continuous decline in the total fertility rate, the cally, a greater burden of disease. In this section we examine total number of births will not decrease. The slight downward the projected trends of the main determining factor of health trend of the 1970s will be reversed and, according to this needs and demand for health care in Mexico. projection, an increase of 31 percent is expected between 1980 and 2010 (table 3-2). Population Projections and Future Health Needs Projections of the Mortality Structure by Cause of Death In the early 1980s the Mexican government (Mexico, INEGI/Ci NAPO, 1985) anticipated that fertility would continue In order to estimate the effect of the aging of the population to decline rather rapidly, at least through the 1980s, and on the relative contribution of the different causes of death, probably also into the 1990s. On the basis of this assumption, the 1980 age-specific death rates for eleven groups of causes of a "programmatic" projection aiming for replacement-level fer- death were applied to all the age groups projected for 2010, tility by 2010 was calculated. The actual fertility reduction of according to the alternative projection. The results are pre- the 1980s, however, was somewhat slower, corresponding sented in the second column of table 3-5, which shows the more closely to the "alternative" projection of remaining at 28 distribution (per thousand) of the selected causes of death. percent above replacement fertility in 2010, the projection Deaths due to malignant tumors would rise from 67 to 81 per used in this chapter. According to this projection, the annual thousand. An increase can be seen in the proportion of deaths absolute additions of population will increase, although the due to heart and circulatory diseases, which would rise from population growth rate will decrease from 3.1 percent in the 169 to 204 per thousand, a 21 percent increase. The total 1970s to 1.7 percent in the period 2000-10. number of deaths due to these causes would increase from The transition from high to low fertility is leading to signif- 74,000 in 1980 to 136,000 in 2010, an 84 percent increase. It icant changes in the age structure. The percentage of children is interesting to note that even though the proportion of deaths under fifteen will decline from 44 in 1980 to 29 in 2010, caused by diarrheal diseases and acute respiratory infections whereas the absolute number will more than only increase would decline from 169 per thousand in 1980 to 130 in 2010, from 30 million to 35 million. The group of adults age fifteen the absolute number of deaths for these two groups would through sixty-four will increase from 53 to 65 percent of the increase from about 78,000 to 86,000. Planning for health population, whereas the absolute number will rise from 37 services in the next thirty years cannot be done without million to 80 million, more than a twofold increase. The reliable estimates of future trends in incidence rates and pos- percentage of the population age sixty-five and above will sible changes in fatality rates of the main diseases and injury. double, from 3 to 6 percent, between 1980 and 2010, and the Projections of the future incidence rates of diseases, accidents, absolute number will more than triple from 2.3 million in 1980 and violence will probably follow trends ofthe past twenty-five to 7.1 million in 2010 (table 3-2). years or so. Infectious diseases will continue to decline to a The aging process will entail an increase in the number of certain floor, and noncommunicable diseases will continue to deaths, larger than any other increase in this century. By the increase, although the pace might be less pronounced. Satura- year 2010 the total number of deaths will be 665,000, about 44 tion or competition among causes will modify the increase. percent more than in 1980. The same can be said for deaths due to accidents and violence. 60 Jose Luis Bobadilla, Julio Frenk, Rafael Lozano, Torras Frejka, and Claudio Stem Table 3-5. Deaths in Mexico by Main Causes, 1980 and Projections for 2010 (per thousand) 2010 Using 1980 Projecting linear Linear trends age-specific trend from 1965, corrected by Cause of death iCD-9 codes 1980 mortality rates 1970, 1981 experts' opinions Malignant tumors 140-165, 170-175, 179-208, 230-234 67 81 89 78 Accidents and violence E800-E848, E850-E869, E880-E888,E890-E999 161 156 287 272 Heart and circulatory diseases 390-398, 401-405, 410-438, 440-459 169 204 409 332 Chronic bronchitis and other chronic respiratory diseases 466-490, 493 32 32 6 21 Acute respiratory infections 460-465, 470-478, 480-487 82 66 0 33 Diarrheal diseases 001-009, 120-129 87 64 5 24 Cirrhosis and other liver diseases 571 35 41 52 44 Other child infections 032-037, 045-055, 138 6 5 0 1 Diabetes 250 37 48 98 80 Perinatal problems 760-779 56 32 20 16 All other causes 268 271 34 99 Total 1,000 1,000 1,000 1,000 Source: 1980: Mexico, Secretaria de Programac[6n y Presupuesto v Secretaria de Salubridad y Asistencia 1984; 2010: Authors' calculations. Two different procedures were applied in estimating the causes of death. In other words, the first three causes of death relative contribution of the eleven groups of causes of death will be responsible for 44 to 79 percent of all deaths in the year forthe year 2010. First, the distribution of deaths was estimated 2010. In any case, the point is made that the Mexican popu- by applying the age-specific incidence rates that would result lation in the twenty-first century will be dying from causes, if past trends were to continue linearly. Three points in time most of which are preventable, that affect mainly adults and were used to estimate the slope and intercept of the linear trend the elderly. The complexity and the costs of preventing, con- (age- and cause-specific death rates for the years 1965, 1970, trolling, and treating many of the main diseases of the period and 1981). The results, shown in the third column of table 3-5, 2000 to 2010 will be considerably larger than those prevailing are not plausible, as can be concluded, for example, from the at present. The implications of this situation for the health virtual disappearance of deaths due to diarrheal diseases and system will be discussed later. acute respiratory infections. For this reason, we followed a second procedure. The opinions of seven epidemiologists on Future Social Change and the Demand for Health Care the future trends (previously described) of each cause of death for all age groups were used to estimate more plausible figures. Up to this point plausible demographic and epidemiologic The aggregated results are presented in the fourth column of changes have been assumed in the estimates of health needs table 3-5. The effect of having introduced experts' opinions is in the year 2010. Other economic and social changes, how- the moderation of the linear trends and the modification of ever, are taking place in Mexican society that will affect future straight lines into curves. The contribution of deaths due to health needs and that should be taken into account in health diarrhea and acute respiratory infections is quite plausible. The planning. In this section we briefly mention some of the trends same can be said for the corresponding data on accidents and in the areas of urbanization, education, employment, and violence. Deaths due to heart and cardiovascular diseases more social inequality and will speculate about their possible effects than double their contribution, rising from 169 to 352 per on future health needs and demands. 1,000 deaths. This is most likely an overestimation. Even though the methodology to estimate the future num- URBANIZATION. According to existing projections (Nufiez ber of deaths by specific causes could be more sophisticated and and Moreno 1986), by the year 2000 approximately 76 percent yield more plausible results, the three alternative projections of the Mexican population will live in communities with 2,500 shown in table 3-5 provide a guide for strategic planning. The or more inhabitants. A significant part of urban growth will calculations obtained using static 1980 rates give an idea of the take place in middle-size cities (50,000 to 500,000 inhabi- minimum number of expected deaths in the three main causes tants), although the large metropolitan areas will also continue of death (malignant tumors, heart and circulatory diseases, to grow significantly. At the same time, tens of thousands of accidents and violence). By contrast, the results obtained by rural localities will persist. A continued process of urbanization applying the regression coefficients of the linear trend provide in a context of greater poverty will have repercussions on the estimates of the maximum expected contribution for the same incidence of diseases and disabilities related to these condi- The Epidemiologic Transiton and Health Pnonties 61 tions, most notably injuries, mental disorders, alcoholism, and to the mid-1970s, will probably continue to diminish, as it did probably drug abuse. in the 1980s. As a result, it is possible to predict a greater polarization of Mexican society. The size of the middle class EMPLOYMENT. According to estimates and projections made will be reduced, and the lower classes might again encompass by Trejo (1988), it will be very difficult to create formal 60 percent or more of the total population. employment in the quantity and at the pace that will be required during the next years. As was shown before, the Implications for the Health Care System working-age population (fifteen through sixty-four years) will grow significantly faster than the total population during the From the foregoing analysis it is possible to conclude that early rest of the century. The annual growth rate of this population in the twenty-first century health needs will be governed in has been estimated at 3.5 percent in 1985-90, 2.7 percent in Mexico by an increase in the number of people who suffer 1990-95, and 2.3 percent in 1995-2000. An average of one from noncommunicable diseases, only moderate reductions million jobs would have to be created yearly during the next in the absolute number of infectious diseases, and an increase twelve years in order to satisfy the requirements of the new in the number of births. Increasing amount and complexity in population entering the labor market. The rates of economic the health needs will have profound implications for the growth required to yield so many jobs, about 6 percent annu- organization and delivery of health services. ally, seem to be difficult to attain in this period. Unemploy- ment and informal employment are likely to increase, leading Allocation of Resources for Competing Health Needs to a rapid growth of population not entitled to social security. In addition, women will continue to enter the labor market at So far, estimates of health needs have been derived only from a faster rate than the general population, imposing additional basic information on mortality. The relation between mortal- pressure on the demand for jobs. ity and causes of death, on the one hand, and morbidity and The incorporation of large numbers of people into the causes for medical care demand, on the other, is far from being informal labor market will generate a much greater demand on straightforward. This is particularly true for countries that are the health subsystem for the noninsured population (that is, in the midst of the epidemiologic transition. The discrepancy services provided by the Ministry of Health, other public between mortality and morbidity indicators is particularly assistance institutions, and to some extent private institu- relevant for the estimation of health needs that are met tions). This, in turn, could lead to more extreme health ineq- through hospitalization. Often the causes for hospital admis- uities between the poor and those with formal jobs, who are sion are not represented in the main causes of death. About 65 entitled to social security. The increasing participation of percent of the budget for health care goes for hospital-related women in the labor market could have repercussions on their expenses. Thus, it is important to introduce a series of consid- health and especially on the health of their children, given the erations into the estimates of health needs derived from mor- scarcity of institutionalized child care. Studies undertaken in tality data. the fast-growing assembly plants along the Mexican border First is the estimate of hospital use due to needs derived from with the United States are showing that children born to reproduction and its regulation. According to hospital records, working women in these industries tend to have lower birth at least 40 percent of the persons hospitalized in Mexico in weight than those of comparable women. Also, the demand 1986 were women who were delivering children, either normal for nurseries might increase considerably. or with complications. This number excludes other related causes of hospital admission such as female surgical steriliza- EDUCATION. If the trend in formal education that took place tion and neonatal admissions, which together might account in the 1970s continues, it might be expected that by the end for 3 to 6 percent of the total number of persons admitted to of the century almost the whole adult population will know hospitals (Mexico, Instituto Mexicano del Seguro Social how to read and write, and the vast majority will have finished 1986). primary school. The average number of years of education grew Second is the fact that at the present time the provision of from3.5to5.5yearsbetween 1970 and 1980 and to6.2 in 1990. hospital services is insufficient to meet the needs of the popu- The average might increase to more than 9 years by the year lation. Planning for future services should take into account 2010. The improvement of education and heightened expo- resources required to close the gap between supply and needs. sure to mass media is likely to increase the demand for health It is almost impossible to estimate the health needs to be met services and for a better quality of services. It can also lead to through hospital services using data from current information a greater participation of the population in taking care of its systems. The magnitude of unmet need is often a matter of own health. speculation and debate. It is, however, possible to estimate the resource equivalents that would be required to meet the health SOCIAL INEQUALITY. Mexico presents one of the highest needs derived from births, given the prevailing health care levels of wealth concentration (Hern6ndez-Laos 1984). The model. According to the 1987 National Health Survey (Mex- degree of social mobility (access of growing sectors of the ico, Direcci6n General de Epidemiologfa 1988), about 30 population to better jobs, salaries, and living conditions), percent of the births were not attended by a health profes- which was a permanent characteristic of Mexican society up sional, which in the Mexican context means an unmet need 62 ]ost Luis Bobadilla. Julio Frenk, Rafael Lozano, Tomnas Frejka, and Claudio Stern of about that magnitude. Considering that the absolute num- able conditions, such as epilepsy due to birth trauma and brain ber of births will increase about 30 percent from 1980 to 2010, cysticercosis. Addressing pretransitional pathology would re- and that 30 percent of the present deliveries are not being inforce the care of posttransitional diseases. cared for properly, the real increase in the requirement of beds Recognizing the need to control the common infectious and for birth care is about 85 percent (in relation to the total reproductive problems should not lead to the conclusion that available). The required budget to build the hospital beds to chronic diseases and injuries will have to wait their turn. Such deliver annually more than a million additional babies in the a conclusion would most likely transform the posttransitional first ten years of the next century is formidable. conditions from emerging to epidemic. To avoid this future Third, the coexistence of noncommunicable diseases and scenario, the control of chronic diseases, especially cancer and problems related to reproduction will most likely pose a di- ischemic heart disease, will require an emphasis and reliance lemma in the allocation of beds, which might not be resolved on primary prevention programs, such as those aimed at reduc- rationally. This problem can be portrayed as a competition ing the number of individuals exposed to the risk factors or the between patients: on the one hand, pregnant or delivering causal agents of disease. Even wealthy countries with large women, a large proportion of whom will come from the de- investments in health care have to turn to primary prevention, prived socioeconomic groups and the rural areas; on the other as the costs of treatment and rehabilitation rapidly increase hand, adult and elderly patients suffering from cardiovascular (L-itvak and others 1987). Thus far, Mexico, as many other disorders, cancer, diabetes, and disabilities due to accidents. middle-income countries, has given a very low priority to such There is evidence that infectious diseases and problems linked programs. The only way to reduce the demand for hospital and to reproduction are more prevalent in the rural areas and other specialized medical services by the year 2010 will be to among the least privileged socioeconomic groups. Past experi- implement vigorous campaigns in order to reduce the con- ence with the decisionmaking process suggests that the out- sumption of tobacco, alcohol, animal fats, sugar, and salt. The come in the allocation of resources between the two groups role of legislation and sanitary regulation as instniments to will be biased toward the needs of the wealthier. Reproductive control the exposure to many noxious agents has not been fully and infectious problems might become neglected health needs exploited. in the period 2000-10 if the health care model and the decisionmaking process continue as now. Reshaping the Health Care Model Changing Priorities in the Delivery of Health Services The limitations of the health care model based on curative services in hospital settings have been clearly demonstrated in Probably the top priority of countries like Mlexico, which are most industrial countries. No single country has been able to undergoing a protracted-polarized transition, is to avoid the cope adequately with the rising costs of medical care. The pernicious competition among types of pathology. Yet, impor- health needs of the Mexican population in the foreseeable tant segments of the health planning community-both at the future suggest that its imported health care model could be national and the international levels-have themselves be- exhausted soon. Aiming at delivering all births in hospitals is come polarized in two bands: those who claim that the first not only financially impossible but also medically unnecessary. order of business is to bring common infections and undernu- In a similar way, performing all surgical operations in hospitals trition under control, and those who see in the rising preva- results in undue costs. The social needs of hospitalized patients lence of chronic ailments and injuries the need for a shift in will have to be met through means other than keeping them priorities. Still, the complex reality of many countries means in hospitals. Many of these changes are being successfully that there is no alternative but to address the pretransitional implemented in developing countries, including Mexico, and the posttransitional problems simultaneously. Basic no- through demonstration projects. Still, changing entrenched tions of equity demand that the gap between knowledge and practice patterns may take many years. Until now the health action be closed for all population groups, so that the "left-over needs of the Mexican population have been growing faster ills" (Frenk and others 1989a) represented by common infec- than the ability to react and adapt to the new conditions. The tions and malnutrition cease to affect the least privileged transfer of care of some health problems from hospitals to segments of society. Eliminating the epidemiologic polariza- health centers and to the homes of the affected would require tion occupies, therefore, a top place in the list of priorities. a strong component of community participation. Of para- Furthermore, there are close links between the two groups mount importance is the strengthening of family and social of pathology. For instance, the most common reason for heart networks to maintain the support for the disabled and the surgery in Mexico continues to be valve replacement to repair chronically ill. damage produced by rheumatic heart disease-a pretransitio- Restructuring the health care model will undoubtedly re- nal condition. If the relatively inexpensive preventive mea- quire innovations. The social response to the complex health sures against rheumatic fever were implemented, specialized needs of the twenty-first century will have to be based on resources could be freed for the treatment of other heart scientifically validated information. Research to design alter- conditions for which prevention is not so effective. Similarly, native modes of effectively meeting the needs of the popula- a large proportion of resources in the complex fields of neurol- tion must be one of the key strategies of any health care system ogy and neurosurgery continue to be used to correct prevent- that aspires to shape the future. The Epidemiologic Transinon and Health Pnorities 63 Notes Mgxico. Direcci6n General de Estadistica. 1973- "Computer Data on Deaths for 1970." M6xico, D. F.: Instituto Nacional de Estadistica, Geograffa e We would like to acknowledge the valuable comments given to an earlier Informatica. version of this chapter by Dean Jamison, Henry Mosley, Joseph Decosas. and . 1984. "Computer Data on Deaths for 1980." Mexico, D. F.: Instituto Stephen Simons. Nacional de Estadistica, Geografia e Informitica. - 1988. "Computer Data on Deaths for 1985." M6xico, D. F.: Instituto Nacional de Estadistica, Geografia e Informatica. References Mgxico, Direcci6n General de Epidemiologia. 1988. Encuesta Nacional de Salud: Resultados nacionales. Mexico, D. F.: Secretarfa de Salud. Alba, Francisco. 1971. La Poblaci6jn de Mdxico: Er'olucicnN Dilensas- Mexico . 1989. Paludismo y otras enfermedades transmindas por vector. Mexico. D. F.: El Colegiis de M6xico. D. F.: Secretaria de Salud. Alt.mir, Oscar. 1974 L medi a eonomiamenteactivade M6xico, Direccl6nGeneralde Planificaci6n Familiar. 1989. EncuestaNacional Altimir, Oscar. 197g4. La medicion de Ia poblaciOn economicamente activa de deecndaySld197CoptraesMxioD.F M6xico, 1950-1970. Demnografia y Economfa 8:50-83. de Fecundidad y Salud 1987. Computer tapes. Mexico, D. F. Mexico, Instituto Mexicano del Seguro Social. 1986. "Computer data on Aparicio-Jimene:, Ricardo. 1988. Niveles, tendencias e impacto demogrAfico hospital discharges by cause and age groups." Nkxico, D. F. de la anticoncepci6n. En: Direcci6n General de Planificaci6n Familiar. Memosria de la Reunion Sohre Avances y Perspectivas de la Investigaci6n Social M6xico, INEGI (Instituto Nacional de Estadistica, Geografia e Informatica). en Planificacioin Familiar ei Mexico. Mexico, D. F.: Secretarfa de Salud. 1980. Incidencia d& la tnortalidad en los Estados LTnidos Mexicanos. Mexico. Bobadilla, lose Lis., and Ana Langer. 1990. "La morralidad infantil en D. F.: Secretariade Programaci6n y Presupuesto. Mexico: Un fen6meno en transici6n." Revista Mexicana de Sociologia . 1984. Agenda Estad(sica. M6xico, D. F.: Secretarfa de Programaciin 1:111-32. y Presupuesto. Camposortega, Sergio. 1988. "El nivel v Ia estructura de la mortalidad en - 1986. Es[ad-sncas Histoncas de Mexico. Mexico, D. F.: Secretaria de MNxico, 1940-1980." In Mario) Bronfrnani andJosiS G6mezde Leosn. eds., La Programaci6n y Presupuesto. mortalidad en NlV_xco: Niveles, tendencias y detenrtnantes. Mexico. D. F.: El . 1992. Peffil Sociodeniografico. Xl Censo General de Poblacion y Colegio de Mexico. Vivienda.Mexico, D. F.: INEGI. Consejo Consultivo del Programa Nacional de Solidaridad. 1990. El Conbate Mexico, INEGI y CONAPO (Instituto Nacional de Estadistica, Geograffa e a la Pobreza. M6xico, D. F.: El Nacional. InformAtica v Consejo Nacional de Poblaci6n). 1985. Proyecciones de la poblacidn de MiSico y de las Entidades Federaives: 198Q-2010. Mvrxico. D. F.: Donabedian,Avedis. 1973.AspectsofMedicalCareAdministration.Cambridge, Secreraria de Programacifln y Presupuesro. Mass.: Harvard University Press. Secretarfa de Programaci6n y Presto e - Me~~~~~~~~~~~lxico, Secretarfa de PrograrracO ...........n v Presupuestoi y Secretarfa cde Salubridad Economic Commission for Latin America and the Caribbean. 1986. Statistical v Asistencia. 1984. Boletin de infornaciOn Esradistica No. I Mexico, D. F. Yearbook for Latin America and the Canbbean. New York: United Nations. Mexico, Secretarfa de Salubridad y Asistencia. 1946. Anuario Estadsrica & Frenk. Julio, Jose Luis Bobadilla, Jaime Sepulveda, and Malaquias L6pez- SalubridadyAsistencia Mexico, D. F.: SecretariadeSalubridadyAsistencia. Cerv-antes. 1989a. "Health Transition in Middle-Income Countries: New Challenges for Health Care." Health Policy and Planniltig 4:29-39. Secreraria de Salusd. Frenk, Julio, Jose Luis Bobadilla, Jainme Sepulveda, Jorge Rosenthal, and Nufiez, Leopoldo, and Lorenzo Moreno. 1986. Proyecciones de la poblacion EnriqueRuelas. 1988b. "AConceptLial Model for Public Health Research." urbana y rural, 1980-2010. M6xico, D. F.: Asociaci6n Mexicana de PAHPR Bulletin 21:60 71. Demograffa M6dica. Frenk, Julio, Tonias Frej ka, Jos6 Luis Bobadilla, Claudio Stern, Jaime Seprlve- Omran, Abdel. 1971. "The Epidemiologic Transition: A Theory of the Epide- da, and Marco Jose. 1989. "The Epidemiological Transition in Latin miology of Population Change." Milbank Memorial Fund Quarterly Am&ica." In iussp (International Union for the Scientific Study of Popu- 49:509-38. lation), International PopulationConference. New Delhi. Liige 1989. Oiirais A. R. 1983. "The Epidemiliologic Transition Theorv: A Prelimsinary Frenk. Julio, Hector Hernandez-Llamas, and Lourdes Alvarez-Klein. 1980. "El Update." Journal of Tropical Pediatrics 29:305-16. mercado de trabajo mOdico: 11. Evolucidn hi5t6rica en Mexico." Gaceta Soberi5n, Guillermo, Jesus Kumate, and Jos6 Laguna. 1988. La salud en MNxico: Mddica de Mdxico 116:20-2 1 and 57-60. Testimonto Vol. 1, Problemas y programas de salud. M6xico, D. F.: Fondo de G6rnez-de-LOon. Jose, Julio Frenk, 1992. Population and Health Challenges in Ctiltura Econ6mica. Mexico on the Eve of the 2 1st Century. Unpublished. Trejo, Sa6l. 1988. Empleo para todos: El reto y los caminos. Mexico, D. F.: Fondo HemAndeT-Laos, Enriquie. 1984. "La desigualdad regional en Mexico (1900- de Cultura EconOmica. 1980)." In Rtolando Cordera and Carlos Tello, eds., La desigualdad en Me_nco. United Nations. 1986. World Populanon Prospects. Population Studies No. 98. Mexico, D.F.: Siglo Veintiuno. New York: United Nations. Jones, G. W. 1975. "Population Growth and Health and Faniily Planning." - 1989a. Case Study in Population Policy: Mexico. New York. In W. C. Robinson, ed., Population and Developrnent Planning. New York: - 1989b. World Population Prospects, 1988. New York. Poppulation Council. Wildavsky, Aaron. 1977. "Doing Better and Feeling Worse: The Political Litvak, Jorge, L. Ruiz, Helena E. Restrepo, and Alison McAlister. 1987. "El Pathologyof Health Policy." ln DoingBetterand Feeling Worse: Health in the problema creciente de las enfermedades no transmisibles, un desaffo para Unired States. Proceedings of the American Academv of Arts and Sciences, los paises de las Americas." Boletin de la Oficina Santiaria Panamericana 106, no. 1. Boston, Mass. 103:433-449. Wilkie. James W. 1978. La Revolucion Menicana: Gasto federal y cambio social. L6pe--Gallardo,Juan 1984."Ladistribuciondel ingreso en Mexico: Estructura Mexico. D. F.: Fondo de Cultura Econ6mica. v evoluci6n." In R. Cordera and C. Tello, eds., La desigualdad en Mexico. World Bank. 1992. World Development Report 1992. Development and the .Mexico, D.F.: Siglo Veintiuno. Environment. Washington, D.C. PART TWO The Unfinished Agenda, I Infectious Disease Acute Respiratory Infecton DiarTheal Diseases Poliornyeliis Hebrdnth Infection Measles Tetanus Rheurnauc Hean Disease Tuberculosis Leprosy Malaria Dengue Hepantis B 4 Acute Respiratory Infection Sally K. Stansfield and Donald S. Shepard Acute respiratory infection (ARI) is the most frequent illness Common diagnostic categories for uncomplicated ARI with globally and a leading cause of death in the developing world. etiologic and clinical correlates are detailed in table 4-1. As Among children under five alone, four million deaths annually suggested by this table, ARI includes the minor upper respiratory are ascribed to ARI, most of which are due to pneumonia. That infections (URIs), such as colds and sore throats, in addition to mortality due to pneumonia is ten to fifty times higher in the more serious (and potentially fatal) acute lower respiratory developing countries suggests that there is ample room for infections (ALRIS) of pneumonia and bronchiolitis. improvement in addressing this important public health prob- Most of the studies of ARI from developing countries have lem. The heterogeneity of the clinical presentations and caus- been conducted among infants and children. Programs which ative organisms in ARI, however, has hampered efforts to design have been developed to prevent or treat ARI have often focused simple and effective interventions. exclusively on children, on the argument that the principal The classification and management of ARI in the industrial- opportunities to reduce ARI mortality are among children under ized world are founded on epidemiologic, radiologic, and mi- five. Although adults, particularly the elderly, may benefit crobiologic data, in addition to clinical history and physical from preventive and therapeutic interventions, the most sig- examination. The syndromes of ARI, which are complex clini- nificant reduction in years of life lost will be seen among cal conditions of varying etiology and severity, are most fre- infants and children. The data and strategies outlined in this quently categorized on the basis of anatomical location. chapter therefore focus primarily on children under five. Table 4-1. Clinical Summary of Acute Respiratory Infections in Infants and Young Children Age at peak incidence Type Diagnosis Most common etiology (months) Mortality Upper respiratory Nasopharyngiris Viral (various) - No infections (coryza, colds) Otitis media (middle ear Bacrerial (pneumococcus, infecrion) Hemophilus influenzae) 6-7 No Pharyngo-tonsillitis Viral (various) and bacterial - No (except diphtheria) (Streptococcus pyogenes, Corynebacterium diphtheriae) Epiglottiris Bacterial (Hemophilus influenzae) 24-47 Yes Lower respirarory Laryngitis (croup) Viral (especially parainfluenza and 12-23 Rare infections measles) Tracheobronchitis Viral and bacterial (various) Constant No Bronchiolitis Viral (RSV, parainfluenza 3) 0-11 Yes Pneumonia Bacterial (pneumococcus, 24-35 Yes Hemophilus influenzae) and viral (RSV, influenza, parainfluenza, measles, adenovirus) - Not available. Source: Authors' data. 67 68 Sally K. Stansfield and Donald S. Shepard Risk Factors for ARI quiring hospitalization among persons from lower socioeco- nomic groups and in more crowded households. Aaby (1988) Treatment of pneumonia clearly reduces ARI mortality, but the has suggested that crowding is a predictor of an increased definitive solution to the problem of high numbers of AR[ case-fatality ratio due to measles, in addition to an increased deaths in developing countries will ultimately be found in risk of infection. prevention of pneumonia. Although the epidemiologic data Both poverty and crowding may, however, be proximate from the developing world are limited, a review of the available measures for other known or as yet unrecognized risk factors. information suggests possible ways to reduce ARI mortality For example, the frequent association of these factors with through reducing the risk of pneumonia. Table 4-2 summarizes lower educational levels, poor nutrition, and certain child-care the following discussion of some of the known and suspected practices further confounds analysis of risks for pneumonia. risks for pneumonia incidence and mortality. Existing evidence suggests, however, that infants with re- stricted respiratory excursion of the chest wall due to obesity Age and Sex (Tracey, De, and Harper 1971) or swaddling (Yurdakok, Yavuz, and Taylor 1990) may have increased risk of pneumo- The incidence of ARI (most of which is URI) is inversely related nia. Family stress also increases the risk of infections such as to age, peaking at four to nine infections in each of the first pneumonia among both children and adults (Foulke and oth- two years of life, dropping to three to four by school age, and ers 1988; Graham and others 1990; Cohen, Tyrell and Smith remaining at two to three per year for adults (Datta Banik, 1991), probably because of interference with immune compe- Krishna, and Mane 1969; Kamath and others 1969; Monto and tence (Kiecolt-Glaser and Glaser 1986). Although chilling is Ullman 1974; Friej and Wall 1977; Spika and others 1989). frequently cited as a risk factor for URI or pneumonia, most The frequency of pneumonia and the case-fatality ratio, how- studies have provided no evidence of this association (Jackson ever, are highest among both the very young and the very old and others 1963; Douglas, Lindgram, and Cough 1968). (Bulla 1978; Berman and others 1983; Ngalikpima 1983). Studies in several developing countries have demonstrated Nutritional Status and Practices that pneumonia occurred 1.5 to 1.8 times as frequently among infants as among children two to fours years of age (Berman Poor nutrition lowers both systemic and local defenses against and McIntosh 1985). There is a slightly increased incidence of ARI, including reduction of the effectiveness of epithelial bar- both overall ARI and pneumonia among male children (Bulla riers, systemic immune responses, and cough reflexes. Nutri- 1978; Berman and others 1983; Narain and Sharma 1987; tional status is inversely related to both the incidence and the Selwyn 1990), although female children have been observed case-fatality ratio for pneumonia (Kielmann and McCord to have a higher case-fatality ratio in some countries, probably 1978; Pio, Leowski, and Luelmo 1982; Berman 1983; Sommer as a result of poorer access and quality of care during illness 1983; Sommer, Katz, and Tarwotjo 1984; Berman and others episodes (Tupasi and others 1990). 1991). Investigators have documented an incidence of pneu- monia twelve to twenty times greater in undernourished chil- Socioeconomic Status and Child-Rearing Practice dren than in children of normal weight-for-age (James 1971; Berman and others 1983; Tupasi and others 1988). Mortality Low socioeconomic status and crowding have been well doc- due to each of these already more frequent episodes of pneu- umented as risk factors for mild respiratory infections in the monia increases two- to thirteenfold for each decile below 80 industrialized world. Studies in developing countries (Verma percent weight-for-age (Escobar, Dover, and Duenas 1976; and Menon 1981; Stansfield 1987; Tupasi and others 1988; Kielmann and McCord 1978; Tupasi 1985). Borrero and others 1990; Vathanophas and others 1990) have While nutritional deficiency diseases augment the chances also demonstrated an increased frequency of pneumonia re- of ARI episodes, so episodes of ARI contribute to nutritional Table 4-2. Risk Factors for Pneumonia Increased incidence Increased case fatality Age less than two years or more than sixty-five Age less than two years or more than sixty-five Male Low socioeconomic status Poor nutritional status Poor nutritional status Low birth weight Low birth weight Lack of breastfeeding (in infants) Lack of breastfeeding (in infants) Smoking, air pollution Lack of matemal education Crowding Reduced access to health care Incomplete immunization Crowding Swaddling Underlying chronic disease Vitamin A deficiency Source: Authors' data. Acute Respiratory Infection 69 deficiency, thus further increasing the risk of subsequent infec- the effect of vitamin A supplementation in children from tion and death. A prospective study in the Gambia (Rowland, areas with endemic vitamin A deficiency (who are, therefore, Rowland, and Cole 1988) showed that pneumonia reduced presumed to be subclinically deficient) have not, however, weight gain in young children by 14.7 grams for each day of demonstrated an effect on ARJ-specific morbidity or mortality infection. Recurrent ARI episodes, as a principal cause of the (Rahmathullah and others 1990; Vijayaraghavan and others weight shortfall during infancy, therefore progressively in- 1990; Rahmathullah and others 1991). The effect of vitamin crease the risk of death due to other childhood diseases. A supplementation on ARI-specific morbidity and mortality Low birth weight, seen in 20-40 percent of infants in many among children who are not clinically xerophthalmic remains developing countries, also increases the risk and case-fatality speculative. ratio of pneumonia. Studies (Datta 1987; WHO 1988) have shown relative risks of mortality due to pneumonia which are Smoking and Air PoUution 2.5- to 8-fold greater among infants of low birth weight. Other than malaria and tobacco chewing, the only factors associated There is a large and expanding literature from industrialized with low birth weight for which cause-and-effect relationships countries on the increase in risk of pneumonia from active and have been established in developing countries (and which are passive smoking. Investigators in both industrialized and de- modifiable over the short term) are low prepregnancy weight, veloping countries have demonstrated a 1.5- to 4-fold in- low gestational weight gain, and low caloric intake (WHO/EPI creased incidence of pneumonia among smokers and among 1987). Short birth intervals, teenage pregnancy, certain geni- children whose parents smoke (Harlap and Davies 1974; Lee- tal infections, and arduous work after mid-pregnancy are other der and others 1976; Ekwo, 1983; Weiss and others 1983; Ware potentially modifiable factors associated with low birth weight. and others 1984; Pedreira and others 1985; Chen,Wanxian, Although reduction of the incidence of low birth weight would and Shunzhang 1986; Burchfiel and others 1986; Lipsky and be expected to reduce AR] mortality, no prospective studies others 1986; Samet, Marbury, and Spengler 1987; USDHHS have demonstrated the feasibility and effectiveness of inter- 1989). Maternal smoking also predisposes to low birth weight ventions to address this important problem. (Martin and Bracken 1986; Ruben and others 1986), thus The few well-conducted studies on infant feeding practices increasing the risk of pneumonia mortality for the infant after and the incidence of pneumonia demonstrate a protective birth. There are no prospective data currently available from effect of breastfeeding. The literature has suffered from wide developing countries to establish that programs to reduce variations in definitions, both of specific feeding practice and smoking will reduce ARI-specific mortality. The recent alarm- of ARI. Although several studies summarized in a review by ing increases in the numbers of persons who smoke in devel- Jason and others (1984) failed to document any protective oping countries, however, argue for prompt intervention, effect of breastfeeding, others have found a two- to fivefold particularly since successful reduction of smoking may be ex- decreased incidence of pneumonia (Chandra 1979; Singhi and pected to yield health benefits beyond the reduction of AR] Singhi 1987) and decreased case-fatality ratio due to pneumo- morbidity and mortality. nia (LePage, Munyakazi, and Hennart 1981). A more rigorous Exposure to both outdoor and indoor air pollution have study in southern Brazil (Victors and others 1987) demon- been suspected to increase the risk of ARI in many developing strated that infants who were completely weaned had a risk of countries (Kamat and others 1980; WHO/UNEP 1988; Chen and death due to pneumonia 3.6 times higher than breastfed in- others 1990). There is growing concern regarding the health fants. effects of the products of combustion (including carbon mon- Vitamin A deficiency, which often accompanies protein- oxide, particulates, and sulfur and nitrogen dioxides) from calorie malnutrition, results in keratinization of the respiratory cooking and heating fires. It has been estimated that 300 epithelium and depression of the immune response, thus pre- million to 400 million people, mostly in the rural areas of sumably decreasing both local and systemic resistance to bac- developing countries, are adversely affected by these organic terial colonization and infection. Still, the literature on fuel emissions (de Koning, Smith, and Last 1985). Although vitamin A deficiency and its association with ARI morbidity there is a clear relation between exposure to such emissions and mortality is sparse and controversial. Two studies (Som- and chronic obstructive pulmonary disease (WHO/EPP 1984; mer, Katz, and Tarwotjo 1984; Bloem and others 1990) have Chen and others 1990), the relation to pneumonia in the suggested a two- to fourfold increase in the relative risk of ARI developing world is less well documented. associated with serologic or ophthalmic signs of vitamin A Indoor particulate concentrations, probably the best single deficiency. In the Sommer study (1983), mortality in the indicator of toxic (noncarcinogenic) effects, are twenty times clinically vitamin A-deficient group was 8.6 times that in higher in the villages of developing countries than in house- non-xerophthalmic children. holds where two packs of cigarettes are smoked per day Several prospective studies have noted a reduction in over- (Pandey, Boleij, Smith, and Wafula 1989). Several studies in all mortality among children whose diets were supple- developing countries have suggested that an increased inci- mented with vitamin A. Vitamin A supplements given to dence of pneumonia is associated with exposure to organic fuel children with severe pneumonia or measles has improved emissions (Sofoluwe 1968; Kossove 1982; Honicky 1985; clinical outcome and reduced mortality (Barclay, Foster, and Campbell, Armstrong, and Byass 1989; Penna and Duchiade Sommer 1987; Hussey and Klein 1990). Prospective studies of 1991), although several studies have had problems with con- 70 Sally K. Stansfield and Donald S. Shepard founding variables such as socioeconomic status and crowding. manifestations may occur in the absence of the typical measles One study in Nepal (Pandey, Neupane, Gautam, and Shrestha rash. Common complications of measles include growth falter- 1989) has demonstrated that the number of episodes of life- ing, chronic diarrhea, oritis media (middle ear infection), threatening pneumonia among children under two is directly encephalitis, and pneumonia. Pneumonia, including primary proportional to the reported hours per day spent near the stove. measles pneumonia as well as superinfection by viruses and Studies in the Gambia suggest that carriage on the mother's bacteria, is the most common complication of measles and back during cooking may predispose children to pneumonia often represents the principal proximal cause of death. (Armstrong and Campbell 1991). Prospective trials are re- In unimmunized populations, epidemics occur in two-year quired to assess the effectiveness of interventions such as cycles with secondary attack rates exceeding 90 percent among improvements in stove design, improved ventilation, and be- susceptible household contacts (Keja and others 1988). Al- havioral change to reduce exposure. though generally a disease of childhood, measles can occur at any age in susceptible populations. Infants in industrialized Clinical Syndromes Causing ARI Mortality countries are not usually affected under the age of six to eight months, presumably because of placentally transmitted mater- The predominant known causes of ARIl mortality are bacterial nal antibodies. In parts of Africa, however, 20 to 45 percent of and viral pneumonia, measles, and pertussis. Additional epi- children are infected with measles before they attain the demiologic data are needed to characterize the importance of recommended age for immmunization at nine months. other clinical syndromes and etiologic agents, including diph- Although improving immunization coverage progressively theria, bacterial pharyngitis, and the "opportunistic" viral and reduces infection rates, it was estimated in 1989 that there bacterial infections which are likely important causes of pneu- were 70 million annual cases of measles, and that 1.5 million monia mortality among the very young, the very old, and those to 2 million of those affected would die during the month immunocompromised by acquired immunodeficiency syn- following infection. Although generally a mild disease in tem- drome (AIDS) or malnutrition. perate climates, an estimated I to 5 percent of all affected children in developing countries will die of measles or its Pneumonia complications. Children who survive the acute episode have an increased risk of mortality for weeks to months following infec- Pneumonia is an inflammatory process of the pulmonary inter- tion. Most investigators, therefore, report deaths which occur stitial space or alveoli which may be diffuse or confined to lung within one month of the measles rash as "measles-associated." segments or lobes. Clinically, patients with pneumonia most Partly because of such variations in methods of ascertaining frequently present with cough and tachypnea (rapid breath- deaths due to or associated with measles, the reported case- ing);retractions (indrawing of the lower chest wall on inspira- fatality ratios vary widely. Williams and Hull (1983) docu- tion) may also be present in more severe cases. Among menred a 5 percent case-fatality ratio during the acute phase neonates and younger infants, however, cough is often absent. of the disease, and a cumulative rate of 15 percent during the Available information from developing countries suggests nine months following the rash. The case-fatality ratios ob- that more than 75 percent of ARI deaths are caused by pneu- tained from prospective population-based studies range from 2 monia, both bacterial and viral (Bulla and Hitze 1978; Berman percent in Bangladesh to 34 percent in Guinea Bissau ('HO/EPI 1991). Microbiologic data is difficult to obtain and of variable 1987). Rates of 50 percent or more have been described in quality, yet most investigators agree that the bulk of ARI mor- severely undernourished populations. It has been suggested, tality among both children and adults is due to pneumonias however, that deaths prevented by measles immunization will caused by two bacteria, Streptococcus pneumoniae and be "replaced" by deaths from other causes, such that measles Hemophilus influenzae (Berman and others 1983; Denny and immunization may prevent fewer deaths than these mortality Clyde 1983; Shann and others 1984; Selwyn 1990; WHO/ARI ratios would suggest. 1991b). Viral agents which cause fatal pneumonia include respiratory syncytial virus (RSV), measles, parainfluenza, influ- Pertussis enza, and adenovirus. Mixed viral and bacterial infections are frequently documented (Berman 1991). Clinical malaria has The majority of cases of whooping cough, or the pertussis also been found to coincide frequently with the clinical and syndrome, are vaccine-preventable infections caused by radiologic diagnosis of pneumonia (Byass and others 1991). Bordetella pertussis. The paroxysms of coughing, often associ- ated with a characteristic inspiratory gasp (the whoop), may Measles persist for four to ten weeks. Pertussis is often associated with dehydration and weight loss; and encephalitis is an occasional Measles is a vaccine-preventable disease causing an acute complication. Pneumonia, resulting either from the organism febrile eruption which occurs naturally only in humans. The itself or from secondary bacterial infection, is the proximal viral infection itself may result in any of several clinical syn- cause of death in over 90 percent of cases. dromes, including croup (laryngotracheobronchitis), bronchi- Although pertussis occurs endemically, it tends to produce tis, bronchiolitis, or even viral pneumonia, particularly in epidemics every three to four years, with up to 90 percent of children immunocompromised by severe malnutrition. These exposed susceptibles developing the disease (Broome 1981; Acute Respiratory Infection 71 Muller, Leeuwenburg, and Pratt 1986). Incidence is higher view of additional concerns regarding the cost and insensitiv- among girls than boys. Population-based studies have sug- ity of the laboratory tests, and the lack of criteria to distinguish gested an annual incidence of I to 5 percent among children streptococcal pharyngitis on clinical grounds, it is currently under fifteen, although infants have a 16 percent chance of difficult to establish in developing countries a strategy for infection in Kenya (Voorhoeve and others 1977). The case- management of pharyngitis which will effectively prevent fatality ratio averages about I percent, although up to 15 poststreptococcal complications. Antibiotic prophylaxis for percent of cases were fatal in studies in Uganda (Bwibo 1971) patients with a history of rheumatic fever has, therefore, been and Santa Maria Cauque (Mata 1978). The highest mortality recommended as the most feasible strategy to prevent rheu- is observed among females and children under two, with an matic heart disease in developing countries (WHO 1988). estimated 500,000 to I million infant deaths annually due to pertussis (Muller, Leeuwenburg, and Pratt 1986; Keja and Other Causes of ARI Mortality others 1988). Additional causes of AR] mortality include viral bronchiolitis Diphtheria and epiglottitis. Bronchiolitis, especially that resulting from RSV and parainfluenza 3, may be responsible for up to one-third The epidemiology of diphtheria in the developing world is of ALRI among children under five, most of which occurs in poorly understood. Although the causative organism, Coryne- infants (Cherian and others 1990). The virology of these bacterium diptheriae, is widely present in Africa, and over 96 infections is apparently similar to that observed in industrial- percent of unvaccinated adults are immune (Ikejani 1961; ized countries (Selwyn 1990). The difficulty of the laboratory Muyembe and others 1972), there are few reported cases of this techniques and lack of cost-effective measures for prevention vaccine-preventable disease. It has been suggested that im- and treatment of these infections have hampered efforts to munity may result from subclinical or misdiagnosed infections, address these important causes of mortality. an explanation supported by the finding of carriage of the Epiglottitis, which is usually caused by Hemophilus influenzae organism in 4 to 9 percent of the population (Ikejani 1961; type b, is an occasional cause of death when the infected Muyembe and others 1972). epiglottis obstructs respiration. Additional epidemiologic in- There are no community-based studies, but data from hos- vestigations are needed to define the role of other organisms pitals suggest diphtheria may be an important cause of pharyn- as causes of mortality due to pneumonia, including group B gitis and croup. Of 180 children hospitalized with respiratory streptococcus, Chlamydia trachomatis and C. pneumoniae, My- infections in Colombia (Escobar, Dover, and Duenas 1976), coplasma pneumoniae, Ureaplasma urealyticum, and Pneu- seven of the nine cases of croup were caused by diphtheria. mocystis carinii. These bacterial and parasitic pneumonias may Investigators in the Gambia found evidence to suggest an be important causes of mortality, especially among neonates annual incidence of 6 per 1,000 children under five. Salih and or persons immunocompromised, such as by malnutrition or others (1985) have reported epidemic diphtheria and suggest AIDS. Tuberculosis and some helminthic infections may also that it is one of the most important diseases of childhood in present as pneumonia; these more chronic infections are often the Sudan. distinguished clinically by their failure to respond to the usual antibiotic therapy. Pharyngitis Public Health Significance Pharyngitis is an upper respiratory tract infection that is most commonly viral and, therefore, self-limited. Bacterial pharvn- Comparison of results from investigations on the public health gitis, although less common, is of greater public health import- significance of ARI in different countries is all but prevented by ance. Though acute bacterial pharyngitis (except when due to wide variations in study design, case definitions, and culture diphtheria) is not a significant primary cause of mortality, techniques. Meaningful comparison of study results is difficult, acute rheumatic fever (ARF) is an occasional late complication for example, when some investigators have used sensitive case of untreated pharytngitis caused by group A betahemolytic definitions which include all coughs and colds, whereas others streptococci (Streptococcus pyogenes). Acute rheumatic fever focus only on the mnore severe ARI that comes to the attention has been reported at rates of 27 to 100 cases per 100,000 per of health care workers. year (WHO 1988), although it is much less frequent in industri- alized countries. Microscopic cardiac damnage during ARF may Current Levels and Trends in the Developing World progress over subsequent years, frequently causing incapacita- tion and, ultimately, death owinig to changes in cardiac function. The few well-conducted community-based prospective studies Antibiotic therapy of bacterial pharyngitis is recommended performed suggest that overall incidence ofARI in the develop- in industrialized countries to prevent ARF and other sequelae ing world is similar to that observed in the industrialized world. of streptococcal pharyngitis. Management of streptococcal pharyngitis has been controversial, however, and even less is MORBIDITY AND MORTALITY LEVELS, CIRCA 1985. Prevalence known of the epidemiology of streptococcal disease to guide figures show that children spend from 22 to 40 percent of its managenment in the developing world (Markowitz 1981). In observed weeks with AR], and from I to 14 percent of observed 72 Sally K. Stansfield and Donald S. Shepard weeks with ALRI, such as pneumonia or bronchiolitis. Acute death where diarrheal disease mortality rates have been suc- respiratory infections account for 20 to 40 percent of adult cessfully reduced (Chen, Rahman, and Sarder 1980; Zimicki outpatient consultations and 20 to 60 percent among children. 1988). Mortality from ARI has increased in relative importance Of all pediatric admissions to hospitals, 12 to 45 percent are even in settings where high coverage with measles immunization for ARI, whereas 20 to 30 percent of adult inpatients have been has been achieved (Greenwood and others 1988; Zimicki 1988). admitted for ARI treatment (Bulla and Hitze 1978; PAHIO 1980; Data from industrialized countries suggest that changes in Leowski 1986). immunization policy may increase the incidence of disease. The reported incidence of ALRI varies widely, from country With intensive control efforts, for example, the incidence of to country as well as with age and nutritional status. Whereas measles in the United States had fallen to the lowest level ever the annual incidence of pneumonia is 3 to 4 percent in chil- recorded in 1983. When expenditure for immunization was dren under five in the industrialized countries, it ranges from reduced in 1984, however, increased outbreaks were observed. 10 to 20 percent in most developing countries, reaching as high A similar resurgence of pertussis has been noted in countries as 80 percent in populations with a high prevalence of malnu- in which changes in public opinion or immunization policy trition and low birth weight. In Papua New Guinea in 1973, have led to a reduction in immunization coverage. for example, there were 72 episodes per 1,000 children of one to four years of age and 1,074 episodes per 1,000 infants (Riley Possible Morbidity and Mortality Patterns: 2000 and 2015 and Douglas 1981). In Costa Rica an annual incidence of pneumoniaof37 per 1,000 children was observed among those As viral upper respiratory infections, which account for the of normal nutritional status, while the rate was 457.8 per 1,000 bulk of ARI morbidity, are unlikely to be eradicated in the among malnourished children (Pio, Leowski, and ten Dam foreseeable future, the overall incidence of ARI is likely to be 1985). The overall incidence of ARI, most of which is coughs substantially unchanged for the next twenty-five years. Con- and colds, is comparable to that in the industrialized world. siderable opportunity exists, however, to reduce the incidence The greater public health importance of ARI in developing of vaccine-preventable ARt and to reduce the case-fatality ratio countries is manifest, however, in the increased frequency of for pneumonia, thereby reducing ARI mortality. lower respiratory tract infections and in the disease-specific Changing demographic patterns, such as birth spacing and mortality rates that are ten to fifty times higher than in consequent improvements in nutritional status would be ex- industrialized countries (WHO 1984; Mohs 1985; Camargos, pected to substantially reduce mortality due to pneumonia Guimaraes, and Drummond 1989). Most vulnerable to death during the next twenty-five years. Increased life expectancies due to pneumonia are the very young and the very old. may later create larger populations of the elderly, among whom Of the estimated 15 million deaths occurring each year pneumonia will likely remain a significant cause of mortality. among children under five, 25 to 30 percent are due to ARL. As Progress in improving the access to and quality of care will be the cause of approximately 4 million deaths annually among instrumental in controlling mortality among both the young this age group alone, ARI often surpasses diarrhea in importance and the elderly. as a cause of mortality (Bulla and Hitze 1978; Balint and Of potential future concern, however, is the evolution of Anand 1979; Shann and others 1984; Pio, Leowski, and ten antimicrobial resistance among the pathogens causing bacte- Dam 1985; Spika and others 1989). Pneumonia causes from 2 to rial pneumonia, which may interfere with the effectiveness of 8 percent of adult deaths in countries for which data are available interventions. Although the development of newer antimicro- (Hayes and others 1989), rankingfrom second to tenth as acause bials has, to date, kept pace with the evolution of resistance to of death among those age fifteen through sixty-four. them, the cost of later-generation antibiotics will not be so easily bome in developing countries. And there is evidence to TRENDS IN THEPERIOD 1970TO 1985. Although surveillance data suggest that inappropriate use of antimicrobials, so frequent for overall ARI morbidity in the developing world are limited, throughout the world, speeds the evolution of resistance. it is likely that these rates have remained unchanged in the Coverage with the vaccines currently included in WHO's past fifteen to twenty years, just as they have in the industrial- Expanded Programme on Immunization (EPI) may be expected ized countries. Reductions in incidence of ARI as a result of to continue to increase, also leading to reduction in the num- improved immunization coverage (with measles, diphtheria, ber of deaths from ARI. In addition, improved vaccine technol- and pertussis vaccines) would have little effect on the overall ogy will likely alter the currently observed patterns of mortality incidence of AR[, because the frequency of viral upper respira- due to ARI during the next twenty-five years. New vaccines, tory infections would remain largely unchanged. too, increase hopes of reducing childhood ARI mortality due to Pneumonia mortality, however, has been reduced signifi- measles and bacterial pneumonias caused by S. pneumoniae and cantly over the past fifteen to twenty years in the United States H. influenzae. for all age groups except the elderly. Similar reductions in mortality would be expected in developing countries where the Economic Costs of AR[ risk factors such as nutritional or socioeconomic status, im- munization coverage, and access to health care have improved. Acute respiratory infections account for an average of 35 percent In many countries, however, AR! has increased in relative of all outpatient visits globally (Bulla and Hitze 1978) and gener- importance, frequently emerging as the first cause ofchildhood ally similar proportions of all hospitalizations among children. Acute Respiratory Infection 73 DIRECT COSTS. The minimal direct cost of ALRI for children in programs to reduce ARI morbidity and mortality. It has been in the first two years of life in the United States has been estimated that deaths due to the four vaccine-preventable estimated at $35.14 per child, 56 percent of which is attribut- respiratory diseases (measles, diphtheria, pertussis, and tuber- able to hospitalization (McConnochie, Hall, and Barker 1988). ' culosis) may account for up to 25 percent of the total mortality In many developing countries, the economic burden of among children under five in the developing world. treatment of ARI already exceeds the expected cost of ARI case Although the immunization programs must be part of any management with improved effectiveness and broader cover- strategy to prevent ARI, available data do not yet justify the age. More appropriate use of existing health personnel and design and implementation of programs to reduce environ- pharmaceutical resources might be expected, in many coun- mental and nutritional risk factors for ARI control. The evi- tries, to avert mortality with little or no additional expendi- dence does suggest, however, that such programs may be ture. For example, the prevalence of the inappropriate use of effective. Several potential preventive interventions that pharmaceuticals for the management of ARJ suggests that a net might be considered for inclusion in AR] control programs have cost savings might be achieved by improving use pattems been included in the following discussion. Those for which (Stansfield 1990; Foreit and others 1991). Frequently, more evidence of feasibility and effectiveness are strongest are in- than half of antibiotic use is unnecessary (Hossain, Glass, and cluded in a comparative model of cost-effectiveness, which is Khan 1982; Stein and others 1984; Quick and others 1988). summarized in table 4-3. It is important to recognize, however, A study in Peru (Foreit and Lesevic 1987) showed that approx- that the actual benefits from these interventions would be imately 50 percent of the expenditure for medications to treat broader than those calculated, since each would reduce mor- ARI episodes was inappropriate, at an excess cost of $18.47 to bidity and mortality resulting from many health problems $21.97 per child covered. The authors of the study estimated beyond ARI alone. Sources for the data used and the methods that an 89 percent reduction in treatment costs would be for calculating the cost-effectiveness estimates are specified in achieved through altering outpatient treatment of ARI to con- the appendix to this chapter. form to WHO guidelines. Both inappropriate prescription of antibiotics and poor compliance probably also contribute to MEASLES IMMUNIZATION. Operational problems in maintain- the development of antimicrobial resistance and will greatly ing the necessary cool temperatures for handling the measles increase the future direct costs of ARI treatment as the use of vaccine are a frequent barrier to maintaining vaccine viability more expensive antimicrobial agents becomes necessary. and efficacy. The World Health Organization has estimated the efficacy of the vaccine to be 90 percent when maintained INDIRECT COSTS. Acute respiratory infections account for an at appropriately cool temperatures (Keja and others 1988). average of one-third of all absences from work (Bulla 1978). Becauseofvariabilityinstudydesignandalterationsinvaccine In Britain, one to two weeks of schooling are lost per child per viability as a result of handling, the measured efficacy of the year due to ARI (Crofton and Douglas 1975). Data from Ghana vaccine may vary broadly, although Hull, Pap, and Oldfield (Ghana Health Project Assessment Team 1981) indicate that (1983) achieved an efficacy of 89 percent in the Gambia. over 94 percent of the fifty-two days of life lost per case of ARI Considerable controversy surrounds the issue of immuniza- is due to mortality rather than disability. Particularly in the tion strategy for measles. Studies with the currently available setting of developing countries, where case-fatality ratios are (Schwartz) vaccine have demonstrated that residual levels of high and access to services limited, the bulk of costs attribut- matemal antibody restrict the effectiveness of the vaccine in able to ARI are indirect costs due to mortality. No such esti- the first few months of life. Available data regarding age-spe- mates are available for the developing world, but ARI also likely cific seroconversion and measles incidence rates suggest that takes a relatively greater toll in these settings in the form of immunization at nine months of age will prevent the maximal growth deficits, malnutrition, and resulting leaming disabili- number of cases (WHO/EPI 1982). These data were the basis for ties. Although these indirect costs of ARI are difficult to quan- the WHO recommendation of one dose of measles vaccine to be tify, it is probable that they greatly reduce the potential given between nine and twelve months of age. productivity of those affected. Yet, in many countries, 20 to 45 percent of measles cases occur among infants before nine months of age, when they are Lowering the Incidence of ARI most vulnerable to measles mortality. It had been suggested that "herd immunity" achieved with adequate immunization Possible preventive approaches to reduce ARI morbidity and coverage among older infants and children may serve to reduce mortality include immunization and alteration of other risk the infection rate among younger infants (Black 1982; factors which predispose children and adults to pneumonia. Heymann and others 1983). Recent evidence, however, sug- gests that in areas of high population density, there is no shift Elements of the Preventive Strategy in the age distribution of cases or reduction in incidence greater than the level of vaccination coverage (Dabis and Although the data are adequate to support the use of im- others 1988; Taylor and others 1988). Particularly in the urban munization in the control of ARI, the limitations of current areas of Africa, the increased transmission rates may lower the knowledge regarding the feasibility and effectiveness of other optimal age for immunization (McLean and Anderson 1988; preventive strategies are, for the moment, a barrier to their use Taylor and others 1988). 74 Sally K. Stansfield and Donald S. Shepard Table 4-3. Calculated Cost-Effectiveness of Interventions for ARi Control (U.S. dollars) Expected Deaths disease- Expected averted in specific Proportion of ARI-specific children Cost per Cost per mortality ARt mortalit-.' mortality under five person in Total cost Cost per disability- reductiona addressed reduction (per million target (per million death adjusted life- Intervention (percent) (percent) (percent) population) population population) averted year saved Case management 60-90 38-52 23-47 351-676 $3.61 $220,000- $379- $37 (80) (49) (39) (585) $940,000 $1,610 ($541,877) ($926) Breastfeeding 50-80 4 2-3.2 15-96 $5.00 $40.00 $417- $38 promotion (72) (2.8) (42) $2,667 ($952) EPJ vaccines 44-80 20-25 8.8-20 66-600 $9.08 $122,580- $409- $40 (65) (22.5) (14.6) (219) $245,160 $1,857 ($217,920) ($995) Reduction of 50-95 70-90 35-85 263-2,550 $15.00 $810,000- $697- $63 malnutrition (80) (80) (64) (960) (malnourished) $1,777,500 $3,080 $11.85 ($1,500,000) ($1,563) (all children) Pneumococcal 0-30 30-50 0-15 0-450 $7.28 $98,280- $437 $67 vaccine (15) (40) (7) (105) $196,560 ($1,664) ($174,720) Note: Mlost likely values in parentheses. a. Disease is pneumonia except for EPI vaccine, where disease is percussis and measles; for pneumococcal vaccine, disease is pneumococcal pneumonia. Source: Authors data. In 1989, the World Health Organization recommended the PERTUSSIS IMMUNIZAT[ON. The vaccine for pertussis is deliv- use of the higher titer Edmonsten-Zagreb vaccine at or before ered together with the diphtheria and tetanus vaccines. The six months of age in areas where measles is a major cause of efficacy of pertussis vaccine for the fully immunized child infant mortality (WHOIEPI 1990). Subsequent reports of in- (three doses) has been recently questioned but is estimated at creased late mortality among children immunized with such 70 to 90 percent in the industrialized world (Church 1979; higher titer vaccines (Garenne, Leroy, and Sene 1991) has, Koplan and others 1979) and 50 to 90 percent in developing however, prompted a suspension of that recommendation. countries. As for measles, however, transmission rates in Altematives to the current measles vaccines must be devel- endemic areas are such that many children are infected oped which are effective in younger infants. and most deaths occur prior to the usual age of completed Ongoing studies of the effectiveness, optimal dose, non- immunization. parenteral routes of administration (in order to overcome Pertussis immunization coverage in some industrialized residual matemal antibodies), and booster response to the new countries has fallen off in the last fifteen years, primarily vaccines will help further to refine immunization policies in because of concern about associated adverse neurological ef- the near future. More studies are needed to explore the cost- fects, most notably encephalopathy (Brahmans 1986), al- effectiveness of a two-dose schedule, such as initial measles though there is also some controversy about the vaccine's immunization with the third diphtheria-pertussis-tetanus vac- effectiveness (Fine and Clarkson 1987). Outbreaks of pertussis cine (DPT) dose (followed by a second dose at six to twelve have been observed in Great Britain, Japan, and Sweden, months of age). High drop-out rates, which are a major barrier where policy changes or public opinion have led to a reduction to the success of immunization programs, also argue in favor of of immunization coverage. Even with the current preparation, using earlier opportunities for immunization. even if children however, the benefit of the vaccine far outweighs the risk of are less than the ideal age for achieving seroconversion or adverse effects (Koplan and others 1979; Cherry 1984). Accel- optimum protection. Some investigators, however, have raised erated research has led to the development of acellular pertus- the concern that earlier immunization may interfere with anti- sis vaccines, which offer hope in the near future for both body response at the time of revaccination (Wilkins and Wehrle improved effectiveness and fewer adverse effects (Miller and 1979; Linnemann and others 1982; Stetler and others 1986). others 1991). Acute Respiratory Infection 75 PNEUMOCOCCAL IMMUNIZATION. The pneumococcal vaccine The newer Hib conjugate vaccines, which link Hib anti- licensedforuse intheUnitedStatesiscomposedofthepurified gens to protein carriers, show improved immunogenicity in polysaccharide extracted from twenty-three of the eighty-four children under two and hold greater promise for preventing types of Streptococcus pneumoniae. These capsular subtypes are H. influenzae disease in the very young. Although experience responsible for approximately 90 percent of invasive pneumo- with use of diphtheria toxoid as a conjugate has been mixed, coccal disease in the United States. Still, over 30 percent of tetanus toxoid carriers may be more effective (Eskola and blood culture isolates from patients with pneumonia in devel- others 1990; Siber and others 1990; Ward and others 1990; oping countries have been pneumococcal serotypes which are Wanger and others 1991). Formulations for developing coun- not included in the current vaccine (Ghafoor and others 1990; tries will need to include additional types (that is, non-b and Mastro and others 1991). In addition, the vaccine induces nonserotypableH. influenzae) whicharenotaprominentcause little immunity in children under eighteen months of age, who of invasive disease in the industrialized world (Funkhouser, are most vulnerable to mortalitydue to pneumococcal infections. Steinhoff, and Ward 1991). Recent studies in Papua New Studies in the United States have suggested that the cur- Guinea (Weinberg and others 1990), Pakistan (Ghafoor and rently available vaccine is 50 to 80 percent effective in pre- others 1990), and the Gambia have shown that approximately venting bacteremia and pneumonia in adults. Results of studies half of all invasive H. influenzae disease is due to nonserotyp- among the very elderly or chronically ill (Simberkoff and able or non-b strains. others 1986) and among children under eighteen months have The costs of the current conventional Hib vaccine is $2.19 been less encouraging. The vaccine has also been tested in per dose, while the conjugate vaccine is $14.00 per dose. Hay Papua New Guinea, which reports that up to 50 percent of and Daum (1987) compared the costs and benefits of rifampin pneumonia is due to pneumococcal infection. Clinical trials prophylaxis of exposed contacts to immunization with the there (Riley and others 1986) among children age six months currently available unconjugated vaccine. Vaccination was through fifty-nine months have shown a 50 percent reduction predicted to be the most cost-effective strategy with a calcu- in pneumonia-specific mortality rates during periods of one to lated overall net savings of $64.8 million, in the setting of an five years after immunization. There appears to have been no anticipated social cost of$ 1.94 billion for H. influenzae disease reduction in pneumonia incidence, and there is little evidence in the 1984 birth cohort. Because of the paucity of data on Hib to suggest that the vaccine was immunogenic in the younger vaccine effectiveness in developing countries, no estimates of age groups. The cause of the mortality reduction is, therefore, cost-effectiveness have been included in table 4-3. not clear, and the results need to be replicated in other devel- oping countries. Also of potential interest was the finding that OTHER IMMUNIZATIONS. Vaccination to induce immunity to infants of mothers immunized during their last trimester had a organisms which cause ARI mortality is clearly an effective 32 percent lower rate of pneumonia (Riley and Douglas 1981). preventive intervention. There is good evidence to support the Such "passive" protection of infants while they await comple- use of vaccines against measles and pertussis, both in the tion of immunization series deserves further investigation. documented importance to public health of these problems and the effectiveness of immunization. Although the expected H. INFLUENZAE VACCINE. Like pneumococcal vaccine, which effect of diphtheria vaccine is difficult to predict because of the is also a polysaccharide vaccine, the current Hl. influenzae lack of information regarding diphtheria morbidity and mor- vaccine has limited immunogenicity in infants and young tality, marginal costs of including the vaccine with pertussis children. The vaccine is made from H. influenzae type b and tetanus (in tPT vaccine) are nearly negligible. polysaccharide, since this type accounts for virtually all inva- Effective vaccines against the viral causes of pneumonia and sive disease in the industrialized world. The effective protec- bronchiolitis would likely avert additional mortality. Attempts tion (measured by the prevention of invasive disease, mainly to develop effective vaccines against the two most important meningitis and bacteremia) found in children over twenty-four causes of mortality, RSV and parainfluenza viruses, however, months of age has ranged from 0 to 90 percent (Granoff and have been frustrating. Still, the mechanism for the adverse others 1986; Harrison and others 1987; Black and others 1988; hypersensitivity responses to RSV antigens has recently been Gilsdorf 1988). The effectiveness of the H. influenzae type b identified, so that purified antigen and recombinant vaccines vaccine in reducing pneumonia morbidity and mortality can- currently under development should offer greater hope for not be estimated from U.S. data, since the frequency of pneu- these important causes of respiratory mortality (Pringle 1987). monia due to Hib is too low. Influenza vaccines have been effective in preventing infec- An increase in early cases after Hib immunization has been tions, particularly among the elderly, but the "antigenic drift," variously ascribed to unmasking of latent infection or shorten- or frequent changes in surface proteins which characterize ing of the incubation period, perhaps because ot transient these viruses make vaccine production and distribution more postvaccination reductions in antibody levels (Black and oth- costly. Such immunization programs for adults have also been ers 1988). One study (Osterholm and others 1987) actu- poorly received and achieve limited coverage. ally calculated an increased risk of H. influenzae infection of 45 percent, leaving the protective effect of Hib vaccine in ENVIRONMENTAL AND NUTRITIONAL RISK REDUCTION. Alter- some doubt. ation of other documented and suspected risk factors for ARI 76 Saly K. Stansfield and Donald S. Shepard mortality, such as poor nutritional status (including low birth primary health care system, avoiding duplication of necessary weight, poor infant-feeding practice, undernutrition, and vita- management, supervision, training, and logistical resources. min A deficiency) and exposure to smoke (including smoke Immunization campaigns, although they result in high short- from active and passive cigarette smoking and from organic- term coverage, may compromise sustainability and divert re- fuel cookfires), have been suggested as additional strategies for sources from the development of the rest of the primary health the prevention of ARI deaths. Although the association of some care infrastructure. of these risk factors with disease is strong, few studies support Global coverage estimates for children immunized during the feasibility of programs using these interventions or their the first year of life are 50 percent for measles and 55 percent effectiveness in preventing ARI. The data show as far more for the DPr series of three doses (Keja and others 1988), feasible and effective the promotion of breastfeeding and re- although figures are considerably lower in Africa. Barriers to duction of malnutrition. achieving improved coverage with the EPI vaccines include difficulties with supply and management systems and the prac- Good Practice and Actual Practice: Are There Gaps? tical problems of maintaining the cold chain. High drop-out rates for immunization series are partly the result of limitations Correct case management is the central strategy of WHO's of resources for social mobilization, the opportunity costs Programme for the Control of Acute Respiratory Infections. to the family in obtaining immunizations, failure of health One of the four objectives, however, is "to reduce the inci- workers to profit from clinic visits by giving immunizations dence of acute lower respiratory infection" (WHo/ARI 1991). (Keja and others 1988), and adverse effects of current vaccine Although intervention to alter some of the nonspecific risk preparations. factors for ARI (table 4-2) is an intriguingpossibi!ity for preven- Increasing attention, especially in Africa, has been paid to tion of pneumonia, immunization remains the only strategy the reuse of syringes and needles in immunizations. These known to be effective in the prevention of morbidity and unsafe immunization practices introduce the risk of transmis- mortality due to pneumonia. sion of blood-borne diseases such as hepatitis and AIDS. Even this proven technology, however, has not been fully Efforts to prevent ARI mortality through reduction of the exploited to prevent ARI mortality. As of 1988, many of the 97 prevalence of malnutrition and low birth weight are hampered countries with an Expanded Programme on Immunization still by the obvious social, economic, and political barriers to devel- had subnational coverage, often neglecting the neediest chil- opment. Promotion of appropriate infant-feeding practice, in- dren in the most remote areas (Keja and others 1988). Vacci- cluding breastfeeding, represents an opportunity to reduce ARI nation efforts are most appropriately integrated with the morbidity aid mortality that deserves greater emphasis. Table 4-4. Diagnosis and Treatment of Pneumonia in Children Aged Two Months to Five Years Disease Signs Treatment Very severe disease Unable to drink Refer urgently to hospital Convulsions Give first does of antibiotic Abnormally sleepy or hard to wake Treat fever, if present Stridor in calm child Treat wheezing, if present Severe undemutrition If cerebral malaria possible, give antimalarial drug Severe pneumonia Chest indrawing Refer urgently to hospital' Give first dose of antibiotic Treat fever, wheezing if present Pneumonia No chest indrawing Advise parent for home care Fast breathingb Give antibiotic Treat fever, wheezing if present Reassess in two days; if child getting worse (unable to drink, chest indrawing, other danger signs), refer urgently to hospital; if child the same, change antibiotic or refer; if child improving (breathing slower, less fever, eating better), finish five days of antibiotic No pneumonia: cough or cold No chest indrawing If coughing more than thirtv days, refer for assessment No fast breathing Assess and treat ear problem or sore throat, if present Advise parent for home care Treat fever, wheezing, if present a. If referral not feasible, treat with antibiotic and follow closely. b. Fast breathing defined as fifty breaths per minute or more in infant age two to twelve months, forty breaths per minute or more in childi age one tn five years. Source: WHO/ARI 199lb. Acute Respiratory Infection 77 Table 4-5. Diagnosis and Treatment of Pneumonia in Infants Less than Two Months Old Disease Signs Treatment Very severe disease Not feeding well Refer urgently to hospital Convulsions Keep infant warm Abnormally sleepy or hard to wake Give first dose of antibiotic Stridor in calm child Wheezing Fever or low body temperature Severe pneumonia Severe chest indrawing Refer urgently to hospiralh Fast breathing' Keep infant warm Give first dose of antibiotic No pneumonia: cough or cold No severe chest indrawing Advise parent to give following home care: No fast breathing Keep infant warm Breastfeed frequently Clear nose if it interferes with feeding Return quickly if breathing becomes fast or difficult, feeding becomes a problem, or infant becomes sicker a. Fast breathing defined as sixty breaths per minute or more. b. If referral not feasible, treat with antibiotic and foilow closely. SOUTLCe: Autliors' data. Case Management Chest indrawing (retraction of the lower part of the chest wall on inspiration) detected in children two months to four years Altlhough research must continue to improve preventive tech- of age indicates the presence of severe pneumonia requiring nologies for the primary causes of ARI mortality, ARI control hospitalization. The algorithms used for diagnosis and treat- programs for the near future will rely principally on improved ment of childhood pneumonia, including additional criteria case management. for referral for hospitalization, are summarized in tables 4-4 and 4-5 (WHO/ARI 1991Ib). Any of four inexpensive antibiotics may Elements of the Case Management Strategy be recommended for the home care of uncomplicated pneumo- nia, including co-trimoxazole (trimethoprim-sulfamethoxazole), The World Health Organization's ARI control program has amoxycillin, ampicillin, and procaine penicillin. taken the lead in promoting intervention to address the prob- Although wheezing (including asthma) is managed within lem of ARI in children. The primary objective of the program this algorithm, there are separate guidelines for care of sore is the reduction of ALRI mortality through effective case man- throats and ear infections. All cases receive general supportive agement. Secondary objectives include the reduction of (a)the care, including fluids, continued feeding, treatment of fever, severity and complications of acute upper respiratory tract and clearing of nasal or ear discharge as needed. Although infections, (b) the inappropriate use of antibiotics and other these findings require further confirmation, studies in Pakistan drugs for the treatnent of ARI, and (c) the incidence of pneumo- have suggested that such supportive measures may actually nia. To improve the case management of pneumonia, WHO has reduce the likelihood of progression of uncomplicated coughs developed guidelines for standard treatment at the most periph- and colds to life-threatening pneumonias (Khan, Addiss, and eral and referral health facilities and at the community level. Rizwan-Ullah 1990). In countries with a high incidence of bacterial pneumonia There is no doubt about the importance of bacterial pneu- (generally those with an infant mortality rate greater than 40 monia as a cause of mortality or about the effectiveness of per 1,000), pneumonia may be relatively reliably diagnosed on antimicrobials in reducing case-fatality ratios. But to address the basis of simple clinical criteria alone. For any child under concerns about whether peripheral health care workers with five with cough ordifficult breathing, tachypnea (rapid breath- limited training could identify and treat cases appropriately, ing) appears to be the best single predictor of pneumonia and several intervention studies were conducted to test the algo- the need for antibiotic treatment (Shann, Hart, and Thomas rithm for case management in an operational setting in several 1982; Campbell, Byass, and Greenwood 1988; Cherian and developing countries. These and another study conducted in others 1988; Campbell, Armstrong, and Byass 1989; Campbell Jumla, Nepal, were recently reviewed (WHO/ARI 1988). Al- and others 1989; Lucero and others 1990). In view of the though each of the studies suffered from design flaws or con- variation of normal respiratory rates with age, 'wio guidelines founding as a result of simultaneous introduction of other recommend a threshold rate of sixty or more per minute in interventions, taken as a whole they present strong evidence young infants (under two months), fifty or more for older of the effectiveness of case management by peripheral health infants (two months through eleven months), and forty or care workers. It was found that ARI-specific mortality declined more for children one through four years of age (WHO 1990). by an average of 41.6 percent (range 18-65 percent), whereas 78 Sally K. Stansfield and Donald S. Shepard overall mortality was reduced in the same five study areas by to successful implementation of an ARI control program. Ap- an average of 22.2 percent (range 11.5 to 40 percent). These propriate case management requires that eachofmany difficult studies, for which further details are presented in table 4-6, conditions be met, including the design and communication confirmed the feasibility and efficacy of providing case man- of culturally appropriate and effective health education for agement of pneumonia through peripheral health workers with family recognition of suspected pneumonia, prompt presenta- limited training. tion to an effectively trained and carefully supervised health These results compare favorably with earlier, more theoret- worker, correct diagnosis and selection of treatment, develop- ical calculations of the expected effectiveness of pneumonia ment and maintenance of reliable logistical systems to ensure case management interventions. For example, Tugwell and adequate supplies of antibiotics, family compliance with ap- others (1985) assumed an efficacy of co-trimoxazole in treat- propriate instructionsforcare, and access tocompetent referral ment of community-acquired pneumonia of 80 percent, a care as required. diagnostic accuracy of 80 percent by the health workers, a These prerequisites for effective case management of pneu- correct treatment rate of 90 percent, 80 percent patient com- monia are inextricably linked to the basic infrastructure for pliance with the medication regimen, and 80 percent access to primary health care. Although ARI control may be introduced appropriate treatment, calculating an expected program effec- as another "vertical" program, it is less conveniently addressed tiveness of 37 percent. outside the context of the health care delivery system as a Lessons learned from the case management intervention whole. Strengthening of systems to reduce pneumonia mortal- trials must be taken into account in program design and ity therefore requires a more comprehensive approach to im- selection of research priorities. For example, the Jumla study proving access and quality of care. documented a mean duration of fatal episodes of pneumonia It will be important, for example, t) rarionalize the use of of three and one-half days (Nils Daulaire, personal communi- antibiotics for other health problems in order to ensure that cation, 1990). Under these circumstances, active surveillance adequate supplies remain to treat cases of pneumonia. Even by health workers is unlikely to detect an adequate proportion when basic antibiotics are unavailable in peripheral health of cases. Control programs for ARj must rely on families to centers, the presence of antibiotics in remote markets provides detect signs and symptoms of pneumonia and bring suspected evidence of the effectiveness of informal systems of distribu- cases to a health worker for evaluation and treatment. Reduc- tion. Such sales of antimicrobials in the informal sector likely tions in deaths due to diarrhea were observed in Jumla, where leads to their inappropriate use in even more than the 50 to 95 only pneumonia cases were treated (WHO/ARI 1988), raising the percent of cases in which inappropriate use is observed in important question of the effect of antibiotic treatment on health centers (Chaulet and Khaled 1982; Hossain, Glass, and concurrent infections such as diarrhea or malaria. Khan 1982; Gutierre: and others 1986). Reduction of the Few operational programs for pneumonia case management inappropriate use of these supplies may actually avert pneumo- have measured cost per case treated or death averted. The nia deaths at no increased cost, both through increasing effec- figures which are available have been obtained in research tive use and reducing adverse effects and the evolution of settings, where expenditure may not be representative. Costs antimicrobial resistance (Stansfield 1990). Although the fea- per case treated in the Philippines have been estimated at sibility of labeling antibiotics in special packages (as solely for $5.15 and $4.37 (Brenzel 1990). Costs per death averted have use in the treatment of pneumonia in children) is being as- ranged from $200 in Indonesia to $350 in Nepal (Brenzel sessed (wHo 1990), the inevitable discovery of the alternative 1990). Using a model to estimate cost-effectiveness outlined uses of these powerful pharmaceuticals will likely render such in the appendix to this chapter, we have calculated an ex- practices ineffective. pected cost per death averted of $926, and a cost per dis- Another obstacle to be anticipated is the resistance of counted healthy year of life saved of $37. physicians to empowering other health care workers with training to diagnose and treat with antibiotics. Narain and Good Practice and Actual Practice: Are There Gaps? Sharma (1987), for example. have presented evidence that over 90 percent of physicians do not agree that nonphysician By the end of 1990, fifty-four countries had prepared plans of health workers should be provided with antibiotics to treat operation for ARI control programs and forty-seven had func- children suffering from pneumonia. Vigilance will also be tioning programs (WHO/ARI, 1991a). Eighteen additional coun- required to prevent commercial drug companies from exploit- tries had designated a national program manager and issued ing new markets by extracting inflated prices for basic pharma- technical guidelines for case management. Therefore, a total ceutical supplies. of fifty-nine countries, most of which are in the Americas and Many countries will require assistance in the development Western Pacific, had taken some steps to establish a national of laboratory capability to ensure the correct selection of ARI control program. antibiotics (at least in reference centers), particularly for refer- Yet it is clear that the intrinsic complexity of the manage- ral patients who have failed treatment with first-line antibiot- ment of ARI will present great challenges in the implementa- ics, through basic bacterial cultures and tests for antibiotic tion of control programs. The significant operational problems sensitivity. These capabilities are also required to maintain the encountered in immunization and diarrheal disease control necessary surveillance for the emergence of significant antibi- programs, for example, are likely to be dwarfed by the obstacles otic resistance pattems, as is evidenced by the alarming resis- Table 4-6. Case Management of ARi in Children: Summary of Intervention Studies Baseline data Measles immunization Case detection Pneumonia treatment MortalitY reduction IMR a coverage Maternal First-line Referral ALRI -specific Overall LocatioTi (dates) Study design (per 1,000) (percent) Case-finding education Source antimicrobial care (percenlt) (percent) Haryana, India Concurrent control; low 210-275 0 Active Yes Community health Penicillin (oral) None 42 24 (1982-84) birth weight only (weekly) worker Abbottabad, Pakistan Concurrent control, 90-100 5.4 Active (every Yes Community health Co-trimoxazole Poor 56 55 (1985-87) subsequient intervention 10-14 days) worker or clinic access in control area Bohol, Philippines Concurrent control 49-63 58-60 Passive No Clinic Co-trimoxazole Yes 25 13 (1984-87) Bagamoyo, Tanzania Concurrent control; 137 53 Passive No Community Co-trimoxazole Yes 30 27 (1985-87) subsequent intervention in health worker control area or clinic Kathmandu, Nepal Before and after 162 11 Active (every Yes Community health Ampicillin Poor 62 40 (1984-87) 2 weeks) worker utilization Kediri, Indonesia Before and after 154 1.5 Active (every Yes Community health Co-trimoxazole Poor 67 41 (1986-87) 2 weeks) worker access a. Infant mortality rate. Source. WHO/ARI/88.2 aind WHO/AR1191.20. 80 Salk, K. Stansfield and Donald S. Shepard tance to commonly used antimicrobials in several countries though most countries have active EPI programs, these pro- (El-Mouza and others 1988; Lataorre Otin, Juncosa Morros, grams must also be strengthened to ensure improved coverage and Sanfeliu Sala 1988; Mastro and others 1991). (Poore 1988). Donor agencies must recognize these gaps when allocating Referral care for pneumonia in persons who have responded resources for ARI control program development. Although the poorly to antibiotic treatment requires adequate laboratory historical lack of donor support in this area has also been an support to obtain bacteriologic cultures, identify organisms, important obstacle to ARI control, the donor community has and determine antibiotic sensitivities. Any national ARI con- recently shown increased interest in strengthening pneumonia trol program must, therefore, allocate adequate resources to case management. Reduction in pneumonia deaths by 25 monitor antibiotic resistance competently in at least one na- percent was included among targets established for the 1990s tional reference center. Ability to conduct vaccine trials will by the World Health Organization and United Nations also depend on laboratory capability in the identification of Children's Fund (UNICEF) Joint Committee on Health Policy. specific serotypes for the main pathogens. This commitment is only beginning to be reflected in in- creased levels of national, bilateral, and multilateral funding Research Priorities to ARI control programs. During program design and implementation, high priority Priorities for ARI Control should also be assigned to establishing a strong evaluation or applied research component to aid in assessing the effective- The many national health plans which emphasize the priority ness of operational programs, refining program priorities, and of interventions to reduce infant and child mortality cannot addressing the many questions which remain regarding opti- long ignore pneumonia which is often a primary cause of this mal strategies for prevention and case management. The mortality. Global commitment to addressing this problem World Health Organization recently reviewed research prior- was reflected in the adoption, at the World Summit for Chil- ities for ARI control (WHO/AR] 1989a, 1990), preparing a list dren in September 1990, of a resolution to reduce deaths due which we have adapted and present below. to ARI by one-third during the final decade of this century (UNICEF 1991). * Assess the effectiveness of interventions and programs to modify the risk of pneumonia, especially through reduc- Priorities for Resource Allocation tion of exposure to biomass fuel emissions. * Document the epidemiology of invasive strains of In view of the effectiveness of the vaccines and of antibiotic Hemophilus influenzae and Streptococcus pneumoniae to guide therapy for pneumonia, it is probable that more than half of the development of effective vaccines. ARI deaths could be averted through use of only the currently Definetherelativeprevalenceandetiologiesofpneumo- available technologies of immunization and improved case nta, sepsis, and meningitis in less immunocompetent groups management. Breastfeeding promotion and reduction of the such as young infants (less than three months of age) and prevalence of malnutrition are also likely to be cost-effective undemourished children. in reducing mortality due to ARI. Interventions for the promo- * Identifythesignsandsymptomswhichindicatetheneed tion of breastfeeding, reduction of malnutrition, and im- * hosptl*cae. munization with EPI vaccines will have a broader effect on child for hospital care. survival through their effectiveness in prevention of mortality * Evaluate the performance of the treatment protocols, due to diseases other than ARI. These three interventions, including for wheezing, at first-level referral facilities. along with appropriate case management, should be given high * Explore the most effective ways to define and teach the priority for implementation, particularly in countries with high reliable distinction between clinical presentations of pneu- infant mortality. Such a combined curative-preventive ap- monia and malaria and to determine the effectiveness of proach is likely to be the most effective as a strategy to reduce co-trimoxazole in treatment of malaria. mortality (Mosley and Becker 1988). * Define the clinical features and optimal treatment of National ARI control programs should be developed or ac- serious bacterial infections (pneumonia, sepsis, and menin- celerated according to the guidelines recently refined by WHO's gitis) in young infants (less than three months of age). Programme for the Control of Acute Respiratory Infections. * Determine the special needs of undernourished children, Intervention studies have provided adequate evidence among children under five that improved case management of pneu- iningad the clinical featuesa causene monia will reduce mortality due to ARI and, possibly, overall monia and the optimal treatment for these children. mortality in that population. As WHO has pointed out (WHO/AR * Examine cultural and other factors which determine the 1988), "there is no technical justification in delaying any ability of families to recognize signs of pneumonia, seek further the expansion of ARI control programmes as an essential appropriate care, and comply with treatment regimens. component of child survival efforts, and with the same priority * Identify optimally effective strategies for the design of attached to the Expanded Programmes on Immunization (EPI) appropriate health education programs, including strategies and the diarrheal disease control (CDD) programmes." Al- for the modification of risk factors for pneumonia. Acute Respiratory Infection 81 Develop inexpensive, simple, and reliable diagnostic children under five. Estimates of cost and effectiveness for technologies to aid in counting respiratory rate and deter- immunization interventions are made using a coverage range mining the etiology of pneumonia, such as by identifying of 45 to 90 percent, as used by Feachem and Koblinsky (1983). viral or bacterial antigens in urine or blood. "Most likely" values for immunization interventions are calcu- * Perform field trials of the available polysaccharide pneu- lated assuming an immunization coverage of 80 percent, the mococcal vaccine and conjugate vaccines for H. influenzae target specified for UNICEF's goal of universal childhood im- type b and for nonserotypable H. influenzae, RSV, and para- munization. Calculations of deaths averted are based on pre- influenza viruses, when available. intervention ARI-specific mortality rates of 5 to 20 per 1,000 (with a most likely value of 10 per 1,000), which yields an Additional research needs that must be addressed include expected 750 to 3,000 (most likely 1,500) deaths among the the development and validation of survey techniques for de- 150,000 children under five in the standard population of I tection of ARI episodes and pneunmonia deaths for use in pro- million. gram evaluation. Studies are also needed to determine the The effect of implementing multiple interventions to pre- effect of antibiotic use on the incidence of and mortality from vent ARI mortality is unlikely to be simply additive. Many other diseases (especially malaria and diarrhea) and the socio- children who die with ARI suffer from several risk factors, such cultural factors which modulate the effectiveness of programs. as the malnourished child who dies of pneumococcal pneumo- Vaccine research issues, in addition to those detailed in the list nia during or within one month of an episode of measles. One above, should include additional efficacy trials of the newer possibility is that such competing risks ofmortality may operate measles vaccines and two-dose schedules of administration. on the same children, so that prevention of one potentially Another issue for operational research will be the effect of mortal event may only leave children vulnerable to other current program emphasis on children under five. Promoting causes of so-called "replacement mortality" (WHO/EPI 1987). the recognition of the need and increasing the demand for Another possibility is that prevention of ARI is actually health services will be essential to the success of ARI control synergistic with other preventive interventions through reduc- programs. Although the opportunity tohave an effect isgreat- ing the cumulative contributions to the frailty (Mosley 1985) est among the program's target group, it will be important to of the child, such as is observed in the growth faltering that assess the benefits and costs to national programs which at- occtirs with recurrent infection. An example of the potential tempt to reserve the attention of health workers and the for synergism among interventions has been suggested by re- supplies of antibiotics for children at the perceived expense ot cent observations of a reduction of mortality from diarrheal the communities' adult decisionmakers and opinion leaders. disease in a program which treats only childhood pneumonias Opportunities to explore mechanisms to achieve financial (WHO 1988). sustainability may be limited for the immunization interven- The calculations presented below consider only the short- tions designed to prevent ARI. For the curative care provided term effects of these interventions upon mortality. It is not in the case management of AR[, however, it will be important known whether the long-term effect of these preventive and to explore mechanisms for cost recovery, such as health insur- curative interventions would be augmented by reduction in ance schemes, taxation, and user fees, to increase the financial the frailty of these children, or offset by replacement mortal- sustainability of national programs. ity. Because of these theoretical problems and the many operational problems associated with predicting the effects of multiple health interventions, the figures presented in Appendix 4A. Sources of Data and Method Used table 4-3 are of use primarily as estimates of the relative cost- to Obtain Cost-Effectiveness Estimates Summarized effectiveness of interventions when implemented alone to in Table 4-3 reduce AR! mortality. Since no prospective data are available for the effectiveness of Expanded Programme on Immunization preventive interventions in the reduction of AR! mortality among children under five, the estimates used in table 4-3 are Most of the ARI mortality prevented by the EPI vaccines is due obtained primarily from retrospective observations of relative to measles and pertussis. Feachem and Koblinsky (1983) esti- risk. These figures, therefore, represent indirect estimates of mate that measles immunization between the ages of nine the potential effectiveness of the intervention rather than a months and twelve months, with an ideal effectiveness of 90 measure of effectiveness achieved in an operational setting. percent and a coverage ofbetween 45 and 90 percent, can avert Cost estimates for these preventive interventions are also 44 to 64 percent of measles cases. In anticipation of the obtained indirectly, through review of data for similar pro- improved effectiveness of the higher-potency vaccines and grams. Cost estimates for case management interventions are immunization before nine months ofage, the upper limit ofthe similarly derived from data available from other programs with proportion of cases averted was adjusted to 80 percent Since similar interventions. pertussis vaccine has a similar ideal effectiveness, it is assumed Estimates assume a standard population of 1 million persons, that a similar proportion of cases would be averted at the same with 15 percent of the population being children under five coverage rates of 45 to 90 percent. Therefore, for pertussis and (approximately 3 percent infants) and 8 percent mothers of measles, an effectiveness range of 44 to 80 percent has been 82 Sally K. Stansfield and Donald S. Shepard used for the model, with an intermediate most likely value of coccal disease in that country is probably not typical of that in 65 percent. most developing countries. On the basis of the 750 to 3,000 It has been estimated that up to 2,596 of ARI mortality may expected deaths among children under five, the number of be preventable if current EPI vaccines are used. Mortality deaths averted may be calculated to be from 0 to 450, with a among children under five due to measles-associated ARI ac- most likely value of 105. counted for approximately 20 percent of all ARI mortality (1.8 The estimated cost per child vaccinated is calculated by per 1,000 out of 9. I per 1,000) in seventeen study areas during reducing the current price of the vaccine ($9.69 per dose) by case management trials for WHO (WHO/ARI 1988). It is therefore one-half (assuming that the cost will be reduced for the inter- assumed that 20 to 25 percent of ARI mortality would be national market in exchange for waiver of liability and once addressed through use of current EPI vaccines. An expected research and development costs are recovered) and adding the A.MR-specific mortality reduction of 8.8 to 20 percent (most cost per dose delivered for the Er' vaccines ($2.44 average), likely 14.6 percent) may be calculated from these figures. since the costs of EPI vaccines ($0.04-$0. 15 per dose) are small These estimates are comparable to the .ARn mortality reduction in relation to the cost of the pneumococcal vaccine. The figures of 5 to 20 percent calculated by Singhi and Singhi resulting estimated cost per dose delivered of $7.28 suggests (1987), although it was observed that measles vaccine pro- that (at coverage levels of 45 to 90 percent) the cost per death vided an effective protection of 22 percent against respiratory averted would be greater than $437, with a most likely value deaths in Bangladesh. Based on the expected number of deaths of $1,664. of 750 to 3,000, the number of deaths averted may be calcu- lated to range from 66 to 600, with a most likely value of 219. Breastfeeding Promotion The cost per child served for EPI immunization interventions to prevent ARI mortal ity is calculated as that portion of the cost Reductions in incidence of and case-fatality ratios for pneumo- of delivering all EPI vaccines, which is proportional to the nia that have been noted with breastfeeding (Chandra 1979; benefit achieved in averting ARI deaths. Tetanus is the only LePage, Munyakazi, and Hennart 1981; Victora and others non-ARI Eri disease which is a significant cause of infant and 1987) suggest that a 50 to 80 percent ARI-specific mortality child mortality. Since tetanus accounts for up to 40 percent of reduction might be realized among breastfed infants. The the overall mortality prevented through EPI vaccines, 60 per- protective effect is observed, however, only below twelve cent ($9.08) of the $15.13 average cost per fully immunized months of age (approximately 20 percent of children under child (Brenzel 1989) was ascribed to ARI prevention. The cost five), and actual prevalence of breastfeeding among intants is per ARI death averted (achieving coverage levels of 45 to 90 generally over 80 percent in high-mortality countries, such percent among the 30,000 infants in the target age group) may, that only about 4 percent of children under five would benefit therefore, be calculated at $409 to $1,857, with a most likely from a program to promote breastfeeding. Even assuming 100 value of $995. percent effectiveness in changing breastfeeding practice, the reduction of ARI-specific mortality among children under five Pneumococcal Immunization resulting from such promotion would be only 2 to 3.2 percent (with, on the basis of Victora's observation of over 70 percent The effectiveness of pneumococcal vaccine, particularly reduction in relative risk, a most likely value of 2.8 percent). among the youngest children, has not been clearly demon- The end result would be an estimated fifteen to ninety-six strated in the developing world. The reported effective range (most likely forty-two) deaths averted through promotion of in adults of 0 to 80 percent and the effectiveness of 69 percent breastfeeding. noted among children in the United States suggest a range of The average cost of a program to promote breastfeeding has 0 to 70 percent. Still, because the vaccine is less immunogenic been estimated at $5 (Feachem and Koblinsky 1984; Feachem among children under two (who constitute approximately 40 1986; Phillips, Feachem, and Mills 1987) per mother. Even if percent of children under five), an estimated 20 percent of targeting of services is only adequate to identit6 the subset ot- invasive, pneumococcal infections in developing counitries are 50 percent of the mothers who are "at risk" of not breast- caused by serotypes not included in the vaccine, and assuming feeding, the population served might be reduced to half of the a 45 to 90 percent coverage (as for the EPI vaccines), the mothers with infants (8,000 in the standard population of I expected disease-specific mortality reduction may be in the million). The estimated cost per ARI death averted for an range of 0 to 30 percent. An intermediate most likely value of educational program to promote breastfeeding may be calcu- 15 percent has been selected. lated to be $417 to $2,667, with a most likely value of $952. Since pneumococcal disease accounts for less than one-third to one-half of pneumonia cases (WHOtRSD 1986), the maximal Reduction of Malnutrition reduction in pneumonia mortality with the presently available vaccine is likely about 0 to 15 percent, with a most likely value Expected mortality reduction with improved nutritional status of 7 percent. Although this range does not include the greater was estimated on the basis of mortality two to twenty times reductions in ARI-specific mortality observed among children higher observed in malnourished children (Kielmann and six months through fifty-nine months of age in Papua New McCord 1978; Tupasi and others 1988). Successful improve- Guinea (Riley and others 1986), the epidemiology of pneumo- ment of nutritional status might be expected to result in a 50 Acute Respiratory Infection 83 to 95 percent reduction in the risk of ARI-mortality among $1,777,500, to serve all 150,000 children expected in the malnourished children. A most likely value of 80 percent sample population. The use of the most likely value of reflects the modest estimate of a fivefold higher relative risk of $1,500,000 reflects the better credibility of the figures from pneumonia deaths among these malnourished children. Since Tanzania. Final evaluation of the Joint Nutrition Support 70 to 90 percent of all pneumonia deaths occur among the Program may yield costs per child as low as $2.50 per year malnourished, expected ARI-specific mortality reductions of 35 (UNICEF, personal communication, June 1991). On the basis of to 85 percent (most likely 64 percent) might be expected the less favorable preliminary figures, however, the cost per with successful improvement of nutritional status. Support death averted is calculated to be $697 to $3,080, with a most for these estimates is provided by the results of a nutritional likely value of $1,563. intervention program in Tanzania (UNICEF 1988b), where a 23 percent reduction (from 48 to 37 percent) in the preva- Case Mancagement lence of mild to moderate malnutrition (less than 80 percent weight-for-age) and a 60 percent reduction (from 5 to 2 per- There is little information available to date regarding the cost cent) in the prevalence of severe malnutrition (less than 60 of operational programs for ARI case management for which percent weight-for-age) were associated with a 64 percent effectiveness has also been assessed. Cost per child treated and reduction in ARI-specific mortality. The expected number of death averted may, however, be estimated from drug costs and ARI deaths averted at this level of effectiveness would be costs for implementing other programs with similar interven- 263 to 2,550, with a most likely value of 960, although these tions. The following model was constructed to provide an figures would be highly dependent on the initial prevalence of estimate of the cost of case management for cost-effectiveness undemutrition. calculations. The lower estimate for the cost of such a program to improve The cost per million population of appropriate case manage- nutritional status is based on expenditure of $15.00 per year ment for AR] is equal to the sum of the costs of the following: per malnourished child under five (Ashworth and Feachem * Health education or sensitization regarding program in- 1986), although effective targeting of the malnourished chil- Henth edi dren would be difficult to achieve. Thejoint Nutrition Support ter-entions (E) . Program in Tanzania (UNICEF 1988b) estimated its costs at * Outpatientcare for coughs and colds (uc,(v+ +z A)), $10.05 per child per year (from both national and donor whereu=theincidenceofcoughsandcolds(per1,000), ,= sources), with the addition of $9.00 per child for start-up costs. the coverage or proportion of coughs and colds in the Annual costs may be estimated at $11.85 per child if the initial community which come to the attention of the health care program start-up costs can be spread over five years. On the system, v = the average cost of an ambulatory care visit or basis of this model, therefore, the cost per million population consultation, M = thecost of nonantimicrobial medications, would likely be between $810,000, for the targeted program for z = the proportion of URI cases inappropriately treated with the expected 54,000 children (36 percent, on the basis of 1990 antimicrobials, and A = the average cost of a course of UNICEF data) (UNICEF 1991) who are malnourished, and antimicrobials). Table 4A- 1. Range of Values for Variables in Model of Cost-Effectiveness of ARi Case Management Symbol Varable Least favorable Most likely Most favorable E Sensitization cost $800,000 $400,000 $80,000 v Cost of one outpatient consultation $2.00 $1.50 $1.00 M Cost per episode of non-antimicrobial pharmaceuticals $3.20 $0.08 0 A Cost per episode ofantimicrobials $7.00 $0.80 $0.16 H Cost of inlpatient or referral care $135 $45 $6 N Average number of visits per episode 2 1.5 I u Cases of cough, anid colds 1,500S000 1,050,000 600,000 p Cases of uncomplicated pneumonia 5,000 7,500 10,000 s Cases of severe or complicated pneumonia 1,000 1,500 2,000 uCl Proportion of URI cases seen and treated by health worker 0.10 0.05 0.02 z Proportion ofURI cases seen and treated with antibiotics 0.50 0.10 0 Cp~ Proportion of uncomplicated pneumonia cases appropriately 0.10 0.40 0.70 diagnosed and treated Proportion of severe pneumonia cases appropriately diagnosed 0.40 0.65 0.90 and treated Fp Case-fatality ratio for untreated uncomplicated pneumonia 0.10 0.13 0.20 F, Case-fatality ratio for untreated severe or complicated 0.25 0.35 0.50 pnetimoniia R, Percent reduction in mortaliry with appropriate antibiotic 0.60 0.80 0.90 treatment Sour.ce: See text of appendix. 84 Sally K. Stansfield and Donald S. Shepard Table 4A-2. Derived Variables and Their Most Likely Values Most likely value Symbol Variable Derivation' (U.S. dollars) c Cost per capita of ARI care E+UC,, -(V+M+ZZA)+P cp 0.54 (M+A+N V)+S C,(V+H) TP Number of uncomplicated pneumonia cases treated P CP 3,000 T, Number of complicated or severe pneumonia cases treated s .c 975 D Number of ALRI deaths averted (Tp * Fp + T F - R 585 CE Cost-effectiveness, or cost per ALRI death averted C, D 926 CE,,ALI Cost per disability-adjusted life-year saved CE/25 37 a. For symbols not defined in this table, see table 4A-1. * Outpatient care for pneumonia (p c (M + A + N . V)), visit in many Latin American countries, the intermediate cost where P = the incidence of pneumonia (per million), cp = of $1.50 and low cost of $1.00 are more typical of costs per the coverage or proportion of pneumonia cases in the com- outpatient consultation in Asia and Africa. Costs for non- munity which are diagnosed and treated appropriately (that antimicrobial pharmaceuticals (m), which are optional in the is, given antimicrobials with or without other medications management of ARI, are estimated at zero for the low-cost for supportive care), and N = the number of consultations figure, $0.08 for intermediate costs typical of five days supply per episode). of a locally made cough syrup, and $3.20 to reflect the often * Inpatient care for severe pneumonia (S. C, (H + V ) larger expenditure on nonantimicrobial pharmaceuticals in where s = the incidence of severe pneumnonia (per million), many developing countries (Quick and others 1988). c, = the coverage or proportion of severe pneumonia in the The values selected for the cost per episode of antimicrobials community which is diagnosed and treated appropriately (A) were based on UNICEF prices in 1988 for a five-day course (that is, given antimicrobials and referred for more special- for a child weighing ten kilograms. These figures were doubled ized care), and H = the cost of referral, generally including to include costs for transport, packaging, and dispensing these impatient hospital care). medications. Basic prices included a low-cost figure of $0.08 for five days of co-trimoxazole for a ten-kilogram child, $0.40 Therefore, the total cost of case management per million for the intermediate figure (Bates and others 1987; Quick and population is: others 1988), and $3.50 for intramuscular penicillin and chlor- amphenicol for the high-cost program. Costs for referral care with hospitalization (H) used in the model include a high-cost E + UCu(V + M + ZA) + PCp.(M + A + N.V)+ sC5(H + V) figure of $135 (three days at $45 per day) from Brazil, an intermediate figure of $45 (three days at $15 per day) from Rwanda (Shepard 1989), and a low-cost figure of $6 (three Clearly, each of the variables in the model has a range of days at $2 per day). The ideal number of visits per episode (N) values. Calculations of cost and effectiveness were made using is 2, although a lower average value of I is used, because some a range of values including "most favorable," for the effect on programs require no follow-up visit (WHO 1988). The interme- cost-effectiveness, "least favorable," and "most likely." The diate figure of 1.5 reflects probable level of compliance with specific values used for each variable are listed in table 4A-1 the recommended follow-up visit. and the "most likely" values for the derived variables in The values used for high, intermediate, and low incidence table 4A-2. of coughs and colds among children under five (u) are ten, Sensitization costs (E) are derived from estimates by Phillips, seven, and four per child per year (Datta Banik, Krishna, and Feachem, and Mills(1987), includingalow-costoption, which Mane 1969; Kamath and others 1969; Friej and Wall 1977; used person-to-person communications at a cost of $1 per Foreit and Lesevic 1987), yielding the numbers of episodes mother (in groups of ten), and a high-cost program, which used specified in table 4A- I among the 150,000 children under five mass media at a cost of $10 per mother (assuming that there in the standard population. For pneumonia incidence (P), the are 80,000 mothers of children under five). The intermediate figures selected were 100, 50, and 25 per 1,000 (Riley and cost program estimate of a cost of $5 per mother equals Douglas 1981; Pio, Leowski, and Luelmo 1982; WHO/ARI those estimates used for breastfeeding promotion and weaning 1989b), suggesting a range of 5,000 to 10,000 cases of uncom- education programs (Feachem 1986; Phillips, Feachem, and plicated pneumonia. For severe pneumonia (s), the figures are Mills 1987). 25, 10, and 5 per 1,000 (Chen and others 1980; Riley and The cost of one outpatient consultation (v) has been de- others 1983; WHO 1984), implying a range of 1,000 to 2,000 rived from figures for diarrheal disease and immunization con- cases of severe pneumonia, with a most likely value of 1,500. sultations (Phillips, Feachem, and Mills 1987), under the Although ideally fewer than 2 percent of coughs and colds assumption that the time spent by the health worker is com- will be brought to the attention of and diagnosed by the health parable. A high-cost figure of $2.00 reflects average costs per worker (cu), a likelier figure is 5 percent, and more than 10 Acute Respirators Infection 85 percent has been observed in some programs (Foreit and ratios (Fp and F,) specified in table 4-6 yield ARI-specific mor- Lesevic 1987). The cost of inappropriate treatment of coughs tality values of 5 per t,000, 10 per 1,000, and 20 per 1,000, and colds with antibiotics is included in the model, since this reflecting probable levels of API mortality (UNICEF 1991; WHO/ARI may be a source of excess costs and of potential savings in 1991b) among children under five in middle-mortality, high- improving case management practices (Stansfield 1990). The mortality, and very high mortality countries. Numbers of percentage of coughs and colds seen by a health worker and deaths averted were, therefore, calculated on the basis of treated inappropriately with antibiotics (z) will ideally be zero, expected numbers of deaths of 750 (5 per 1,000), 1,500 (10 per although in many programs up to half of such cases receive 1,000), and3,000 (50per 1,000) formoderate-, high-, and very antimicrobials. A most likely value of 10 percent should be high mortality countries, respectively. The range of assump- achievable with careful training and supervision of health tions for effectiveness of treatment (R) includes 90 percent workers. A study in Lesotho (Redd, Moteetee, and Waldman for a highly efficacious program (Berman and McIntosh 1985), 1990) found that 6 to 15 percent of practitioners (before being 80 percent for the internmediate level of effectiveness (Institute retrained underwHoguidelines) reported that they would treat of Medicine 1986), and 60 percent for the lower level of a cough or cold with antimicrobials, so it is likely that observed effectivenesss, such as may be seen in settings where compli- rates of inappropriate use of antimicrobials would exceed these ance is poor. reported rates. The additional variables defined in table 4A-2 were derived For pneumonia coverage (cp), the figures selected for the from the values for each variable specified in table 4A-1. model are 70 percent, 40 percent, and 10 percent. The World Calculation of the cost per capita of ARI care (c) using these HealthOrganizationestimates that 12 percent ofall childhood figures yields a most likely per capita value of $0.54, or pneumonias were treated with antibiotics in 1990, and it $541,877 for the nample population of I million. The ranges projects increases to 40 percent in 1995 and 60percentby 2000 for total cost (of $220,000 to $940,000) and for the cost-effec- (WHO/ARI 199 Ia). Although there are no good data from oper- tiveness figures used in table 4-3 reflect values obtained from ational setting, it has been estimated that from 40 to 90 percent the sensitivity analysis, which is summarized in table 4A-3, (with a most likely value of 65) of severe pneumonias may be obtained by varying one parameter at a time. The sensitivity seen and diagnosed by a health care worker (c,). Incidence and analysis data indicate that program costs and cost-effectiveness coverage figures selected for URI and pneumonias suggest that are most sensitive to the costs for health education, or "sensi- 4 percent (least favorable) to 50 percent ("most favorable")- tization." The program effectiveness (as measured by deaths with a most likely value of 17 percent-of all cases of ARI averted) is most sensitive to the proportion of uncomplicated presenting to health facilities would be diagnosed as pneumo- pneumonia cases appropriately diagnosed and treated. nia. These figures reflect such measurements made in opera- A most likely value for cost-effectiveness for ARI case man- tional settings (WHO/ARI 1990; Foreit and Lesevic 1987; Quick agementof$926(perdeathaverted)wascalculated.Useofthe and others 1988). The incidence (p and s) and case-fatality extreme figures for incidence and case-fatality ratios (rather Table 4A-3. Least and Most Favorable Costs, Effectiveness, and Cost-Effectiveness for Each Variable in Case Management Model for ARI Cost-effectiveness Cost per capita (U.S. dollars) Deaths averted (cost per death averted, U. S. dollars) Variablea Leastfavorable Mostfavorable Leastfavorable Mostfavorable Leastfavorable Mostfavorable E 0.94 0.22 585 585 E,610 379 v 0.57 0.51 585 585 976 877 M 0.72 0.54 585 585 1,222 919 A 0.59 0.54 585 585 1,014 917 H 0.63 0.50 585 585 1,076 861 N 0.54 0.54 585 585 930 922 U 0.58 0.50 585 585 990 862 r 0.54 0.55 481 689 1,120 791 S 0.53 0.56 494 676 1,066 824 Cu 0.63 0.49 585 585 1,075 837 z 0.56 0.54 585 585 955 919 Cp 0.53 0.55 3-51 819 1,524 670 C, 0.52 0.56 480 690 1,093 811 Fp 0.54 0.54 513 753 1,056 720 Fs 0.54 0.54 585 702 926 772 R 0.54 0.54 439 658 1,235 823 Note: Figures enclosed in boxes represent the minimum and maximum values for cost, effectiveness, and cost-effectiveness. a. For definition of symbols, see table 4A-1. Source: Authors' data. 86 Sally K. Stansfield and Donald S. Shepard Table 4A-4. Cost-Effectiveness of ARI Case Bates, J. A., Bimo, P. Tengko, P. Foreman, B. Santoso. D. Adi, D. Ross- Management for Low, and High-Mortality Countries Degnan, and J. D. Quick. 1987. "Child Survival Pharmaceuticals in Indo- nesia: Opportunities for Therapeutic and Economic Efficiencies in Low-mortality High-mortatit-s Pharmaceutical Supply and Use." Management Sciences for Health. Mortality couUtre country Yayasan Indonesia Sejahtera, Ministry of Health. Boston, Mass. Cost per capita of AR] care $0.52 $0.56 Berman. Stephen. 1991. 'Epidemiology of Acute Respiratory Infections in Cost per capeta pofuaI cae $0.59 $0.61 Children of Developing Countries." Reciews of Infectous Diseases 13(Sup- C ost per target population $3.49 $3.61plmn :S5-2 [>eaths averted 338 1,160 plement 6):S454-62. Cost per death averted $1,152 $483 Berman, Stephen, A. Duenas, A. Bedoya, and others. 1983. "Acute Lower Cost per disability-adjusted life- $46 $19 Respiratory Tract Illnesses in Cali, Colombia: A Two-Year Ambulatory year saved Studv ." Pediamcs 71:2110-] 8. Berman, Stephen, and Kenneth Mcintosh. 1985. "Selective Primary Health Source; See text of appendix. Care: Strategies for Control of Disease in the Developing World. XXI. Acute Respiratory Infections." Reviews of Infectiou Diseases 71(5):674-91. Black. F. L. 1982. "The Role of Herd Immunity in Control of Measles." Yale than varying one parameter at a time, as in the sensitivity Journal of BiologN and Medicine 55:351-60. analysis), suchi as may be observed in high- and low-mortality Black, S. B.. H. R. Shinefield, R. A. Hiatt, B. H. Fireman, and the Kaiser countries, yields cost and effectiveness figures specified in table Permanente Pediatric Vaccine Study Group. 1988. "Efficacy of Haemophilus 4A-4. These figures underline the fact that interventions to influenzae Type B Capsular Polysaccharide Vaccine." Pedianmic Infectious improve ARI case management are of highest priority in the Disease 7:149-56. countries with high overall and ARI-specific infant and child Bloem,. N-W., M. Wedel, R. J. Egger and others. 1990. "Mild Vitamin A motait rte.These estimates are higher than the estimates Deficiency and Risk of RespiratoryTract Diseasesand Diarrhea in Preschool mortality rates. X nese estlmates are nlgne tnan tne estimates dand Schoiol Children in Northeastem Thailand." AmencanJournal of Epi- of $350 per death averted obtained in a field study in Nepal demiology 131:332-39. and $131 per death averted obtained in the Philippines (John Borrero,1 ., L. Farjardo P., A. Bedoya M.. A. Zea, F. Carmona, and M. F. de Snow International, Linpublished data). Borrero. 1990. "Acute Respiratory Tract Infections amoing a Birth Cohort Calculations ofcost per discounted healthy year of life saved of Children from Cali, Colombia, Who Were Studied through 17 Months are made using a life expectancy of fifty years, an average age of Age." Reviews of Infectious Diseases 12(Supplement 8):S950-56. at death of two years, and a discount rate of 3 percent per Brahmans, D. 1986. "Does Pertussis Vaccine Cause Brain Damage?" Lancet annum. Ranges for the intermediate variables and a summary 2:286. of the cost-effectiveness calculation are presented in table 4-3 Brenzel, Logan. 1989. "The Costs ofEpi: A Review of Cost and Cost-Effective- in a format for comparison with ttie analogous figures for the ness Studies (1979-1987)." Resources for Child Health (REACH) Project, In a format for comparison with the analogous figures for the Rsln a onSo.Ic,Arl other AR] interventions. ~ ~~~~~~~Rosslyn, Va.- John Snow, Inc., April. oh A_neeis1990. "Cost and Cost-Effectiveness Issues for ARI Case Management." In ALRI and Child Survival in Developing Countries: Understanding the Current Status and Directions for the 1990s. Proceedings of a workshop given b5 the Notes Johns Hopkins University School of Hygiene and Public Health. Washing- ton, D.C.: Institute for Intemational Programs, and the U.S. Agency for The authors wouild like to express their gratitude for the invaluable contri- Intemational Development, August 2-3. butions of the many colleagues and friends who reviewed and commented on Broome, C. V. 1981. "Epidemiology of Pertussis, Atlanta, 1977." Journal of drafts of this chapter, with special recognition (of assistance provided by C. J. Pediamzs 98:362-67. Clements, Nils M. P. Daulaire, Floyd W. Denny, Don de Savigny, Ralph R. Bulia, A., and K. L. Hitze. 1978. "Acute Respiratory Infections: A Review." Frerichs, Y. Ghendon, Sandy Cove. Davidson Gwatkin, Dean Jamison, Joel Bulletin of the World Health Organization 56:481-98. A. Lamounier, Nancy Pielemeier, Antonio Pio, Mark C. Steinhoff, Tessa L. Burchfiel, C. M., M. W. Higgins, J. B. Keller, W. F. Howatt, W. J. Butler, and Tan-Torres, and James Toilloch. 1. T. T. Higgins. 1986. "Passive Smoking in Childhood." Amencan Review of Respiraros Diseases 133:966- 73. Bwibo. N. C). 1971. "Whooping Cough in Uganda." Scandinavian Journal of References Infectious Diseases 3:41-43. Byass, P., H. Campbell, T. J. O'Dempsey, and B. M. Greenwood. 1991. Aaby, Peter. 1988. "Malniutrition and Overcrowding/lntensive Exposure in "Coincidence Lof Malaria Parasitemia and Abnormal Chest X-ray Findings Severe Measles Infection: Review of Comimunity StLdies." Revieus of infec- in Young Gambian Children." Journal of Tropical Medicine and Hy.giene tiots Diseases 10:478-91. 94.22 23. Armstrong, J. R. NM., and H. Campbell. 1991. "Indoor Air Pollution ExposuIre Camargos, P. A., M. D. Guimaraes. and E. de F. Drummond. 1989. "Mortali- and Lower Respiratory Infections in Young Ganmbian Children." Interna- dade por pneumonia em cnancas menores de conco anos de idade em ntinal . ulotol (If EplemniooLgy 20(2):424-29. localidade do estado de Minas Gerais (Brasil), 1979-1985." Revue de Saude Ashworth. Ann and Richard G. Feacheni. 1986. Interventions for the Control Puiatca 23(5):398-94. ofDiarrhoeal Diseases among Y)ungChildre7n GiouthMonit(rng Programmes Campbell. H., J. R. M. Armstrong. and P. Byass 1989. "Indoor Air Pollution Geneva: Wiorld Health Organization. in Developing Countries and Acute Respiratory Infection in Clsildren." Balint, O.. aind K. Aisand. 1979. "Infectious and Parasitic Diseases in Zambian Lance L 1: I 1 2. Childreni." Tropical Doctor 9:99-103. Campbell, H., P Bvass, and B. Greenwood. 1988. "Simple Clinical Signs for Barclay, A. J. G., A. Foster, and A. Sommer. 1987. "Vitamin A Supplemien- Diagnosis of Acute Lower Respiratory Tract Infections." Lancet 1:742-43. tation and Meortality Related to Measles: A Randomised Clinical Trial." Campbell, H.. P. Byass. I. M. Firgle,and N. Lloyd-Evans. 1989. "Clinical Signs Bsitish MedicalJ(ournal 294:294-96. of Pneumoniia in Children." Lancet 1:899-900. Acute Respiratorv Infection 87 Campbell, H-, P. Byass, A. C. Lamont, and others. 1989. "Assessment of Feachem, R. G A. 1986. "Preventing Diarrhoea: What Are the Policv Clinical Criteria for Identification of Acute Lower Respiratory Tract Infec- Options ?" Health Policy anid Planning 1(2): 109-17. tions in Children." Lancet 1:297-99. Feachem, R. G. A., and M. A. Koblinsky. 1983. "Interventions for the Control Chandra, R. K. 1979. "Prospective Studies of the Effect of Breastfeeding on of Diarrhoeal Diseases among Young Children: Measles Immtinization." the Incidence of Infection and Allergy." Acta Pediat-ica Scandinatica Bullerin of the World Health Organization 61(4):641-52. 68:691-94. - 1984. "Interventions for the Control of Diarrhoeal Diseases among Chaulet, P., and N. Ait Khaled. 1982. "Enquete sur les infections respiratoires Young Children: Promotion of Breastfeeding." Bulletin of the World Health aigues et la prescription d'antibiotiques dans les unitrs sanitaires de base en Organization 62(2):271-91. Algeria, au cours de l'annee 1980." Revue Franqats des Maladhes Respiratoires Foreit, Karen (G., D. Haustein, M. Winterhalter, and E. LaS Mata. 1991. "Costs 10(1):45-52. and Benefits of Implementing Child Survival Services at a Private Mining Chen, B. H.. C. J. Hong, M. R. Pandey, and K. R. Smith. 1990. "Indoor Air Company in Peru." AmericanJournial of Public Health 81(8): 1055-57. Pollution in Developing Countries." WorldHealh StatisLsQuarterly 43:127-38. Foreit, Karen G., and B. Lesevic. 1987. "Btisiness Analysis: Implementing Chen, Lincoln C., Mizanur Rahman, and A. M. Sarder. 1980. "Epidemiology Child Survival Services at a Private M ining Company in Pent." Technical and Causes of Death among Children in a Rural Area of Bangladesh." Information on Poptilation and the Private Sector (TiPits) Project, John InternationalJournal of Epidemiology 9:25-33. Short and Associates and the U.S. Agency for International Development. Chen Yue, Wanxian Li, and Shunzhang Yu. 1986. "Influence of Passive Sponsored by USAIt, Washington, D.C. Smoking on Admissions for Respiratory Illness in Early Childhood." British Foulke, F. G., K. G. Reeh, A. V. Graham, and others. 1988. "Family Function, Medicaljournal 293:303-6. Respiratory Illness and Ortins Media in Urban Black Infants." Family Med- Cherian, Thomas, T. Jacob John, Eric Simoes, Mark C. Steinhoff, and Merav icine 20: 128 32. John. 1988. "EvaluationofSimpleClinicalSignsfortheDiagnosisofAcute Freij, L., and S. Wall. 1977. "Exploring Child Health and Its Ecology: The Lower Respiratory Tract Infections." 1Lancer 2:125-28. Kirkos Stuidy in Addis Ababa: An Evaluation of Procedures in the Measure- Cherian, Thomas., Eric A. F. Simoes, Mark C. Steinhoff, K. Chitra, Meray mentofAcuteMorbidityandaSearchforCausalStructur." ActaPediatrica John, P. Raghupathy, and T. Jacob John. 1990. "Bronchiolitis in Tropical Scandinavica 5:267. South India." AmericanJournal of Diseases of Childreni 144. Funkhouser, A., M. C. Steinhoff, and ]. Ward. 1991. "Haemnophilusinfluenizae Chetry, . D. 1984. In Current Prohlens in Pediatrics 14:68. Disease and Immunization in Developing Couintries." Reviews of Infectious Church, M. A. 1979. "Evidence of Whooping CoUgh Vaccine Efficacv from Diseases 1 3(Supplement 06:542-54. the 1978 Whooping Cough Epidemic in Hertfordshire." Lancet 2:188-90. Garenne, Michel, Odile Leroy, Jean-Pietre Beau, and Ibrahiiam Sene. 1991. Cohen, Sheldon, David Tyrell, and Andrew Sniith. 1991. "Psychological Child Mortality after High-TitTe Measles Vaccines: Prospective Study in Stress and Susceptibility to the common cold." The New England journal of Senegal. Lancet 338(2):903-6. Medicine 325(9):606-12. Ghafoor, A., N. K. Nonuani, Z. Ishaq, S. Z. Zaidi, F. Abnwar, M. 1. Burney, A. Crofton, ]., and A. Douglas. 1975. Respirator'y Diseases. Oxford: Blackwell. W. Oureshi, and S. A. Ahmnad. 1990. "Diagnoses of Acute Lower Respira- Dabis, Fran,ois, A. Sow, Ronald Waldman, and others. 1988. "The Epidemi- tory Tract Infections in Children in Rawalpindi and Islamabad, Pakistan." ology of Measles in a Partially Vaccinated Population in an African City: Reviews of Infectious Diseases I 2(Supplement 8):S907- 14. Implications for Immunization Programs." Arniericanjournrl of Epidentiology Ghana Health Assessment Project Team. 1981. "A Quantitative Method of 127:171-78. Assessing the Health Impact of Different Diseases in Less Developed Datta, Neena 1987. "Acute Respiratory Infection in Low Birth Weight In- Countries." InternationallourralofEpidemiolog-y 10:73-80. fants." Indian Journal of Pediatrics 54:171-76. Gilsdorf, J. R. 1988. "Haemnophilus influenzae Type B Vaccine Efficacv in the Datta Banik, N.D., R. Krishna, and S. . S.NMane. 1969. "A Longitudinal Study United States." Pediatric Infectious Disease 7:147-48. of Morbidiry and Mortality Pattem of Children under Age of Five Years in Graham, N. M. H., A. J. Woodward, P. Ryan, and R. M. Douglas. 1990. "Acte an Urban Community." Indian Journal of Medical Research 57:948-57. Respiratory I llness in Adelaide Children. I 1: The Relationship of Matemal de Korting, H. W., K. R. Smith and J. R. Last. 1985. "Biomass Fuel Consuinp- Stress, Social Supports and Family Functioning." Internotionul ournal of tion and Health." Bulletin of the World Health Organization 63(1): 11-26. Epideiriokog.y 19(4):9 37-44. Denny, F. W., and W. A. Clyde. 1983. "Acute Respiratory Tract Infections: Granoff, D. M., P. G. Shackelford, B. K. SUarex, and others. 1986. An Overview." Pediatric Research 17:1026-29. "Hoemorphilusinfluenzae Type b lisease in Children Vaccinated with Tvpe Douglas, R. G., K. M. Lindgram, and R. B. Cough. 1968. "Expostire to Cold B Polvsaccharide Vaccine." New' EnglandJournal of Medicine 315:1583-90. Environment and Rhinovirus Cold: Failure to Demonstrate an Effect." New Greenwoood, B. M., A. M. Greenwood, A. K. Bradley, S. Tulloch, R. Hayes, EnglandJoutral of Medicine 279:742-47. and F. S. J. Oldfield. 1988. "Deaths in Infancy and Early Childhood in a Ekwo, E. E., M. W. Weinberger, P. A. Lachenbruch, and W. H. HLntley. 1983. Well-Vaccinated, Rural, West African Population," AAnnals of Tropical "Relationship of Parental Smoking and Gas Cooking to Respiratory Disease Paediamcs 7:91-99. in Children." Chest 84:662-68. Gutrirrez, G., M. C. Martinez, H. Guiscafre, G. Gomez, A. Peniche, and 0. El-Mowua, M. I., K. Twan-Danso, B. H. AI-Awamy, G. A. Niazi, andc M. T. Mufno:. 1986. "Encuesta sobre el uIso de antimicrohianos en las infecciones Altork. 1988. "Pneumnococcal Infections in Eastern Saudi Arabia: Serotvpes repiratorisS agudas en la poblaci6n rmral mexicana." Boletiin Medico-Hospital and Antibiotic Sensitivity Pattems." Tropical and Geographic Medicine InfantildeMixtcn 43(12):76l-68. 40:213-17 Harlap, S., and A. M. Davies. 1974. "Infant Admissions to Hospital and Escobar, J. A., A. S. Dover, and A. Duenas. 1976. "Etiology of Respiratory Matemal Smoking." Lancet 1(875):529-32. Infections in Childre in Call, Colombia." Pediatrics 57:123-30. Harrison, L. H., A. W. Hightower, S. Gaventa, and others. 1987. "Case- Escola, J. H. Kiyhty, A. K. Takala, H. Peltola, P.-R. Ronnherg, E. Kela, E. Control Efficacy Study of the Polysacchanrde Haemophilus influenizae Type Pekkanen, P. H. McVerry, and P. H. Makela. 1990. "A Randomized, B (Hib) Vaccine." Abstract 317, presented at the 27th Interscience Prospective Field Trial of a Conjugate Vaccine in the Protection of Infants Conference on Antimicrobial Agents and Chemotherapy, New York, and Young Children against Invasive Haemnophilus influenzae Type B Dis- October 5. ease." New Englandjournal of Medicine 323(20):1381-87. Hay,J. W., and R. S. Dauni. 1987. "Cost-Benefit Analysis of Two Strategies Fine, P. E., and J. A. Clarkson. 1987. "Reflections on the Efficacy of PertLssIs for Prevention of Haemnophilus influenizae Tvpe B Infection." Pediamcs Vaccine." Reviews of Infectious Diseases 9(5):866-83. 80(3): 319-29. 88 SallN K. Stansfield and Donald S. Shepard Hayes, Richard, Thierrv Mertens, Geraldine l ockett, and Laura Rodrigues. Lepage, P., C. Munyakazi, and P. HIennart. 1981. "Breastfeedinig and Hospital 1989. "Causes of Adult Deaths in Developing Countries: A Review of Data Mortalitv in Children in Rwanda." Lancet: 409-11. and Methods." enrr (Policy, Planning and Research) Working P'aper. World Linnemann, C. C., M. S. Yine, G. A. Roselle, and others. 1982. "Measles Bank, Population Health and Nutrition, Population and Human Re.ources Immunity after Revaccination: Results in Children Vaccinated before 10 Department. July. Monthsof Age." PedianTcs69:332-35. Hcvmann, David L., G. K. Mayben, K. R. Murphy, B. Guyer, and S. 0. Foster. Lipsky, Benjamin A., E J. Bkylo, Thomas S. Inul, and others. 1986. "Risk 1983. " Measles Control in Yaounde: Justification of a Onie Dose, Nine- Factors tor Acquiring PneUmrococcal Infections." Archives of Internl Medi- Month Mittiniuni Age Vaccination Policy in Tropical Africa." Lancet cutie 146:2179-85. 127:788-94. -Licero, M. G., T. E. TUpasi, M. L. 0. Go5me:, G. L. Beltrin, A. U. Cris6stomo, Honicky, R. F , J. S. Osborne, and C. A. Akpom. 1985. "Symptonms of V. V. Romano, and L. M. Rivera. 1990. "Respiratorv Rare Greater than 50 Respiratory Illness in YoungChildren and rhe Use ofWood-BurnmigSroves per Minute as a Clinical Indicator of Pneumonia in Filipino Children with tor Indoor Heating." Pediatrics 75( 3): 587-93. a Cough." Revieus of Infectious Diseases 1 2(Supplement 8:SIQ081-83. Hossain, NI. NI., R. 1. (Glass, and kM. R. Khan. 1982. "Antibiotic Use in a Rural McConnochie, K. MN., C. B. Hall, and W. H. Barker. 1988. "Lower Respiratory Community in Bangladesh." hia'rnationalJounal ofEpdemiology I I (4):402-5. Tract 11lness in the First T wo Years of Life: Epidemiologic Patterns and Costs Hull, Harry F., J. W. Pap, anid F. Oldfield. 1983. "Measles Nvortaliht and its a Suburban P'ediatric Pratctice." American Journal of Public Health Vaccine Efficacy in Rural West Africa." Lancet 1:972-75. 78(1): 34-39. Hussey, G. UD., and M. Klein. 1990. "A Randomized, Controlled Trial of 1McLean,A.R., and R. NAnderson. 1988. "Measles in DevelopingCountries. Vitamin A tn Children with Severe Measles." Neu. England Journal of Part 11. The Predicted Impact of 1N1ass Vaccination." Epidemriology and Medicine 323(3):160-64. Infection 100:419-44. Ikejani, 0. 1966. "Immunity against Diphtheria in Nigerian Children." West NMarkowit7, Milton. 1981. "Observations (n the Epidemiology and Pre- African MedicalJournal: 272-77. ventibility of Rheumatic Fever in Developing Countries." Cinic(al TheTa- Institute of Medicine. 1986. "The Burden of Disease Resulting from Acite teutics4(4):240-51. Respiratory Illnesses." In New, Vaccine Develoipment: Establishing Prionties Martin, T. R., and M. L). Bracken. 1986. "Association of Low Birth Weight Volumes I and 2. Washington, D.C.: National Academy Press with Passive Smoking in Pregnancy." Amencan Joinmal of Epidemiolog Jackson. G. G'., R. L. Muldoon, (G. C. Johnson, and others. 1963. "Contrrbu- 124:633-42. tion of Volunteers to Studies of the Common Cold." American Retiew oif Mastro, Timothy D., Abdul Ghafoor, Nasreen Khalir Nomani, Zahir Ishaq, RespiratorN Diseases 88(Supplement): 120-27. Favzal Anwar, Dani M. Granoff, John S. Spika, Clyde Thomsberrvt, and James, J. W. 1971. "Longituidinal Study of the Morbidity of Diartheal and Richard R. Facklam. 1991. "Antimicrobial Resistance of Pneumococci in Respiratory Infections in Malnourished Children." Amencan Journal of Children with Acute LowerRespiratoryTract Infection in Pakistan." Lancet Chnical Nutrition 25:690-94. 3 37.156-59 Jason,Janine NI., Philip Nieburg,J. S. Marks, and others. 1984. "Mortality and Mata, Leonardo J. 1978. The Children of Santa Mana Cauque: A Prospective Infectious Disease Associated with Infant Feeding Practices in Developing Field Study of Health and Growth. Cambridge, Mass.: MIT Press. Countries." Pediatnics 74(Supplenient):702-27. Miller, E., L. A. E. Ashworth, A. Robinson, P. A. Wright, and L. 1. Irons. 1991. Kamat, S. R., K. D. Godkhindi, B. W. Shah, and others. 1980. "Correlation "Phase 11 Trial of Whole-Cell Pertussis Vaccine vs. an Acellular Vaccine of Health Morbidity to Air Pollution Levels in Bombay City: Results of a Containing Agglutinogens." Lancet 337:70-73. Prospective 3 Year Stirvey at One Year." Journal of Postgraduate Medicine Mohs, Edgar. 1985. "Acute Respiratory Infections in Children: Possible Con- 26:45--62. trol Measures." PAHCO Bulletin 19:82-87. Kamath, K. R., R. A. Feldman, P. s. s. Sundar Rao, and others. 1969. Monto, A. S., and B. M. Ullman. 1974. "Acute Respiratory Illness in an "Infection and Disease in a Group of Souths Indian Families." American American Community: The Tecumseh Study." lournal of the Amencan JournalofEpidemilog-y 89:375-83. Medical Association 227:164-69. Keja, K., C. Chan, G. Hayden, and R. Hf. Henderson. 1988. "Expanded Mosley,W. Henry. 1985."Child Survival Research and Policy." WorldHealth Programme on Immunization." WVorld Health Statistics Quarterly 41:59-63. Forum 6:352460. Khan, J. A., D. G. Addiss, and Rizwan-Ullah. 1990. "Pneumonia and Com- Mosley, W. Henry, and S. Becker. 1988. "Demographic Models for Child munity Health Workers." Lanicet 336(8720):939. Survival: Implications for Program Strategy." Paper presented at the semi- Kiecolt-Glaser, J. K., and R. Glaser. 1986. "Psychological Influences on narChild Survival Programs: Issues for the I990s,Johns Hopkins University Immunity." PsNchosomatics 27:621-24. School of Hygiene and Public Health. Washington, D.C.: Institute for Kielmann, Ainfried Arndd, and Coln McCord. 1978. "Weight-for-Age as an Intemational Programs, November 21. Index of Risk of Death in Children." Lancet 1:1247-350. Muller, A. S., J. Leeuwenburg, and D. S. Pratt. 1986. "Pertussis: Epidemiology Koplan, J. P., S. C. Schsenhaum, MN. C. Weinstein, and D. W. Fraser. 1979. and Control." Bulletin of the World Health Organizanon 64(23:321-31. "Pertussis Vaccine-An Analysis of Benefits, Risks and Costs." The Neuw Muyembe,J.J.,F.GattijJ.Spaepen,andJ. Vandepitte. 1972. "L'epid6miologie EnglanUd]ournal of Medicine 301(17):906-I 1. de la diphtrrie en la R6publique du Zaire: le rn3le de l'intection cutanee." Kossove, D. 1982. "Smoke-Filled Roooms anid Lower Respiratory Disease in AnLales de la Socmcid BeIge de Medecine Tropicale 52:141-52. Infaiits." Sa Mediese Tydskn 24 April:622-23. Naran,J. P., andT. L). Sharma. 1987. "Acute Respiratory Infections in Kangra Lataorre Otin, C., T. Juncosa Morros, and t. Sanfeliu Sala. 1988. "Antibiotic District: Nlagnitude and Current Treatment Practices." Indian JounaIl of Susceptibility of Streptococcuo pneumoniae Isolates from Pediatric Patients." Pediatrncs 54(3):441-44. jouumal ofAntimicr-ohialChemnotherapy 22:659465. Ngalikpima, V. F. 1983. "Review of Respiratory Infections in a L)eveloping Leeder, S. R., R. Corkhill, L. NI. Irwig, W. W. Holland, and J. R. T. Colley. (Country." EuropeanJournalofRespiratory Diseases 64:481-86. 1976. "Influence of Family Factors on the Incidence of Respiratorn Infec- Osterholm, M. T., J. H. Rambeck, K. E. White, and others. 1987. "Lack of tionsin the First YearofLife." BntishJournalofPrevennveandSocialMedicine Protective Efficacy and Increased Risk of Disease within 7 Days aftcr 30(4):203-1 2. Vaccination Associated with Haermophilusinfluemmzae Type B (Hih) Polysac- Leowski,Jerzy 1986. "NMortalityfrom Acute Respiratory Infections in Children charide (PS) Vaccine Use ini Minnesota." Abstract 318, presented at the under Five Years ofAge: Global Estimates." W'orldHealthStanstics Quarierl) 27th Interscience Conference on Antimicrobial Agents and Chemorher- 39:13844. sp, New York, October 5 Acute Respiratory Infection 89 PAHO (Pan-American Health Organization). 1980. "Acute Respiratorv Infec- Salih. M. A., F. El Hakim, G. 1. Suliman, S. A. Hassan, and A. El Khatim. tions in the Americas." PAHO Epidemiological Bulletin 1(5):1-4. 1985. "An Epidemiological Study of the 1978 Outbreak of Diphtheria in Pandey, M. R., and AR] Study Team. 1988. "Acute Respirat(ory Infections in Khartoum Province." Journal of Tropical Pediamcs 31:812. Rural Hill Regioin of Nepal: A Prospective Pilot Intervention Stu(ly." Samet, J. M., M. C. Marbury, and J. D. Spengler. 1987. "Health Effects and Mrigendra Medical Trust, Kathmandu. Sources of Indoor Air Pollution." American Review of Respiratory Diseases Pandey. M. R.. J. S. M. Boleij, K. R. Smith, and E. M. Wafula. 1989. "Indcoor 136:1486I1508. Air Pollution in Developing Countries and Acute Respiratory Infection in Selwyn, B. J., on behalf of the Coordinated Data Group of BOSTIDResearchers. Children." L.incet 1:427-29. 1990. "The Epidemiology of Acute Respiratory Tract Infection in Young Pandey, M. R., R. P. Neupane, A. Gautam. and 1. B. Shrestha. 1989. "Domestic Children: Comparison of Findings from Several Developing Countries." Smoke Pollution and Acute Respiratory Infections in a Rural Community Reviews of Infectious Diseases 12(Supplement 8):S870-88. of the Hill Region of Nepal. " Envilronmenit lnteniational. Shann, F.,,M. Gratten. S. Germer, and others. 1984. "Aeriology of Pneumonia Pedreira, F. A., V. L. Guandolo, E. F. Feroli, G. W. Mella, and 1. P. Weiss. in Children in Goroka Hospital, Papua New Guinea." Lancet2:537-41. 1985. "Involuntary Smoking and Incidence of Respiratory lllness during the Shann, F., K. Hart. and D. Thomas. 1982. "Acute LowerRespiratory Infecrions First Year of Life." Pediatncs 75:594-97. in Children: Possible Criteria for Selection of Patients for Antibiotic Penna, M. L. F., and M. P. Duchiade. 1991. "Air Pollution and Infant Mortalit., Therapy and Hospital Admission." Bullenn of the World Health Organizanon from Pneumonia in Rio de Janeiro Metropolitan Area." PAHO Bulletin 62:749-53. 25(1):47-54. Shepard,D. S.,R. Robertson,C.Cameron, P. Satarus,M. Pollock.J. Manceau, Phillips, M.A., R. GA. Feachem, andrA. Mv1ills. 1987. "Options forDiarrhoea P. Meissner, and J. Perroni. 1989. "Cost Effectiveness of Routine and Control: The Cost-Effectiveness of Selected Interventions for the Preven- Campaign Vaccination Strategies in Ecuador." Bulletin of the World Health tion of Diarrhoea." Publication 13, EPC (Evaluation and Planning Centre Organization 67:649-62. for Health Catre), London School oif Hygiene and Tropical Medicine and the Siber, G. R., M. Santosham, G. R. Reid, C. Thompson, J. Almeido-Hill, A. Diarrhoeal Diseases Control Division, WHO (World Health Organiation). Morell, G. DeLange, J. K. Ketcham and E. H. Callahan. 1990. "Impaired Pio, A.J. Leowski, and F. Luelmo. 1982. "Bases forwHoProgramme on Acute Antibody Response to Haemophilus influenzae Type B Polysaccharide and Respiratory Infections in Children." 25th World Conference on Tubercu- Low IgG2 and IgG4 Concentrations in Apache Children. New England losis, Buenos Aires, December 15-18. Jounial of Medicine 323(20):1387-92. Pio, A.,J. Leowski, and H. G. ten Dam. 1985. "The Magnitude of the Problem Simberkoff, M. S., A-P. Cross, M. Al-Ibrahim, and others. 1986. "Efficacy of of Acute Respiratory Infection." In R. M. Douglas and E. Kerby-Eaton, eds., Pneumococcal Vaccination in High Risk Patients." New England]ournal of Acute Respiratory Infections in Childhood. Proceedings of an International Medicine 315:1318-27. Workshop. Adelaide: University of Adelaide. Singhi, S., and P. Singhi. 1987. "Prevention of Acute Respiratory Infections." Poore, P. 1988. "Vaccination Strategies in Developing Countries." Vaccine Indian Joumal of Pediamcs 54:161-70. 6:393-98. Sofoluwe, G. 0. 1968. "Smoke Pollution in Dwellings of Infants with Bron- Pringle, C. R. 1987. "Progress towards Control of the Acute Respiratory Viral chopneurnonia." Archives of Environmental Health 16:670-72. Diseases of Childhosd." Bulletin of the World HealdiOrganizat(mi 65(2):133-37. Sommer, Alfred. 1983. "Mortality Associated with Mild Untreated Xeroph- Quick, Jonathan D., P. Foreman, D. Ross-Degnan, and others. 1988. "Where thalmia." Transacuons of the Amnencan Ophthalniological Society 81:825-53. Does the Tetracycline Go?: Health Center Prescribing and Child Sur- Sommer, Alfred,J. Katz, and 1. Tarwotjo. 1984. "Increased Risk of Respiratory vival in East Java and West Kalimantan, Indonesia." Report of Consul- Disease in Children with Preexisting Mild Vitamin A Deficiency." Amneri- tatiion. Drug Management Program, Management Sciences for Health, canJournal of Clinical Nutrinon 40:1090-95. Boston. Spika, John S., M. H. Munshi, B. Wo)tyniak, D. A. Sack, A. Hossain, M. Rahniathullah, L., B. A.Underw'ood.R. D.Thulasiraj, andR.C. Milton. 1991. Rahman, and S. K. Saha. 1989. "Acute Lower Respiratory Infections: A "Diarrhea, Respiratory Infections and Growth Are Not Affected by a Major Cause of Death in Children in Bangladesh." AnnuiLs of Tropical Weekly Low-Dose Vitamin A Supplement: A Masked, Controlled Field Paediamtcs9:33 39. TrialinChildreninSouthernIndia." AmencanJournalofClinicalNumtion Stansfield, S. K. 1987. "Acute Respiratory Infections in the Developing 54:568-77. World: Strategies for Prevention, Treatment and Control." Pediatric Infec- Rahmathullah. L., B. A. Underwood, R. D. Thulasiraj, R. C. Milt(n, K. tious Diseases 61:624-29. Ramaswamy, R. Rahmathullah, and G. Babu. 1990. "Reduced Mortality -. 1990. "Potential Savings through Reduction of Inappropriate Use of among Children in Southem India Receiving a Small Weekly Dose of Pharmaceuticals in the Treatment of Acute Respiratory Infections." In ALRI vitamin A." New Englandjournal of Medicine 323(14):929-35. and Child Surcrival in Developing Countmes: Understanding the Current Status Redd, Stephen, Mpolai Moteetee, and Ronald Waldman. 1990. "Diagnosis and Directions for the 1990s. Proceedings of a workshop given by the Johns and Management of Acute Respiratory Infections in Lesotho." Health Policy Hopkins University School of Hygiene and Public Health, Institute for and Planning 5(3):255-60. Intemational Programs, and the U. S. Agency for International Develop- Riley, Ian D., and R. M. Douglas. 1981. "An Epidemiologic Approach to ment. Washington, D.C., August 2-3. PneLImococcal Disease." Reviews of Infectious Diseases 3:233-45. Stein, C. M., W. T. A. Todd, D. Parirenyatwa, J. Chakonda, and A. G. M. Riley, Ian, Emmanuel Carrad, Helen Gratten, and others. 1983. "The Status Dizwani. 1984. "A Survey of Antibiotic Use in Harare Primary Care of Research on Acute Respiratorv Infections in Children in Papua New Clinics."Journal ofAntmnicrobialChemotherapy 14:149-56. Gumea." Pediatnc Research 17:1041-43. Stetler, H. C., W. A. Orenstein, R. H. Bemier, and others. 1986. "Impact of Riley, Ian D., Deborah Lehmann, MN. P. Alpers, and tithers. 1986. "Pneumo- Revaccination of Children Who Initiallv Received Measles Vaccine hefore coccal Vaccine Prevents Death frsm Acute Lower Respiratorv Tract Infec- IQ Months (f Age." Pediatrics 77:471-76. tions in Papua New Guinea Children." Lancet 2:877-81. Taylor, W. R., Ruti-Kalisa, Manibu ina-Disu, and J. M. Weinman. 1988. Rowland, NI. G. NI., S. G. H. G. Rowland, and T. !. Cole. 1988. "Impact Of "Measles Control Efforts in Urban Africa Complicated by High Incidence Infection on the Growth of Children from 0 to 2 Years in an Urban West of Measles in the First Year of Life." Arnencan Journal of Epidremiology African Conimisnity." Amencanjournal of Clnical Nutrition 47:134-38. 127(4):788-94. Rubeni, D. H., J. M. Leventhal, P. A. Krasilnikoff, B. Weile, and A. Berget. Tracev, V. V., N. C. De, and J. R. Harper. 1971. "Obesity and Respiratory 1986. "Effect of Passive Snioking on Birth Weight." Lancet 2:415-17. Infections in Infants and Young Children." Bintsh MedicalJournal I 1:16-18. 90 Sally K. Stanisfield and Donald S. Shepard Tugwell, Peter, Kathryn J. Bennett, 1). L. Sackett and R. B. Haynes. 1985. Ware, J. H., D. W. Dockery, A. Spiro Ill. F. E. Spaizer, and B. G. Ferris, Jr. "The Measurement Iterative Loop: A Framework for the Critical Appraisal 1984. "Passive Smoking, Gas Cooking, and Respiratory Health of of Need, Benefits and Costs of Health Interventions." Journal of Chronic Children Living in Six Cities." American Review of Respiratory Diseases Diseases 38:339-5 1. 129:366-74. Tupasi, Thelma E., 1985. "Nutrition and Acute Respiratory Infections." In R. Weinberg, G. A., D. Lehmann, T. E. Tupasi. and D. M. Granoff. 1990. NI. Douglas and E. Kerby-Eaton, eds., Acute Respiratory Infecions in Child- "Diversitv of Outer Menibrane Protein Profiles of Nontypeable Haetnophilus hood. Proceedings of an International Workshop. Adelaide: University of influenzae Isolated fronm Children from Papua New Guinea and the Philip- Adelaide. pines." Reviews of Infectious Diseases 12(Supplenient 8):S 1017-20. Tupasi, Thelma E., M. G. Lucero, D. M. Magdangal, N. V. Mangubat, Weiss, S. r., 1. B. Tager, M. Schenker, and F. E. Spaizer. 1983. "The Health NI. E. S. Sunico, C. U. Tortes, Lillian E. de Leon, J. F. Paladin, L. Baes, and Effects of Involtintary Smoking." Amenican Review of Respiratory Diseases M. C. Javato. 1990. "Etiology of Acute Lower Respiratory Tract Infections 128:933-42. in Children from Alabang, Metro Manila." Reviews of Infectious Diseases Wilkins, J., and P. F. Wehrle. 1979. "Additional Evidence against Measles 1 2(Supplement 8):S929-39. Vaccine Administration to Infants Less than 12 Months of Age: Altered Tupasi, Thelma E., Melicia A. Velmonte, Maria Elnor G. Sanvictores, Leticia Inimune Response Following Active/Passive Immunization." Journal of Pe- Abraham, Lillian E. de Leon, Susan A. Tan, Cynthia A. Miguel, and diatncs 94:865-69. Mediadora C. Saniel. 1988. "Determinants of Morhidity and Mortality due Williams, P. J., and F. H. Hull. 1983. "Status of Measles in the Gambia, 1981." to Acute Respiratory Infections: Implications for Intervention." Journal (of Rev'eius of Infectious Diseases 5:391-95. Infectious Diseases 157(4):615-23. WHi) (World Health Organization). 1984. "A Programme for Controlling LINEP/WHO (United Nations Environment Program/WLorld Health Organiza- Acute Respiratoiry Infections in Children: Memorandum from a l.si oMeet- ton). 1988. "Global Environment Monitoring System: Assessment of ing." Bullen of the World Health Organiizaton 62:47- 58. Urban Air Quality." Geneva. . 1988. "Rheumatic Fever and Rheumatic Heart Disease." Technical UNICEF (United Nations Children's Fund) 1988. Essential Drugs Price List. Report Series, 764. Geneva. Copenhagen. - 1990. "Respiratory Infections in Children: Management at Small 1988b. Joint wHO/ULINICEF Nutntion Support Programme in Iringa, Tan- Hospitals." Draft. Geneva. zanua 1983-1988 Evaluation Report. Dar es Salaam: Government of the \WH0LARI (World Health Organization/Programme for Control of Acure Re- Republic of Tanzania, World Health Organization. spirators Infections). 1988. "Case Nianagemeist of Acute Respiratory Infec- -. 1991. The State of the World's Children. Oxford: Oxford University nion in Children: Intervention Studies." Publication 88-2. Geneva. Press for UNICEF. 1 I989b. "Programme for Control of Acute Respiratory Infections: USDHH.S (U. S. Department of Health and Human Services). 1989. "Reducing Programme Report, 1988." Publication 89-3. Geneva. the Consequences of Smoking: 25 Years of Progress." Publication (CDC89- 8411 Atlnta Pubic Halt Setce Cntes to DislseContol- 1990. "Programme fo)r the Control of Acute Respiratory Infections: 841 1. Atlanta: Public HeaLth Service, Centers for Disease Control. Case Management Resea,rch Priorities." Publication 90--1. Geneva. Vathanophas, K., R. Sangchai, S. Raktham. A. Pariyanonda, J. Thangsuvan, 1991a. "PrOgramme ((r Conrol sf Acute Respiratsry Infections: P. Bunyaratabhandu, S. Athipanyakom, S. Suwanjutha, P. Jayanetra, C. Interim Programme Report." Publication 91-19. Geneva. Wasi, M. Vorachit, and P'. Puthavathana. 1990. "A Community-Based fntr theowamoeRecrt.niendations o9 the Study of Acute Respiratory Tract Infection in Thai Children." Reviews of . 1991b. "Technical Bases for the WHO Re-onmendation on the Infectiows Diseases 12(Supplement 8):S957-65. Management of Pneumonia in Children at First-Level Health Facilities." Verma, l. C., and P. S. N. Menon. 1981. "Epidemiology of Acute Respirator' Publication 91-20. Geneva. Disease in North India." Indian Journal of Pediatrics 48:37-40. WHO/EPI (World Health Organization/Expanded Programme on Imminiza- Victora, Cesar G., P. G. Smith,). P. 'Vaughn, L. C. Nobre, C. Lombardi, A. tion). 1982. "The Optimal Age for Measles Vaccination." Weekly Epidenii- NI. B. Teixeira, S. NI. C. Fuchs, L. B. MNireiral, L. P. Gigante, ancd F. C. ologicRec(erd57:21-31. Barros. 1987. "Evidence for Protection by Brea.srfeeding against Infant . 1987. "Key Issues in Measles Immunization Research: A Review of Deaths from lnfecrtious Disease in Brazil." Latucet 2:319-21. the Literatuire." Report of the Global Advisory Grnup Meeting, November. Vijayaraghavan, K. G. Radhaiah, B. S. Prakasam, K. V. R. Sarma, and V. - 1990. "MeastesControl in the 1990s: Immunization before 9 Months Reddy. 1990. "Effect of Massive Dose Vitamin A ont Morbidity and Mortal- of Age." Publication 90-3. Geneva. ity in Indian Children." Lancet 336:1342-45. WHO/EPP (World Health Organization). 1984. "Biomass Fuel Consumption Vocorhoeve, A. M., A. S. Muller, T. W. 1. Shulpen, W. Manetje. and M. and Health." Geneva. Vangens. 1977. "The Epidemiology of Pertussis." Tropical and Geographic WHO/RSI' (World Health Organization). 1986. "Respiratory Infections in Medicine 30:121-24. Children: Management at Sniall Hospitals, Background Notes and a MNan- Wanger, J. D., Richard Pierce, Katharine A. DeaVer, Brian D. Phkavris, ual for Dictors." Publication 86-26. Geneva. Richard R. Facklam, Claire V. Bnrome, and the Haemophilus inJluenzae YUrdakik, K., T. Yavuz, and C. E. Taylor. 1990. "Swaddling and Acute Vaccine Efficacy Study Group. Lancet 338(8764): 395-98. Respiratorv Infections." Amencati Journal of Public Health 80(7):873-75. Ward, J., G. Brenneman, G. W. Letson, W. L. Hevward, and the Alaska H. Zinsicki, Susan. 1988. "L'enregistrement des causes de d6ces par des non- influenzae Vaccine Study Group. 1990. "Limited Efficacy of a Haernophilus m6decins: deux exprriences auL Bangladesh." In J. S. Vallin and A. Palloni, influenzae Type B Conjugate Vaccine in Alaska Native Infants." New eds., Mesure et analyse de la morralit&: nouvelles approches. Paris: Presses EnglandJournal of Medicine 323(20): 1393-1401. Universitaires de France. 5 Diarrheal Diseases Jose Martines, Margaret Phillips, and Richard G. A. Feachem Diarrhea is a complex of symptoms and signs, usually defined as malaria, measles, and respiratory infections and is a promi- as an increased number of stools of liquid or semiliquid consis- nent feature in acquired immunodeficiency syndrome (A1IDS). tency passed during a twenty-four-hour period. Although a The pathogenesis of infectious diarrhea has been exten- considerable variety of definitions can be found in the litera- sively studied, and the primary pathogenic mechanisms are ture (Snyder and Merson 1982), recent studies have tended to summarized in figure 5-1. Some of these pathogenic processes, consider more than three stools passed in twenty-four hours of especially those relating to the production and action of the observation as an indication of diarrhea after the age of three enterotoxins of E. coli and V. cholerae, are understood in great months (WHO 1988c). The word "diarrhea" is also used in detall. This understanding has opened the way to the produc- programmatic and public health contexts, although not typi- tion of sophisticated vaccines, including the live, genetically cally in clinical contexts, to embrace dysentery. Dysentery is engineered, vaccine strains. usually characterized by the presence of blood in the stools, The risk of diarrheal morbidity and mortality is greater with or without excessive looseness or frequency. among families of lower socioeconomic status and in condi- Considerable differences of opiniion still exist as to the tions of poor personal and domestic hygiene. Low family minimum number of healthy days that marks the end of a income (Manderson 1981; Stanton and Clemens 1987b), lack diarrheal episode, ranging from forty-eight hours (wHo 1988c) of luxury items (Huttly and others 1987), living in a one-room to two weeks (Scrimshiaw and others 1961). Episodes tend to house (Stanton and Clemens 1987b), living in a house with be self-limiting, generally lasting for less than seven days. an earthen floor (Bertrand and Walmus 1983), lower maternal Studies in several developing countries have shown that 3 to education (Bertrand and Walmus 1983), lower occupational 20 percent of acute diarrheal episodes in children under five status of the head of the family (Islam, Bhuiya, and Yunus may become persistent, lasting for at least fourteen days (WHO 1984), and unclean living conditions (Bertrand and Walmus 1988b). 1983; Huttly and others 1987; Taylor and others 1986) have A variety of enteric pathogens-including bacteria such as all been associated with increased risk of diarrheal morbidity enterotoxigenic Escherichia coli, enteropathogenic E. coli, Sal- and mortality. monella, Shigella, Vibrio cholerae, and Campylobacter jejuni; vi- The risk of diarrheal morbidity and mortality is higher ruses such as rotavirus; and protozoa such as Giardia lamblia, among infants who are not breastfed (Feachem and Koblinsky Entamoeba hystolytica, and Cryptosporidium-can cause diar- 1984). More recent studies in developitig countries have con- rhea (Farthing and Keusch 1989). The specific pathogens of firmed the very substantial role of breastfeeding in protecting greatest importance to public health vary according to age of infants against diarrheal incidence, severity, and mortality the patient and the geographical setting. For example, (Briend, Wojtyniak, and Rowland 1988; Clemens and others rotavirus is a significant cause of severe diarrhea in children 1986; Huttly and others 1987; Mahmood, Feachem, and under two years of age in developing countries, and Salmotwlla Huttly 1989; Martines 1988; Victora and others 1987). Several and Campylobacter are primary causes among adults in indus- studies indicate a risk of increased diarrheal duration and trialized countries who consume poultry raised in factory-farm- severity among the malnourished (Bairagi and others 1987; ing conditions. Black, Brown, and Becker 1984a; Black and others 1984; Other causes of diarrhea, considered to be relatively unim- Chen, Huq, and Huffman 1981; Palmer and others 1976). portant in developing countries, include food intolerance, Recent studies in Sudan and Mexico have suggested that especially lactase deficiency and allergies to animal protein, malnutrition also increases the risk of frequent diarrheal epi- granulomatous diseases of the gut, and tumors elaborating sodes (Samani, Willet, and Ware 1988; Sep6lveda, Willet, and gastrointestinal hormones (Rohde 1986). Diarrhea may also Munioz 1988). Although information is scarce on the role of be associated with infections outside the intestinal tract, such low birth weight as a determinant of diarrheal morbidity (Ash- 91 92 Jose Martines, Margaret Phillips and Richard G. A. Feachem Figure 5-1. Primary Pathogenic Mechanisms attributed to the severity or duration of diarrheal episodes, the for Infectious Diarrheal Diseases nutritional effectiveness of case management, and the extent of catch-up growth (Black, Brown, and Becker 1983; Brown and others 1985; Keusch and Scrimshaw 1986; Khan and Ingestion of Ahmad 1986; Miall, Desai, and Standard 1970; United Na- microorganisms tions University 1979; Whitehead 1977). 1 Despite diarrheal incidence and severity being lower among Survival in gastric acid adults, diarrhea may still represent a significant portion of total adult morbidity in developing countries. In Nigeria, diarrhea was associated with 20 percent of all illnesses in adults (Huttly and others 1987) and accounted for a median of 23 percent Viruses Bactena (range 5-41 percent) of disability days per year among adults I 1 age fifteen through forty-four in Pakistan, Indonesia, Nepal, Adhesion to and Ghana (Grosse 1980). In rural Bangladesh, diarrhea ac- Mucosal invasion Adhemucosa counted for 14 percent of the total mortality among the pop- -r- I ulation over forty-five years old surveyed between 1980 and Destruction of jejunal P 1983 (BRAC, 1987). mucosal cellsPrlfato The Public Health Significance of Diarrhea The current levels and trends of diarrhea morbidity and mor- loss rdasaccsa glucose nt nson Other tality in the developing world, and the probable pattems of transportproductin morbidity and mortality in the next century are examined in this section. Source: Candy and Phillips 1986. Current Levels and Trends in the Developing World worth and Feachem 1985b), the association between intra- Information on the current levels and trends in the incidence uterine growth retardation and impaired immunocompetence, and mortality due to diarrhea is reviewed in the following and the strong association between low birth weight and paragraphs. diarrheal mortality in infancy in developing countries (Barros and others 1987), suggest that low birth weight is a risk factor MORBIDITY. Measurements ofdiarrheal incidence depend on for diarrheal severity and mortality. such factors as what definition of diarrhea is used, the fre- Diarrhea can lead to dehydration and early death, particu- quency of surveillance, and the location of the study popula- larly in young children (Gordon and others 1968; Snyder and tion (Snyder and Merson 1982). Since 1981, the Diarrhoeal Merson 1982), and to impaired growth and nutritional status Diseases Control Programme of the World Health Organiza- among the survivors. The effect of diarrhea on infant and child tion (WHO/CDD) has promoted a series of surveys to measure growth and nutritional status is the outcome of a complex diarrheal morbidity and mortality rates using a consistent interplay of host, pathogen, and sociocultural factors, which methodology (wybo 1987b, 1988c). Results from 276 surveys may cause decreased food intake (Briscoe 1979; Brown and conducted in sixty countries between 1981 and 1986 are others 1985; Hoyle, Yunus, and Chen 1980; Khan and Ahmad summarized in table 5-1. They indicate that, on average, 1986; Martorell and others 1980; Mata and others 1977; children under five years of age in developing countries suffer Nabarro and others 1988), malabsorption (Chen 1983; Molla, 3.5 episodes of diarrhea per year. Excluding the Americas, Molla, and Khatun 1986; Rosenberg, Solomons, and Schnei- where relatively few surveys have been undertaken, the high- der 1977; Scrimshaw 1977), loss of endogenous nutrients est rates were found in Sub-Saharan Africa and the lowest in (Chen 1983; Sarker and others 1986), and increased metabolic Asia, especially in China and other countries in the Westem rate (Beisel 1977; Chen 1983; Keusch and Katz 1979; Keusch Pacific region. and Scrimshaw 1986). In order to estimate diarrheal morbidity rates among older Longitudinal studies have detected a significant effect of age groups, for which information is scarcer, the estimates of diarrheal episodes on the growth of infants and children relative incidence of diarrheal episodes in the older age groups (Bairagi and others 1987; Black, Brown, and Becker 1984b; adopted by the Committee on Issues and Priorities for New Cole and Parkin 1977; Condon-Paoloni and others 1977; Vaccine Development of the Institute of Medicine of the Guerrant and others 1983; Martorell and others 1975; United States (U.S. Institute of Medicine 1986) were applied Martorell and others 1977; Nabarro and others 1988; Rowland, to the WHO/CDD morbidity rates. Diarrheal incidence was esti- Rowland, and Cole 1988; Zumrawi, Dimond, and Waterlow mated to be four to six times lower for those between five and 1987). The wide range of effects recorded have been variously fourteen years old, and thirteen to sixteen times lower for those Diarrheal Diseases 93 Table 5-1. Morbidity in 276 Surveys in Children months and 2.5 episodes per year among adults). Diarrheal Aged Four Years and Younger Using the WHO/CDD rates described by Georges and others (1987) in the Central Methodology, 1981-86 AfricanRepublic (7episodesperchildperyearamongchildren Annual incidence under five), although higher than the average of those pre- (episodeslchild/year)a sented forthe region, were still within the range ofthe WHO/COD Region Surveys Countries Median Range survey findings from which the presented values were derived. A lower incidence (1.9 episodes per child per year)-but also Latin America and the Carribean 1 2 8 4.9 0.8-10.4 within the range measured in the WIIO/CDD surveys-was de- Sub-Saharan Af1ca 67 22 4.4 1.6-9.9 scribed among children under four years of age in rural Ghana Middle East and (Biritwun and others 1986). Kirkwood (1991) provides a com- North Africa 47 10 2.7 2.1-10.8 prehensive review of knowledge of diarrhea epidemiology in Asia and the Pacific 150 20 2.6 1.1-5.7 Sub-Saharan Africa. Incidence rates presented for North India - 1 2.7 Africa and the Middle East are similar to those described China 1 1.2 by el-Alamy and others (1986) among adults and children in Other - 1 2.6 rrlEyt All regions 276 60 3.5 0.8-10.8 rural Egypt. Recently the epidemiology of persistent diarrhea (diarrhea - Not available. lasting at least fourteen days) has been the focus of particular Note: Surveys were conducted mainly in geographically limited areas. If investigation. In rural norther India, Bhan and others (1989) more than one survey was conducted in any country, the median for the country was used to) calculate the regional and global medians- report a low incidence of persistent diarrhea of 0.06 episodes a. Survey estimate was adjusted for seasonality when appropriate data per child per year in children under six years of age, with a peak were available. o 0 e p l p Source: Unpublished World Health Organization data. Incidence of 0.3 episodes per chld per year m iants. Persis- tent episodes made up 10 percent of all episodes in children under six years and 22 percent of all episodes in infants under beyond the age of fourteen years, than that observed among one year. In rural Bangladesh, Huttly and others (1989) report children under five (table 5-2). an incidence of persistent diarrhea of 0.6 episodes per child per Independent longitudinal studies largely confirm these esti- year in children under five years of age (16 percent of all mates of diarrhea incidence in Latin America presented in episodes). In the first half of infancy, persistent episodes com- table 5-2. In a poor periurban population in the north of Brazil, prised one-quarter of all episodes. Giugliano and others(1986) recorded 4.8 episodes per yearfor No evidence could be found in the literature of a fall children from birth to thirty-five months and 0.2 episodes per in incidence rates of diarrhea during the past fifteen years. year for adults. Guerrant and others (1983), studying a poor Snyder and Merson (1982), using active surveillance studies urban group in northeast Brazil, found similar rates for young conducted between 1954 and 1979, estimated that diarrheal childrenbutsubstantiallyhigheronesforadults,forwhomthey incidence in developing countries averaged 2.2 episodes estimated more than one episode per adult per year. per child per year. Comparing this value with the 3.5 epi- Diarrheal rates presented forSub-Saharan Africa (table 5-2) sodes per child per year derived from cross-sectional survey are substantially lower than those estimated from cross- data collected between 1981 and 1986, we must conclude that sectional surveys by Huttly and others (1987) in Nigeria (14 it is unlikely that any significant reduction in diarrheal inci- episodes per year among children six months through eleven dence has occurred in the period between 1970 and 1985, particularly as cross-sectional surveys are likely to underesti- Table 5-2. Estimated Annual Diarrheal Morbidity mate morbidity rates. Rates Region and Age, 1985 IMORTALITY. Available data on diarrheal mortality tend to be Age (years) less reliable than those on morbidity. The only global data Region 0-4 5-14 15-59 > 59 source available was generated during the WHO/CDD morbidity Latin America and and mortality surveys (table 5-3). Some of the information on the Carribean 4.9 1.2 0.3 0.3 mortality was considered implausible by WHO, a situation that Sub-Saharan Africa 4.4 0.9 0.3 0.3 was attributed to lack of training or supervision of the surveyors Middle East and (WHO 1988c). In addition, the survey method used, the inter- North Africa 2.7 0.5 0.2 0.2 viewing of heads of households, who were asked to recall the Asia and the Pacific 2.6 0.4 0.2 0.2 deaths of family members in the past year, is regarded skepti- All regions (median) 3.5 0.7 0.2 0.2 cally by some demographers and may greatly underestimate Note: Rates were calculated as episodes per person per year. Rates for chil- mortality rates (Timaeus and Graham 1989). The information dren aged four years and younger were derived from CmD mobidity survey in table 5-4, although the best presently available, should be (table 5-I ). Other rates were estimated according to relative incidence rates by age adopted by the Institute of Medicine 1986. regarded with caution. Estimates of diarrhea case-fatality rates Source: Unpublished World Health Organization data. Rates were calcu- by age that were made by the U.S. Institute of Medicine (1986) lated by the authors. were used to calculate mortality rates of those beyond the age 94 Jos6 Martines, Margaret Phdilips, and Richard G. A. Feachem Table 5-3. Mortality in 276 Surveys in Children Aged Four Years and Younger, Using WHO/CDD Methodology, 1981-86 Diarrhea Mortality Rate Diarrhea deaths as (deaths/I ,000 children/year) percentage of totai Region Surveys Countries MkiedianT Ranige deaths (median) Latin America and the Carribean 1 2 8 4.2 1.2-9.2 35 Sub-Saharan Africa 67 22 10.6 3.1-54.9 38 Middle East and North Africa 47 10 5.8 1.0-25.3 39 Asia and the Pacific 150 20 3.2 0.0-17.2 29 India - 1 3.2 - -- China - I 0.0 Other - 18 3.3 - All regions 276 60 6.5 0.0-54.9 36 - Not available. Note: Stirvevs were conducted mainly in geographically limited areas If more than one survey was conducted In any country, the medcian f(tr the country was used to calculaie the regional and global medians. Source: Unpuhlished Vlorld Health Organization dat,i. of four years shown in the table. Case-fatality rates were The data from Egypt and Costa Rica indicate that this reduc- estimated to fall from 2 per 1,000 among children under five, tion in diarrheal mortality rates has been accompanied by to 0.4 per 1,000 among those between five and fourreen reductions in general childhood mortality rates during the years and 0.3 per I ,000 among persons between fifteen and same period, in a way that suggests that diarrhea has not been fifty-nine years. A slight increase, to 0.5 per 1,000, was esti- replaced by other causes of death. mated for members of the population age sixty years and older. Diarrheal mortality in children under five was calculated by Estimates of adult mortality due to diarrhea for developing Snyder and Merson (1982) to be 13.6 deaths per 1,000 from countries with adequate vital registration systems are reported data collected between 1956 and 1979, whereas a median by Feachem and others (1992). Estimates of the risk of death value of 6.5 deaths per 1,000 was estimated by the WHO/CDI) from diarrhea between ages fifteen and fifty-nine vary from morbidity-mortalitystirveys (table 5-3). These latterdata may, 0.01 percent (for example, in Argentina) to 3 percent in as explained above, be substantial underestimates. The case- Guatemala. The proportion of deaths due to diarrhea was fatality rates in children tinder five years old used by the U.S. reported to be roughly 4 percent in India in persons aged Institute of Medicine (1986) increases the mortality rate to 7.1 fifteen through fifty-four years and 2 percent in Kenya in per I,0QQ,,sti148percentlowerthanthatcalculatedbySnyder persons aged fifteen through sixty-four years (Feachem and and Merson (1982) for the period 1956-79. This fall in diar- others 1992). rheal mortality can be attributed reasonably to a fall in the L)iarrheal mortality appears to have fallen substantially in case-fatality rate during the time. CLrrent case-fatality rates many areas during the past fifteen years. Indications of this vary widely: for children under five years they were found to trend are reported from Egypt (el-Rafie and others 1990; be 0.8 per 1,000 in Bangui, Central African Republic (Georges Egypt, National Control of Diarrheal Diseases Project 1988), and others 1987), 3 per 1,000 in rural Egypt (el-Alamy and Thailand (Phonboon and others 1986), Brazil (Benicio) and others 1986), 3.6 per 1,000 in rural northern India (Kumar, others 1987; Monteiro and Benicio 1989), Cuba (Riveron- Kumar, and Dutta 1987), and 5.4 per 1,000 in rural Indonesia Corteguera and Mtuioz 1982) and Costa Rica (Mata 1981). (Nazir, Pardede, and Ismail 1985). Nevertheless, these values Table 5-4. Estimated Annual Diarrhea Mortality Rates, by Region and Age, 1986 Age (years) _ ____ Region 0-41 5-14 15-59 >59 Latin America and the Carribean 4.2 0.5 0.1 0.1 Sub-Saharan Africa 10.6 0.4 0.1 0.1 Middle East and North Africa 5.8 0.2 0.1 0.1 Asia and the Pacific 3.2 0.2 0.1 0.1 All regions (median) 6.5 0.3 0.1 Q0.1 Note: Rates were calculated as deaths per I ,000 persons per year. Rates for children aged four and youinger were derived from (ruu mortality survey.> (table 5-3). Other rates were estimated according to case-fatality rates by age adopted by the Institute of Medicine 1986. a. Mortality rates (per 1,000 children) estimated according to the Institute of Medicine's ( 1986) case-fatality ratios are as follows: Lattin America, 9.8; Sub- Saharani Africa, 8.8; North Africa, 5.4; Asia, 5.2. Source Unpublished World Health Organization data. Ratios were calcutated by the authors. Diarrheal Daseases 95 are all lower than the 6 per 1,000 estimated by Snyder and rehydration salts (ORS) has increased from less than 10 percent Merson (1982) from data collected before 1980. in 1982 to 51 percent in 1985 (WHO 1988c) and to 63 percent In summary, therefore, there is evidence of a fall in diarrheal in 1989 (WHO 1991), nearly 40 percent of the population mortality rates in children in recent years, but little evidence remain uncovered. The level of mortality decrease that can be of a parallel fall in incidence rates. This implies, on the one expected from expandingoRT programs will depend on the risk hand, falling case-fatality rates and, on the other, stagnating of diarrheal death from dehydration among those as yet with- levels of the key risk factors for diarrhea morbidity. Central out access to ORS and on the potential for improvement in the questions for policy that emerge are, first, to what extent the quality of case management required to make a difference. declining case-fatality rates have been created by the interven- Boosting coverage from current levels and, even more so, tions of national cirDr programs and, second, whether there are improving the correctness ofORT use, may be difficult. The CDD reasonable prospects for reducing morbidity over the next program has recently instituted a new applied research initia- decade. We returm to these questions later in this chapter. tive to deal with these issues and has intensified efforts to train health care providers (doctors, nurses, paramedical staff, phar- Likely Morbidity and Mortality Patterns-Next Century macists, traditional healers) in diarrhea case management (WHC 1989). This will require, nevertheless, that governments Although risk factors for diarrhea morbidity and mortality or agencies be prepared to devote the necessary financial and have been identified, and in some cases quantified, and the manpower resources for it. effects of certain interventions to control diarrheal diseases Improvement in the rates of use of ORT is unlikely to reduce measured, only very general predictions of the likely future mortality due to persistent or dysenteric diarrhea. Better over- pattern of diarrheal morbidity and mortality are possible. In all case management is necessary and may require new meth- the first place, explorations of many of the relationships be- ods, with special attention to feeding during the episode and tween risk factors and diarrhea outcome have been conducted recovery. The same points apply to complicated diarrheas, at a relatively simplistic level-ignoring, for example, interac- such as diarrheas accompanied by measles, acute respiratory tions between various causes-and in relatively few locations. infections, or malaria. These complicated diarrheas represent Second, it is not clear how or how much various factors a considerable proportion of all inpatient diarrhea cases in influencing diarrhea morbidity or mortality will change in the some hospitals; the more so since the establishment of ORT- future. Evidence of past trends suggests, although only weakly, based outpatient facilities for the treatment of uncomplicated that morbidity levels have remained stagnant over the last acure watery diarrheas. fifteen years (see previous section), despite changes at the global level in several of the risk factors, such as substantial Summary of the Public Health Significance of Diarrhea increases in measles immunization coverage and more modest increases in the proportion of households having improved Despite developments in the last decades in the understanding water supply and sanitation. These global trends, however, of the etiology and pathogenesis of diarrheal diseases and the disguise what are probably considerable variations within re- discovery of an effective oral rehydration solution to treat the gions and countries, and not all these risk factors have moved majority of the patients with watery diarrhea, morbidity rates in a direction associated with reduictions in diarrhea rates. For among young children are still high, and diarrhea remains a example, breastfeeding rates in rural areas in developing coun- significant cause of adult illness. In 1985, nearly 3 billion tries have tended to decline and are unlikely to have increased diarrheal episodes are estimated to have occurred in the devel- significantly during the period in urban areas (Popkin and oping world, leading to the death of 5 million persons, 80 Bisgrove 1988), and nutritional status has remained stagnant percent of them under five years of age (see Lopez, chapter 2, in many countries, especially in Sub-Saharan Africa (Ash- this collection). Although evidence exists of a reduction in worth and Dowler 1991). diarrheal mortality during the past fifteen years, no indication Nevertheless, diarrheal mortality rates in children have has been found of reductions in diarrheal incidence during the fallen in the last fifteen years. In the same period, overall infant same period. This lack of decline points to the need for the and child mortality rates have also fallen throughout the development, implementation, and evaluation of effective developing world for reasons that are not entirely understood measures to lower diarrheal morbidity. (Feachem, Timaeus, and Graham 1989). It is not clear how much of the diarrhea mortality decline has been brought about Lowering Disease Incidence by increased use of oral rehydration therapy (ORT); how much by general improvement, access to, and quality of health care The promotion of ORT for the case management of acute services at the periphery; and how much by nonspecific factors diarrheal diseases may have contributed to the reduction of such as improving parental education (leading to more prompt diarrheal mortality rates observed in developing countries in attention to worsening symptoms of a sick child). the last fifteen years. Still, ORT cannot be expected to have a Present coverage rates and estimated use rates suggest that significant effect on the incidence of diarrhea. If morbidity there is potential for mortality to decline through more effec- rates are to be reduced, primary preventive strategies must be tive delivery of cDD programs. Although global access to oral identified and implemented. 96 Jose Martines, Margaret Phllips, and Richard G. A. Feacheni Elements of a Preventive Strategy * Rotavirus immunization (when an effective vaccine be- comes available) A systematic study of interventions, excluding case manage- * Cholera immunization (in selected countries, when a ment, that might play a role in diarrhea control was initiated more effective new vaccine becomes available) in 1982 by the Diarrhoeal Diseases Control Programme of WHO * Measles immunization (Feachem, Hogan, and Merson 1983). Eighteen interventions * Improvement of water supply and sanitation facilities were evaluated, with particular emphasis on their effectiveness a Promotion of personal and domestic hygiene and feasibility (table 5-5). Of these, four interventions-im- proving lactation, supplementary feeding programs, chemo- These interventions, except for improved water and sanita- prophylaxis, fly control-were found to be either ineffective tion, were the object of more detailed analysis of cost-effec- or too costly for incorporation in programs ofdiarrheal diseases tiveness (Phillips, Feachem, and Mills 1987). Information control in developing countries. Seven other interventions- regarding the potential effectiveness and cost-effectiveness of prevention of low birth weight, use of growth charts, increase these interventions is summarized in tahle 5-6. in child spacing, vitamin A supplementation, improvement of food hygiene, control of zoonotic reservoirs, epidemic con- PROMOTION OF BREASTFEEDING. The hypothesis that breastfed trol-were considered of uncertain effectiveness, feasibility, or infants may have a reduced risk of diarrheal morbidity and cost and required further research before their potential role mortality is supported by evidence indicating the immunologi- could be properly assessed. Two of these potential interven- cal and antimicrobial properties of breast milk, its contribution tions-increase in child spacing and control of zoonotic reser- to good nutritional status during the first few months of life, voirs-are still under review. and the risk associated with contaminated feeds. The literature Seven interventions were identified for which evidence of on the risks of diarrheal morbidity and mortality among chil- adequate effectiveness and feasibility were reasonably strong: dren, according to mode of feeding, was reviewed by Feachem Promotion of breastfeeding and Koblinsky (1984). More recent studies, in which con- founding was carefully controlled, have shown even higher Improvement in weaning practices relative risks of severe diarrhea and diarrhea mortality associ- Table 5-5. Potential Nonclinical Interventions for Control of Diarrhea among Young Children Strategy Intertvention Reference' Matemal Health Preventing low birth weight Ashworth and Feachem 1985 Improving Lactation Ashworth and Feachem 1985a Child health Promoting breastfeeding Feachem and Koblinsky 1984 Improving weaning practices Ashworth and Feachem 1985c Supplementary feeding programs Feachem 1983 Using growth charts Ashworth and Feachem 1986 Increasing child spacing n.a. Vitamin A supplementation Feachem 1987 Immunization and Rotavirus immunization de Zoysa and Feachem 1985 chemoprophylaxis Cholera immunization de Zoysa and Feachem 1985 Measles immunization Feachem and Koblinsky 1983 Chemoprophylaxis de Zoysa and Feachem 1985 Interrupting transmission Improving water supply and Esrey, Feachem, and Hughes, 1985 sanitation facilities Promoting personal and Feachem 1984 domestic hygiene Improvtng food hygiene Esrey and Feachem 1989 Controlling zoonotic reservoirs n.a. Controlling flies Esrey 1991 Epidemic control Epidemic surveillance, n.a. investigation, and control n.a. Not applicable. Note: The general rational for the review series is described in Feachem, Hogan, and Merson, 1983. Policy conclusions are discussed in Feachem. 1986. Cost-effectiveness analysis of selected interventions is presented in Phillips, Feachem, and Mills, 1987. a. Some of these publications are also available in French and Spanish. Source: See table (last column) and note above. Diarrheal Diseases 97 ated with nonbreastfeeding. For example, in Basrah, Iraq, for Breastfeeding promotion projects have been shown to be the nonbreastfed infant the risk of hospitalized diarrhea was effective in increasing the number of women initiating thirty-seven times greater than for the exclusively breastfed breastfeeding and the total duration of exclusive or partial infant in the first three months of life and twenty-four times breastfeeding (Hardy and others 1982; Huffman and Combest greater in the second three months (Mahmood, Feachem, and 1988; Winikoff and Baer 1980). Methods of breastfeeding Huttly 1989). In Brazil in 1985, the risk of diarrheal mortality promotion have included the training and education of health was found to be fourteen times greater for nonbreastfed infants professionals; the changing of hospital practices to facilitate than for exclusively breastfed infants in the first year of life. early suckling following birth, rooming-in, limited supplemen- The risk of infant mortality was highest among the non- tary and glucose-water feeds, and restrictions on the distribu- breastfed during the first two months of life, being twenty- tion of formula samples; and the provision of assistance to three times greater than among the exclusively breastfed breasrfeeding support groups. Large-scale projects in Brazil, (Victora and others 1987). No evidence could be found by Honduras, Indonesia, Panama, andThailand were reviewed by Feachem and Koblinsky (1984) that indicated that protection Huffman and Combest (1988). Although few adequately de- against diarrhea extended after the termination of breastfeed- signed evaluations have been carried out, the results point ing (a conclusion supported by Mahmood, Feachem, and consistently to the success of these promotional programs in Huttly 1989) or after the first year of life. Recent studies have decreasing rates of nonbreastfeeding. suggested, however, significant protection from breastfeeding Theoretical calculations based on the above data (Feachem against the severity of diarrhea even during the third year of and Koblinsky 1984) indicate that promotion of breastfeeding life (Briend, Wojtyniak, and Rowland 1988; Clemens and can reduce diarrhea morbidity by 8 to 20 percent in the first others 1986). six months of life and by I to 4 percent for children under five Table 5-6. Effectiveness and Cost-Effectiveness of Interventions for Diarrhea Control among Young Children Cost-effectivenessh Strategy Intervention Effectiveness' (percent) range (median) (1982 U.S. dollars) Child health Promoting breastfeeding Reduction in nonbreastfeeding 0-2 months: 40 3-5 months: 30 6-11 months: 10 Reduction in diarrhea morbidity 10-75 45 0-6 months: 8-20 0-59 months: 1-4 Reduction in diarrhea mortality 400-10,750 1000 0-6 months: 24-27 0-59 months: 8-9 Improvingweaningpractices Reductioninpercentageofchildren 50- 2000 1070 < 75 percent weight-for-age 6-59 months: 50 Reduction in diarrhea mortaliuy 6-59 months: 2-12 Immunization Rotavirus immunization Reduction in diarrhea incidence: 4 3-30 5 Reduction in diarrhea mortality: 13 140-1400 220 Cholera immunization Reduction in diarrhea incidence: 0.2 90-1450 174 Reduction in diarrhea mortality: 2.8 1,075-16,710 2,000 Measles immunization Reduction in diarrhea incidence: 3 3-60 7 Reduction in diarrhea mortality: 22 66-1,156 143 Interrrupting Improving water supply Reduction in diarrhea incidence: 27 No meaningful estimates; transmission and sanitation Reduction in diarrhea mortality: 30 multiple benefits Promoting personal and Reduction in diarrhea incidence: 14-48 5-500 10 domestic hygiene - Not available. a. For children from hirth to age fifty-nine months, unless otherwise specified. b. Only considers diarrhea deaths or episodes averted in children under age five vears. c. Assumes 1lO percent coverage with 80 percent vaccine efficiency. d. Assumes 100 percent coverage with 70 percent vaccine efficiency. e. In Bangladesh. f. Assumes 100 percent co verage with 85 percent vaccine efficiency. Source See table 5-5. 98 Jose Martines, Margaret Phillips, and Richard G. A. Feachem years. Mortality rates can be reduced by 24 to 27 percent in the and Becker 1984a; Chen, Huq, and Huffman 1981; Palmer and first six months, and by 8 to 9 percent for children under five others 1976). Studies in Nigeria (Tomkins 1981), Mexico years. Results from more recent studies indicate that these (Sepulveda, Willet, and Mufnoz 1988), and the Republic of the calculations are conservative, especially with reference to the Sudan (Samani, Willet, and Ware 1988) suggest that malnour- prevention of diarrheal mortality, and that breastfeeding pro- ished children may also experience increased incidence of vides greater and more extended protection against diarrheal diarrhea. The effect of nutritional status on the severity (de- deaths and severe diarrheal episodes, even up to the third year gree of dehydration and purging rate) of diarrhea has also been of life (Briend, Wojtyniak, and Rowland 1988; Clemens and investigated in clinical studies (Black and others 1981; Black others 1986; Mahmood, Feachem, and Huttly 1989; Martines and others 1984). Diarrheal mortality has also been shown to 1988; Victora and others 1987). be higher in malnourished children. Phillips, Feachem, and Mills (1987) constructed a range In Bangladesh, children who had diarrheal episodes and of plausible costs for a variety of different breastfeeding pro- whose weight-for-age was below 65 percent of the expected motion strategies (changes in hospital routine, face-to-face were 3.7 times more likely to die during the subsequent twenty- education, mass media campaigns, discouragement of four months than their better-nourished counterparts (Chen, breastfeeding alternatives, provision of facilities for working Chowdhury, and Huffman 1980). In northern India, case- women), using information on the characteristics of these fatality rates in children with diarrhea were significantly interventions, their likely resource requirements (principally greater among those who were severely malnourished (7.7 in terms of staff time), and salary estimates. Even with salaries percent) than among moderatel) malnourished (2.2 percent) set at a constant average level and the use of a standard input or well-nourished children (0.3 percent) (Bhan and others package (except for mass media campaigns), the cost per child 1986). affected varied considerably within each strategy, depending Poor weaning practices can be the result of lack of access to on such factors as scale (hospital size in the case of hospital- appropriate foods. Such practices, however, also appear to based interventions; population size in the case of mass media derive from food taboos or ignorance and to be potentially campaigns) and on assumptions concerning the coverage of susceptible, therefore, to education. Few weaning education the intervention and the degree of secondary effect. Phillips, programs have been evaluated in terms of improvement pro- Feachem, and Mills concluded that, with a judicious selection duced in nutritional status (Ashworth and Feachem 1985c). of interventions (excluding the relatively costly strategies in- The information available suggests that a weaning education volving the provision of facilities for working women), it program can halve the proportion ofchildren who are less than should be possible in most countries to provide a breastfeeding 75 percent weight-for-age, children who are likely to have a promotion package for about $5 per infant exposed.' Using this diarrheal mortality rate twice as high as other children (Ash- estimate and adopting conservative estimates of effectiveness worth and Feachem 1985c). If the effect of weaning education based on Feachem and Koblinsky (1984), we estimated the is manifest only during the weaning period, eighteen months cost-effectiveness of breastfeeding promotion to range be- to twenty-three months of age marking its upper limit, reduc- tween $10and $75 (median $45) perdiarrheal episode averted tions in diarrhea mortality in children under five years old of and between $400 and $10,750 (median $1,000) per diarrheal 2 to 8 percent may be expected when the prevalence rates of death averted in children under five years of age. moderate and severe malnutrition in the community, before the intervention, range from 10 to 50 percent of children. If IMPROVED WEANING PRACTICES. The hypothesis that under- the effect of weaning education can be extended to 59 months, weight children may be predisposed to diarrhea is supported by and indeed a high proportion of undemutrition up to this age evidence suggesting that protein-energy malnutrition causes maybe the result offailure tocatch up after the weaning period, impaired cellular (Chandra 1986) and secretory immune re- reductions of diarrheal mortality of 2 to 12 percent may be sponses (Sirisinha and others 1975). This hypothesis is further expected. According to mostofthe evidence available, the risk corroborated by evidence from patients with congenital or of diarrheal incidence is not likely to be reduced with im- acquired immunodeficiency syndromes, who have increased provetnents in the child's nutritional status, although more diarrlheal risk (Arbo and Santos 1987). Nonspecific protective recent studies challenge these conclusions (Samani, Willet, mechanisms, such as gastric acid production and intestinal and Ware 1988; Sepulveda, Willet, and Mufioz 1988). Im- mucosal renewal, may also be affected by malnutrition provements in food hygiene resulting from weaning education (Brunser and others 1968; Guiraldes and Hamilton 1981). are likely to increase the effect of the intervention, reducing Increased risk of diarrhea has been described in hypochlorhyd- diarrheal incidence and, possibly, the severity of episodes ric individuals (Gianella, Broitman, and Zamcheck 1973), and (Esrey and Feachem 1989). recovery from diarrhea may be delayed by an impaired mucosal From the few available data on the cost of implementing renewal (Butzner and others 1985). weaning education programs, Phillips, Feachem, and Mills Epidemiological studies conducted in Bangladesh indicate (1987) found that weaning education activities have been that protein-energy malnutrition is a significant determinant mounted for between $0.50 and $10.00 per child benefiting, ofdiarrhea severity and duration in infancy and childhood, but though these estimates are very sensitive to assumptions con- not of its incidence (Bairagi and others 1987; Black, Brown, cerning the number ofchildren benefiting and the comprehen- Diarrheal Diseases 99 siveness of the costings. The cost to the family, in the form of CHOLERA IMMlUNIZATION. Intestinal infections with Vibrio purchase and preparation of food, which may be substantial, is cholerae 01 are associated with especially severe diarrheal dis- not included in these estimates. Using effectiveness data from ease that occurs in both epidemic and endemic form (Feachem Ashworth and Feachem (1985c), the cost-effectiveness of 1981; Miller, Feachem, and Draser 1984). In recent years, weaning education was estimated to range between $50 and strains resistant to multiple antibiotics have been described in $2,000 per diarrhea death averted, with a median estimate of Bangladesh and East Africa. $1,070. The cholera vaccines that are available at present need to be given parenterally and are of low efficacy (WHO 1986c). ROTAVIRUS IMMUNIZATION. Rotavirus is the most frequent Current research on cholera vaccines is directed toward the etiological agent isolated in children under two years who development of an effective oral vaccine that uses either attend treatment centers for diarrheal diseases in developing nonliving bacteria and purified bacterial antigens or living countries (Black and others 1980; Levine and others 1986; avirulent mutants or genetically engineered strains. The hope Mata and others 1983; Stoll and others 1982). Rotavirus is that these will stimulate the substantial immunity that is vaccines can thus be expected to have a role in reducing known to follow clinical cholera (WHO 1986c). Field trials of diarrhealincidenceandmortalityamongyoungchildreninthe new nonliving oral vaccines in Bangladesh indicated overall developing world (de Zoysa and Feachein 1985b). protection for three years of 50 percent-but only 23 to 26 No vaccine is currently available for full-scale application. percent protection for children under six years (WHO 1989). Attempts are under way to develop live, attenuated, oral The field is controversial and fast moving, and it remains vaccines against disease caused by human rotavirus (Flores and unclear which type of vaccine for widespread application will Kapikian 1989) and reassortant rotaviruses that incorporate in emerge from the field trials (Levine 1989). an animal rotavirus a gene segment encoding for the VP7 The administration schedule for the new cholera vaccines surface antigen of human rotavirus serotypes 1,2, and 4. More is still tentative. Two or three oral doses in the second year of recently, the naturally attenuated human "nursery strain" life is one possible schedule. The vaccine is likely to be admin- rotavirus has also been considered for use as an oral vaccine. istered at a new contact and at an age when healthy children Field trials conducted to date have shown that vaccination are not normally brought to health services, which may in- with live rotavirus vaccines, in industrial countries, can reduce crease costs and decrease coverage. De Zoysa and Feachem the incidence of clinically significant rotavirus diarrhea by (1985b) investigated the potential of cholera vaccination in nearly 80 percent. Vaccine efficacy in developing countries, Bangladesh, which has relatively high rates of endemic chol- for reasons that are nor fully understood, has been much lower. era. For a cholera vaccine of 70 percent efficacy and an average The level of protection against all rotavirus diarrhea has in age at full vaccination of two years, 0.21 percent of diarrhea most studies been lower than that against severe rotavirLs episodes and 2.8 percent of diarrheal deaths in vaccinated diarrhea (de Zoysa and Vesikari 1990). children under five years could be averted. In countries that Exploring the potential of rotavirus vaccines, de Zoysa and had a lower proportion of diarrhea episodes and deaths caused Feachem (1985b) suggested that a rotavirus vaccine of 80 by cholera, the effects would be less. percent efficacy given to children whose average age at full Phillips, Feachem, and Mills (1987) calculated costs using vaccination was six months could achieve reductions of 4 the methodology described for rotavirus and assuming a vacci- percent in diarrheal incidence and 13 percent in diarrheal nation schedule of three doses given in the second year of life, deaths among children under five. a constant dropout rate between each contact of 10 percent, On the basis of actual cost data available for several national and a vaccine cost per dose similar to that of measles ($0.20). vaccination programs conducted in nine developing countries, The results suggested that the cost per child fully vaccinated and taking into account various features of the vaccines used with a new oral cholera vaccine would range from $1.70 to in those programs (whether orally administered or injected, $27.00, with a median value of $3.20. The cost per diarrhea with another vaccine at the same visit or in the same injection, episode and diarrhea death averted through cholera vaccina- and the number ofdoses), Phillips, Feachem, and Mills (1987) tions, assuming 70 percent efficacy of the vaccine and cost of calculated an average cost per "injected dose equivalent." $4.00 per child fully vaccinated, was calculated to be $1 74.00 They then applied this cost unit to the rotavirus vaccine (as per diarrhea episode and $2,000.00 per diarrhea death averted well as to the measles and cholera vaccines discussed in the in children under five years of age in Bangladesh. This esti- following sections). Taking account of the relevant character- mate, however, relates only to routinely administered vaccina- istics of that vaccine and assuming the presence of an ongoing tion and not to vaccination in the presence of epidemics, vaccination program, they derived estimated costs per child when the cost-effectiveness would probably improve. The cost fully vaccinated of between $1.20 and $12.00, with a median of vaccination is the highest of the vaccines evaluated, possi- of$ 1.90. Employing effectiveness results presented by de Zoysa bly twice as expensive as rotavirus or measles. The relatively and Feachem (1985b), they calculated the cost per diarrhea poor performance of cholera vaccination in terms of cost- episode averted in children under five to lie in the range $3 to effectiveness is largely attributable to the rarity of the disease. $30 (median $5) and the cost per diarrheal death averted to Cholera constitutes a small proportion of the total diarrheal be between $140 and $1,400 (median $220). cases, even in Bangladesh, and even with equal costs per child 100 Jose Marnnes, Margaret Phillips, and Richard G A. Feachem vaccinated its cost-effectiveness would be about five times disposal of excreta has the potential to play a role in controlling lower than that of measles or rotavirus vaccines. their transmission. Environmental improvements of these kinds probably contributed to the reduction in diarrheal mor- MEASLES IMMUNIZATION. A marked association between bidity and to the control of epidemic cholera and typhoid in measles and diarrhea has been described in developing coun- Europe and North America between 1860 and 1920. tries (Koster and others 1981; Morley, Woodland, and Martin A review of sixty-seven studies from twenty-eight countries 1963; Scrimshaw and others 1966), and measles immunization on the effect of water supply and sanitation on diarrhea, related has been considered a potential intervention for diarrhea infections, nutritional status, and mortality was conducted by control (Feachem and Koblinsky 1983). It is the only one of Esrey, Feachem, and Hughes (1985). A median reduction of the three vaccines discussed here that is already commercially 27 percent in diarrhea morbidity and 30 percent in diarrheal available and widely used. A single injected dose at the age of mortality with the provision of improved water supply and nine months is recommended in most developing countries. excreta disposal was found in a subset of studies selected for This is an age at which children are likely to be brought to the their better design. Improvements in water quality appeared to health services for various reasons, though no other vaccines have a lower effect than improvements in water availability or are scheduled for this age. excreta disposal. No adequate data could be located, however, Feachem and Koblinsky (1983), on the basis of the propor- on the effect of improvements in water quality and availability tion of diarrheal episodes and diarrheal deaths that are together with excreta disposal. It is possible that well-designed measles-associated and the proportion of measles cases averted projects combining water supply, excreta disposal, and hygiene in the first five years of life by measles immunization, suggest education may achieve reductions of 35 to 50 percent in that a measles vaccine that is 85 percent efficacious, given at diarrheal morbidity (Esrey, Feachem, and Hughes 1985). It is the age of nine months through eleven months, with coverage expected that, except in areas where other interventions have of between 45 and 90 percent, can prevent 44 to 64 percent of substantially reduced diarrheal mortality, the effect will be measles cases, 0.6 to 3.8 percent of diarrheal episodes, and 6 to larger on mortality rates than on morbidity. 26 percent of diarrheal deaths among children under five years A recent case-control study in the south of Brazil found that of age. This estimate may be too conservative, however, as those infants whose homes had piped water had a diarrhea indicated by recently reported reductions of over 80 percent mortality rare 80 percent lower than those from homes with in total mortality among children nine months through thirty- no easy access to piped water. No difference in mortality rates nine months in Haiti (Holt and others 1990) and of 59 percent was detected, however, between infants receiving treated or in diarrheal mortality among vaccinated children age ten untreated water, suggesting that beneficial effects of piped months to sixty months in Bangladesh (Clemens and others water may be related to the easy availability rather than its 1988). quality (Victora and others 1988). In addition to this Brazilian Phillips, Feachem, and Mills (1987) estimated the cost per study, roughly a dozen studies of the effect of water and child vaccinated against measles to range from $0.60 (Indone- sanitation on diarrhea have been conducted since the review sia, mixed strategy: static/mobile delivery) to $12.00 (Ghana, by Esrey, Feachem, and Hughes (1985). They were conducted outreach delivery strategy), with a median of$1.40. The cost in Bangladesh, the Gambia, Lesotho, Malawi, Nicaragua, Ni- per diarrheal episode averted in children under five through geria, the Philippines, and Sri Lanka. Half of them employed measles vaccination was calculated as $7.00, and the cost per a case-control methodology. The studies have been reviewed diarrheal death averted as $143.00. Discounting diarrhea epi- recently by Cairncross (1990) and Huttly (1990). sodes and deaths averted in years following vaccination makes Analysis of cost data from eighty-seven developing coun- relatively little difference to the results, because the primary tries suggests median annual costs of $14 per capita for rural effect of the vaccine on diarrhea in children under five oc- water supply and latrine projects and $46 per capita for a curs within two years of vaccination. These values are roughly combination of in-house water supply and sewerage in an similar to that of rotavirus vaccine, which is more cost- urban area (Esrey, Feachem, and Hughes 1985). The multiple effective in terms of diarrheal morbidity but less cost-effective benefits deriving from water-supply and sanitation interven- in terms of diarrheal mortality. tions, including the reduction of diarrheal morbidity and mor- tality in other age groups, the reduction of the incidence of IMPROVED WATER SUPPLY AND SANITATION FACILITIES. All other infections, and other benefits not directly related to major infectious agents that cause diarrhea are transmitted by health (Briscoe 1984), make interpretation of simple cost per the fecal-oral route. These enteric pathogens can be transmit- diarrhea death or episode averted problematic (Okun 1988). ted via contaminated water, and water-bome transmission has Further studies on ways to overcome these analytical difficul- been documented for most of them. Improvement in water ties are needed. quality is, therefore, a potentially important intervention. Improvement in water quantity and availability is also impor- IMPROVED DOMESTIC AND PERSONAL HYGIENE. Poor personal tant as an aid to hygienic practices which may interrupt the hygiene of food handlers, inadequate cooking, and storage of fecal-oral transmission. As all principal infectious agents of food at incorrect temperatures for long periods are the most diarrhea are shed by infected persons via the feces, hygienic common contributing factors to food-borne diarrheal out- D[arrheal Diseases 101 breaks in industrial countries (Frank and Bamhart 1986). antidiarrhea programs can be expected to yield greater bene- Similar factors probably apply to food contamination in devel- fits. The emphasis on education in the current AIs-control oping countries, where personal hygiene is likely to be mark- strategy can only accelerate this process. The only new inter- edly impaired by restricted water availability, overcrowding, ventions that are expected are new vaccines. The next decade and poor sanitation. Shortages of fuel, heavy workloads, and may see the incorporation into immunization programs of new lack of access to refrigerators are probably additional contrib- vaccines against Escherichia coli and Shigella, besides the vac- utory factors in developing countries (Esrey and Feachem cines against Vibrio cholerae and rotavirus that were reviewed 1989). earlier in the chapter. Unclean living conditions have been found to be associated with an increased risk of diarrheal incidence in children and ESCHERICHIA COLI VACCINES. The incidence of diarrhea asso- adults (Bertrand and Walmus 1983; Huttly and others 1987; ciated with enterotoxigenic E. coli in developing countries Taylor and others 1986). Infant-feeding bottles and bottle appears to be highest in children under two years, in which one nipples are often highly contaminated (Elegbe and others or even two episodes per year have been noted. The incidence 1982; Hibbert and Golden 1981; Phillips and others 1969; remains high in older children. Because partial immunity Surjono and others 1980), especially in lower-income house- appears to develop after childhood, the target population for holds (Mathur and Reddy 1983). Feeding bowls and other vaccination would be children during the first six months of feeding utensils have been found contaminated with Es- life. It is difficult to estimate the numberof strains the vaccine cherichiacoli (Rowland, Barrell, and Whitehead 1978). Storage would need to cover, and a small proportion of disease would of food at high ambient temperatures increases the risk of still occur at an early age, before vaccination could confer full contamination with fecal organisms (Barrell and Rowland protection. 1979, 1980; Black and others 1982; Capparelli and Mata Taking these two factors into account, the U.S. Institute of 1975). In these conditions, bacterial counts increase substan- Medicine (1986) estimated that 50 to 60 percent of en- tially with length of storage (Black and others 1982; Rowland, terotoxigenic E. coli episodes would be vaccine preventable. Barrell, and Whitehead 1978). Recent advances in the development of vaccines against diar- Few studies have been found that quantify the effectiveness rhea due to enterotoxigenic E. coli were reviewed by Levine of hygiene education interventions on diarrheal morbidity. (1989, 1990). The first experimental vaccines were expected Those that have been located (Bartlett and others 1985; Black to be ready for testing in human volunteers in 1990 (wi4o and others 1981; Khan 1982; Stanton and Clemens 1987a; 1989). Studies in Swedish volunteers have shown that the Torun 1982) focus mainly on the promotion of handwashing. prototype E. coli vaccine is safe and immunogenic. At least 80 Reductions of diarrheal incidence ranged from 14 to 48 per- percent of the volunteers developed intestinal antibody re- cent, suggesting that hygiene education, and handwashing sponse to the antigens after receiving two or three doses of promotion in particular, has a marked effect on diarrheal rates. vaccine. (WHO 1992) Information on the cost of hygiene education is scarce. Costs are lower than for the provision of water supply and excreta SHIGELLA VACCINES.The development of effective ShigeUla disposal facilities, but the success of the promotion of hand- vaccines deserves special priority. Of all the diarrhea-causing washing may depend on the presence of these improved facil- agents, it is the most life-threatening. S. dysenteriae is respon- ities (Feachem 1984). Analysis of three educational sible for an especially serious form of dysentery which can interventions for which data are available reveals a range of prove difficult to manage, even in sophisticated clinical set- cost-effectiveness of under $20 per childhood case averted for tings. Major outbreaks of drug-resistent S. dysenteriae have village-based group education and supervision of day-care cen- occurred in the past thirty years in Central America, Central ters, to $300 to $500 per case averted for individual education Africa, and Bengal (East and West) and have resulted in of families presenting one or more cases of shigellosis (Phillips, hundreds of thousands of deaths in all age groups. Nalidixic Feachem, and Mills 1987). acid was the antibiotic of choice in these circumstances, but resistance to this product is now reported. Development of a Preventive Strategy in the Next Decade The incidence of shigellosis in developing countries is high- est for children age two to four years, but it is already a problem During the next decade, the mainstay of diarrhea prevention from six months of age. All age groups are susceptible and could will probably continue to be the interventions discussed above benefit from immunization. A vaccine incorporating protec- and listed in table 5-6. The change that can be expected is in tive antigens from the most common infecting strains in a the effectiveness (and the cost-effectiveness) of these inter- given geographic area should prevent 80 to 90 percent of the ventions. The goal pursued in the use of the interventions, Shigella infections, depending on the prevalence of these except for the immunizations, is substantial behavior change strains and assuming total coverage of the target population through education. Public health education is a relatively new with a vaccine that is 100 percent effective and is delivered at field, and most countries currently lack the ability to deliver the earliest feasible age (U.S. Institute of Medicine 1986). health education messages effectively to appropriate target Shigella vaccines are reviewed by Levine (1989). A candidate groups. As experience grows and technology improves, these vaccine, developed from a live attenuated S. flexneri strain, is 102 Jose Martines, Margaret Phillips, and Richard G. A. Feachem being tested for safety and immunogeneity in the United specific interventions. Much of the work reported in this States. Other efforts are under way in France and Sweden (WHO chapter on diarrhea prevention has been initiated or influ- 1989). enced by WHO/CDD. It is important that this effort continue with an increased focus on research that will assist in the design and Ideal Prevention and Current Prevention: Closing the Gaps delivery of cost-effective interventions. The WHO/CDDprogram is taking substantial steps to influence The following section describes the technical and institutional the ability and commitment of govemments to implement these aspects involved in closing the gap between current preventive known effective interventions. Measles immunization and im- practices and ideal prevention. provements in water supplies and sanitation are under way in any case and are not the responsibility of national CDD pro- TECHNICAL ASPECTS. Although a set of behaviors associated grams. The other interventions discussed here (table 5-6) exist with lower diarrheal incidence and reduction of severity has in a very limited fashion in most countries. A great challenge already been identified, there is still a long way to go before for the WHO/CDD program during the next decade lies in recti- these desirable behaviors are widespread and commonplace fying this situation. Promotion of exclusive breastfeeding in and before the necessary improvements in water, sanitation, the first four to six months of life is the first preventive and kitchen equipment are available to most families. Exclu- intervention chosen by WHO/CDD for implementation efforts, sive breastfeeding during the first four to six months of life, which started in 1990. followed by partial breastfeeding during the remainder of the first year of life, can be expected to reduce infant diarrheal Defining the Optimal Preventive Strategy morbidity and mortality significantly. Trends in developing countries have, however, been toward lower prevalence and What preventive measures should reasonably be taken for any duration of breastfeeding in urban areas (Popkin, Bilsborrow, and Akin 1982; Popkin and Bisgrove 1988), and early food particular population will depend on a range of factors-the supplementation for breastfed infants is the rule, rather than mortality the relative importance of different etiologies and the exception. Early weaning and Inadequate supplemenra- motlt;terltv motnc fdfeeteilge n concentat of the risk factors that predispose children and adults to diar- tion, the use of foods of low energy and nutrient concentration rhea; the nature and extent of the existing infrastructure; the selection of single foods of low nutritional value, the use rnmenature dext ent of ongoing intrve; of contaminated foods, feeding at infrequent intervals, and government priorities for development and ongoing interven giving infants a disproportionately small share of the family thons; relative and absolute price leels; and, very Important, food are still common practices in developing countries (Ash- the level of relevant budgetary constraints. worth and Feachem 1985c). ~~~~Several of the preventive strategies, in addition to being worth and Feachem 1985c). . relatively cost-effective inpreventingdiarrhea (still one of the Measles vaccination should reduce diarrheal mortality in children under five. Coverage, however, remains low in some most Important audeat in many d ev efing cond regions, averaging 46 percent in Africa and 51 percent in have important additional potential health benefits beyond Sourheast Asia (WHO 1990a), and the effectiveness of vaccines diarrhea control. It would seem reasonable, therefore, that at Sotes Asi (w, 1 9a, an the efeciens least one-twentieth of a country's health budget be spent on may be reduced by an inadequate "cold chain." Improvements eff in water supply an sanitation can lead to reductios iefforts to prevent diarrhea-say $1 per head for high-mortality rheal morbidity and mortality in children under five of about countries (whose health budgets are generally small) and $3 30 percent Access to betterfacilities does not imply improved per head for lower-mortality (and generally higher-income) use, hower,ent. Andess water adanitatiesdons inoterventimron ved countries. A rough order of priorities for preventive interven- use, however, and water and sanitation interventions have tions, based substantially on the cost-effectiveness of currently failed to reduce diarrheal rates in some settings (Nigeria, Imo available technology, is set out below: State Evaluation Team 1989; Ryder, Reeves, and Sack 1985). Personal and domestic hygiene behaviors that may reduce * Measlesvaccinationforatleast80percentofthenine-to diarrheal incidence have been identified. Handwashing before twelve-month-old population. At a cost of about $2.00 per preparing and offering food and careful disposal of child feces, additional child vaccinated (assuming a vaccination pro- two potentially effective practices, may depend on changes in gram is already in place), and possibly double that to cover attitudes and in the availability of water and soap. In the least accessible final 20 percent, this amounts to some Bangladesh, for example, it was observed that mothers washed $0.10 per head, or up to 10 percent of the proposed budget their own hands before offering food to young children in 53 allocation for diarrhea prevention in high-mortality coun- percent of feedings but in almost all cases without using soap. tries. Such a vaccination program would be expected to Other caretakers were observed washing their hands when avert 23 percent to 29 percent of diarrheal deaths in those offering thefood inonly 1 Ipercent of the feeds (Guldan 1988). under five years of age. It would also avert substantial measles mortality and, in areas where measles is severe and INSTITUTIONAL ASPECTS. The WHO/CDD program has played a associated with high risks of xerophthalmia, an additional major role since 1980 in promoting research on diarrhea epi- benefit would be the prevention of a substantial proportion demiology, vaccine development, and the cost-effectiveness of of blindness. D arrheal Diseases 103 * For areas where breastfeeding rates are low or early and antidiarrheals and their limited role in curative and pal- supplementation is a common practice, investment in liative treatment of diarrhea. breastfeeding promotion aimed at achieving exclusive breastfeeding up to four or six months should also be a SECONDARY PREVENTION: REHYDRATION. Until the late 1960s priority. Some interventions, such as changing hospital the primary medical method of rehydration was intravenous routines, can be implemented relatively cheaply (under (Iv) therapy. The development of oral rehydration solutions, $1.00 per delivery-or less than $0.05 per head even if all combining electrolytes and glucose, sucrose, or rice powder, children were delivered in a hospital and routines changed has been an important breakthrough (Darrow and others 1949; in all hospitals). For about $0.20 per head per year it should Nalin and others 1968; Pierce and others 1968; Lancet 1981). be possible to implement a package of hospital-based, legis- The composition for ORS recommended currently by WHO is 3.5 lative, and mass media measures designed to improve grams of sodium chloride, 2.9 grams of trisodium citrate dihy- breastfeeding rates. drate, 1.5 grams of potassium chloride, and 20 grams of glucose * With the remaining funds available, promotion of im- (WHO 1987c). Clinical studies have shown that ORS can suc- proved weaning and hygiene practices for mothers (both cessfully rehydrate 90 to 95 percent of cases of acute watery feasible for about $0.30 per head each), should be next on diarrhea, substantially reducing the requirements for iv rehy- the agenda (the former particularly for high-mortality coun- dration (Hirschhorn 1980; Levine and others 1986). Not only tries), although adoption of hygiene practices may be con- are ORS as effective as intravenous rehydration in the vast strained by water availability. majority of cases, they are also less hazardous, especially in * Improvements in water and sanitation facilities are rel- settings in which the risk of infection in the hospital is high. atively expensive, estimated at between $14 to $46 per head In efforts to identify ways of promoting early and increased annually. Even if the whole of the $1 to $3 per head use of rehydration therapy, researchers have explored the notionally set aside for diarrhea prevention were allocated efficacy of fluids which can be prepared with ingredients avail- to water supply, only one-fifth to one-forty-fifth of individ- able in the home. Solutions of sugar and salt (sss) have been uals would be covered by the interventions. Some countries shown to correct fluid volume deficits in noncholera diarrhea could partially overcome this difficulty by requiring that in adults and children (Clements and others 1981; Islam and individual consumers be responsible for some or all of the others 1980). They correct acidosis slowly, however, and are costs of these services, which they may, indeed, provide for not adequate for the correction of hypokale[nia or the replace- themselves. Self-provision of Ventilated Improved Pit la- ment of electrolyte losses in cholera (Islam and Bardhan 1985). trines in Lesotho is a case in point and has achieved notable Home-based fluids (such as gruels, soups, and diluted yoghurt health benefits (Daniels and others 1990). drinks) which contain some salt and a source of glucose, can be effective in preventing dehydration as long as the sodium concentration lies in the range of 30 to 80 millimolars, and the Case Management osmolarity is less than 300 milliosmoles per kilogram of water (WHO 1987a). The glucose range is not relevant if a complex The elements of the current case management strategy, its carbohydrate is used. likely development over the next decade, the gap between Studies of the effectiveness of ORT promotion under field good and actual case management, and the definition ofan op- conditions are not uniformly positive in their findings (for timal case management strategy are examined in this section. example, Tekce 1982). This issmall wonder, given the reliance on adequate supplies ofORS and the multiple steps required by Elements of a Case Management Strategy caretakers in the preparation and use of ORT, where practice can readily stray from the ideal. In a later section we provide Case management involves several dimensions: the way diag- some evidence of the extent to which practice diverges from noses are made, the nattire of medication and advice offered, the ideal. Nevertheless, the overall findings on effectiveness and the level and location of services. We concentrate princi- tend to be positive. Studies in Egypt (Kielman and others pally on the effectiveness and costs of medication options, 1985) and India (Kielman and McCord 1977) described referring more briefly to the way these are influenced by various marked decreases in diarrhea-associated mortality rates with options for making diagnoses and for providing services. the use of Qiss by village health workers for the treatment of A serious complication of some diarrhea episodes is dehy- diarrhea. It is not clear, however, if the results were due to the dration, which can lead to death, particularly in small children. use ofsss at home, an increase in the use ofORs in the treatment Correct case management involves, at a minimum, the preven- of more severe cases, or the repeated home visits by health tion and correction of dehydration. Considerable effort has workers. Recent evaluations of the effect of oral rehydration been devoted in the last twenty years to the development and therapy in Egypt associate a sharp reduction in diarrheal mor- promotion of rehydration solutions which can be taken orally. tality rates among children under five with improved diarrheal These advances, and the effectiveness and cost of the alterna- case management by mothers and doctors (Egypt, National tive methods of prevention and correctiotn of dehydration, are Control of Diarrheal Diseases Project 1988; el-Rafie and others outlinied below. We follow this with a discussion of antibiotics 1990). The constancy of death rates from other causes, the 104 Jose Martnes, Margaret Phillips, and Richard G. A. Feachem nature of the change in the seasonal pattem of total mortality, associated household costs in using ORS could, however, be and the lack of change in diarrhea incidence or nutritional reduced by regular stocking of household or neighborhood status are all compatible with the hypothesis that improved supplies. Both oRs and home-based fluids probably place con- case management is the explanation for the reduction in siderable time demands on mothers. Evidence is scarce, but a diarrhea mortality. Nevertheless, uncertainty as to the accu- rough calculation suggests that a mother with young children racy of mortality statistics may indicate a need for caution in could spend 10 percent of the year treating diarrhea in her the evaluation of these results. children with ORT (Leslie 1989). Data on the effect of ORT in a hospital setting summarized Oral rehydration therapy is considerably less costly than by WHO/CDD from twenty-eight reports from twenty-one coun- intravenous rehydration, not only in terms of the ingredients tries indicate median reductions of 61 percent in diarrhea and equipment but, more important, because of the reduced admission rates, 71 percent in overall diarrhea case-fatality demand for hospital beds and all the associated inpatient costs. rates, and 41 percent in inpatient diarrhea case-fatality rates. Oral rehydration therapy can generally be provided at the Reductions of nearly 70 percent in the proportion of hospital- lowest level of health facility or in the home, as can the clinical ized cases treated with iv fluids after the introduction of ORT diagnosis required both to distinguish watery from other diar- were reported from Malawi and Tanzania. In Stanley Hospital, rheas and to identify the degree of severity of dehydration. At Madras, India, the average number of days of hospital stay Safdargang Hospital in New Delhi, India, diarrhea treatment decreased from 6 to 1.5 in association with the gradual increase costs were reduced by 69 percent and at San Lazaro Hospital, ofORT use in a period of seven years (WHO ] 988a). In a hospital Manila, the Philippines, by 62 percent, when ORT was intro- in Haiti, for example, diarrhea mortality fell from 35 percent duced (WHO 1988c). In a children's hospital in Mexico City, to 14 percent and then to less than I percent with the creation the opening of an oral rehydration unit reduced the number of of an ORT unit (Allman and Rohde 1985). inpatients with diarrhea by 25 percent, giving rise to consid- The effectiveness of ORT, high in preventing death from erable potential savings, (Phillips, Kumate-Rodrigues, and dehydration caused by acute watery diarrheal episodes and, Mota-Hemandez 1989). Another study in Mexico City also possibly, in aiding recovery of appetite and thus increasing food identified savings as a result of the introduction in two clinics intake, is probably low in the prevention of death from persis- of therapeutic norms stressing the importance of ORT (Castro tent and dysenteric episodes. and others 1988). In the first two years following staff training The cost of ORS is low; UNIPAC (the central warehouse of and the establishment of an ORT unit at the Kamuzu Central UNICEF in Copenhagen) provides I-liter sachets for $0.07. A Hospital, Lilongwe, Malawi, there was a 50 percent decrease course oftreatment using 2 to 3 liters would cost $0.14 to $0.21 in the number ofchildren admitted to the pediatric ward with per episode. The cost of the ingredients for home-based solu- diarrhea, a 56 percent decrease in the use of intravenous fluids tions is likely to be lower than the cost of packaged ORS, and to rehydrate such children, and a 32 percent reduction in certainly the cost is lower for the health services if ORs are recurrent hospital costs attributable to pediatric diarrhea otherwise supplied free or at a subsidized rate. Still, other costs (Heymann and others 1990). The results, however, are not to the health services may he greater when home-based solu- uniformly positive. In a study in Indonesia, neither hospital- tions rather than ORS are used. The effectiveness and safety ization rate nor intravenous fluid expenditure were related to of sss depend, for example, on the ability of mothers to learn the rate of ORS use in the community (Lerman, Shepard, and and retain the skills required in its preparation. This calls Cash 1985).The time demands ofORT probably remain higher for educational interventions which may need to he more than for iv rehydration, which may partly explain why in many substantial than those required for oRks. Education of mothers cases children hospitalized with diarrhea who could be treated on the use of sss has been shown to be short-lived in its effect with ORT receive Iv therapy instead. on behavior (AED 1985; Chowdhury, Vaughan, and Abed Relatively few studies have attempted to measure both the 1988; WHO 1986b), which suggests the need for more effective costs and the effectiveness of field-based ORT programs, and or repeated education. How this compares with the kind of with few exceptions measures of health effects (for example, effort required to educate mothers about ORS is, however, not deaths averted) have not been used. "Diarrhea cases treated" documented. is the most common direct measure of effectiveness. The Oral rehydration salts and home-based fluids differ in their nature of the cost analysis in these studies has not been cost implications for families, and the net result is difficult to consistent. Shepard, Brenzel, and Nemeth (1986) identified judge. On the one hand, ingredients for home-based fluids are studies with adequate cost data in only four countries and likely to be fully paid for by the family, whereas ORS may well reanalyzed the cost data to generate reasonably compatible be wholly or partly subsidized. Household costs for preparing estimates of full economic costs. The results from these studies food-based fluids may be particularly high if regular prepara- and three additional ones are shown in table 5-7 and reveal tion of small quantities (with the associated time and fuel costs per child treated in the range ofapproximately$1 to $10. demands) is required in order to avoid spoilage. On the other Much of the variation in cost between countries can be ex- hand, if ingredients are at hand, home-based fluids may be plained by differences in gross national product per capita. The more convenient to mothers, reducing costs in time and trans- rest is probably the result of (a) differences in the level of portation associated with visits to health facilities. Treatment- services provided (Horton and Claquin [1983] found an ap- Diarrheal Diseases 105 Table 5-7. Cost of Treatment with ORT severe diarrhea, then effectiveness would increase. If the pre- Average cost ORT case-fatality rates were lower, 1 per 1,000, for example, and per child treated ORT were only 50 percent effective, deaths would be averted in Location (1982 U.S. dollars) Reference 0.05 percent of the treated cases. Table 5-8 shows how costs The Gambia 0.70 Shepard, Brenzel, and per death averted vary for different costs per episode treated Nemeth, 1986 and deaths averted per case treated. The range of most likely Indonesia 0.77 Shepard, Brenzel, and estimates appears to be between $1,000 and $10,000. Most of Nemeth, 1986 the estimates of Shepard, Brenzel, and Nemeth (1986) lie Malawi 1.86 Qualls and Robertson, below $10,000, and some lelow $1,000. 1989i Honduras 2.94 Shepard, Brenzel, and Nemeth, 1986 CURE AND PALLIATIVETREATMENT-D)RUGs. A large variety of Bangladesh 3.30 Horton and Claquin drugs is currently promoted commercially for the management 1983 of diarrhea. They include antimotility drugs, antisecretory Egypt 5.56h Shepard, Brenzel, and drugs, adsorbents, and antibiotics. The majority of these drugs Nemeth, 1986 have no proven benefit and can be positively dangerous, Swaziland 6.28 Quails and Robertson, particularly for small children. Most have a minimal effect on Turkey 9.66 Brenzel, 1987 the clinically important aspects of diarrhea, some have nega- Turkey___________9____ 66_____ Brenzel______________1987 _____- tive side effects, and all are suspected of diverting attention a. Study reports data collected in 1984. away from life-saving rehydration therapies. Table 5-9 outlines b. Cost per child in the population. th e Source: See last column of table. the evidence on the efficacy of selected drugs commonly promoted as antidiarrheals. Case management for adult sufferers of diarrhea poses differ- proximately tenfold difference in cost between treatment in ent challenges from those of childhood diarrhea. Not only are basic facilities and that in relatively sophisticated facilities in there differences in the relative incidence of different etiolo- Bangladesh); (b) the scale of the operation's fixed costs, which gies of diarrhea and in the average severity and length of tends to be relatively high (for example, 96 percent of the episodes (U.S. Institute of Medicine 1986) but also in the Turkish ORT program costs were fixed [Brenzel 1987]), giving significance of different characteristics of diarrhea. For exam- rise to substantial changes in average costs as the program pie, the inconvenience of poor control over defecation is likely expands; (c) the efficiency with which the program is con- to be an important motivation for adults seeking treatment for ducted; and (d) the costing methodology employed, particu- diarrhea. Furthermore, the negative side effects of anti- larly with respect to the measurement of personnel costs, diarrheals are generally less severe for adults. Palliative treat- which are often the largest component and the least straight- ment, strongly discouraged for young children, given the forward to measure. currently available selection of drugs, has a role to play in adult The effectiveness of the programs in averting deaths is likely treatment. to vary, depending on the nature and severity of all the PatientswithAII[Dscommonlysufferfromsevereandchronic diarrhea cases, the degree of selectivity in treatment, how well watery diarrhea, associated variously with protozoan organ- the intervention is delivered, and the extent of use and effec- isms such as Isospora belli and Cryptosporidium (Young 1987), tiveness of alternative treatments. Case-fatality rates prior to with mycobacterial organisms such as Mycobacterium avium the widespread use of ORT were estimated to be of the order of intracellulare (Carswell 1988), and occasionally with well- 6 per 1,000 (Snyder and Merson 1982). If ORTwere 100 percent recognized agents such as Giardia, Salmonella, Shigella, and effective and targeted at the diarrhea group with a case-fatality Campylobacter. Some reports have suggested improvements of ratio of 10 per 1,000, then deaths averted would be I percent diarrhea in patients who have been treated with either of treated cases. If case-fatality ratios were higher or ORT spiramycin or a-eflomithine, though the evidence is not con- programs were successfully targeted at those with even more sistent (Kaplan, Wofsy, and Volberding 1987). At the present time maintenance of hydration and, in adult patients, symp- tomatic control with opiate derivative antidiarrheals are the Table 5-8. Cost per Death Averted for Different only helpful interventions (Kaplan, Wofsy, and Volberding Treatment Costs and ORT Effectiveness 1987), though a role for prostaglandin inhibitors has been Costs per episode Cases treated that prevented one death suggested (Young 1987). treated (U.S. dollars) 0.05 0.10 0.50 1.00 For all ages there is a legitimate, though limited, role for antibiotics in the treatment of diarrhea: antibiotics can signif- 0.50 1,000 500 100 50 icantly diminish the severity and duration of diarrhea due to 1.00 2,000 1,000 200 100 cholera and Shigella and shorten the period of pathogen excre- 0.00 20,000 1,0000 2,000 ,000 tion in the case of dysenteric episodes. Antibiotics have no proven value for the routine treatment of acute watery diar- Source: Authors' calculations from assumFptions based on figures of table 5-7. rhea, however, and their use, besides being inappropriate, may 106 Jose Marrnes, Margaret Phillips, and Richard G. A. Feacheni Table 5-9. Efficacy and Side Effects of Selected Drugs Promnoted for Treatment of Diarrhea Drug EfficacN Side effects Antimotilits Loperamide (Imodium) Trials have failed to demonstrate clinically Can prolong infectious diarrhea, cause toxic significant effects on daily stool volume or megacolon, central nervous system toxicity rehydration requirements Diphenoxylare (Lomotil) No clinically important reduction in number or Can worsen clinical course (prolong fever and volume of stools, length of episode, or Shigella excretion); central nervous system intravenouis liquids required toxicity at rec(mmended doses Anrsecretory Bismuth subsalicylate (Pepto-Bismol) No effect on quantity of liquid or total stool Required doses too large to be practical or safe weight; diminished number of liquid stools and subjective complaints in young adults with travelers' diarrhea Aciduric bacteria No therapeutic or prophylactic benefits demonstrated Adsorbents Kaolin Can increase consistency of stool; no effect on May interfere with efficacy of antibiotics weight, liquid content, or frequency of stools Charcoal Ineffective Interferes with effects of tetracycline Attapulgite and Smectite Can increase consistency of stools; no conclusive May bind and inactivate other drugs evidence of effect on fluid or electrolyte losses Cholestyramine Conflicting evidence; not recommended Interferes with fat and vitamin absorption - Not availahle. SNurLe. VHO 1986 anld 1990. even be dangerous (WlHo 1986a). In Basrah, for example, antidiarrheals (discussed later), accounts for the stIbstantial Mahmood and Feachem (1987) found that antibiotic therapy levels of expenditure on these dnmgs. Sixtypercent ofdrug costs was significantly associated with prolonged hospitalization of for treating simple diarrhea were spent on antimicrobials in infants with diarrhea caused by enteropathogenic Escherichia one area of Indonesia (Quick and others 1988). In another coli. In table 5-10 it is noted which antibiotics are recom- study in Indonesia (Lerman, Shepard, and Cash 1985), more mended for the treatment of diarrhea associated wvith specific than five times as much was spent on antimicrobials and pathogens. antidiarrheals as on ORS ($1.01 compared with $0.18 per child). On grounds of efficacy alone i-nany of the drugs marketed as For the relatively few cases of diarrhea in which antibiotics antidiarrheals can be dismissed. Furthermore, they can be quite can appropriately be prescribed, recommendations concerning costly: a bottle of tetracycline syrup, for example, costs more the preferred choice from the range of available drugs have than six times that of a liter-size package of ORS in Indonesia. largely been based on relative efficacy (taking into account the This, together with the high rate of use of antibiotics and problem of resistance) and severity of side effects. Costs have Table 5-10. Antibiotic Treatment for Specific Diarrheal Diseases Disease __ _ _ Recommended treatment Alternatives Cholera (proven or suspected) Tetracycline or doxycycline Furazolidone or sulamethoxazole-trimethoprim Dysenitery (no coproculture necessary) Sulfamethoxazole-trimethoprim Nalidixic acid or ampiciliin Amebiasis (trophozoite in stool) Metronidazole In severe cases, dehvdroemetine hydrochloride Giardia lamblia (trophozoite or giardial antigen in stool) Metronidazole Quiniacrine Carnpyloboicter Erythromycin' None Severe Yersinia Chloramphen ,ol None Escherichia coli rotavirus No antihiotics None a. Antibit,tic treatment is reconmmended ionly if rapid diagnosis n possible andi treatment can begin on the first day h. Antibiotic- can lead to harimful overgrowth of organisim, 5 >8 follows we assume that, prior to control, transmission was at Wucherenabancrofti >5 >5 moderate to intense levels. Source: Authors. Recent analytical work on the transmission dynamics of helminth parasites in human communities has led to the eradicated. In practice, however, spatial factors often result in derivation of a "control criterion" of a fairly general character, reinvasion from neighboring areas where control effort is less which defines the frequency (in units of time) and the coverage effective. (the proportion of the community treated during each inter- Return times to pretreatment infection levels are inversely vention cycle) of treatment ideally required to reduce trans- correlated with the life expectancy of the parasite in the mission, community-wide, to very low or negligible levels. This human host (Anderson 1986; Anderson and May 1985; Bundy level of control should be significantly greater than that re- and others 1 988b). If life expectancy is short, the return time quired to reduce morbidity. For an anthelmintic drug of effec- is fast, and vice versa. Current estimates (often very crude and tiveness h, whose effectiveness defines the proportion of an approximate) of the life expectancy of some important human individual's worm load killed by a single or a short course of helminths are given in table 7-9. treatment (for most drugs h > 0.9), the proportion of a human The intrinsic properties of a parasite to reproduce and trans- community, p, that must be treated per unit of time (for mit from host to host may be measured by reference to its basic example, year or month) to block transmission is given approx- reproductive rate, Ro. This quantity defines the average num- imately by the relationship ber of female worms (for a diecious species) produced by one female throughout her reproductive life span that themselves (3) p > {I - exp l(l - R.)/L ] / h gain entry to a host and survive to reproductive maturity in a susceptible human population. Clearly if R, is less than one the Here Ro is the basic reproductive rate and L is parasite life parasite will not persist endemically in the human community. expectancy in the human host (Anderson 1986). If effective- This constraint sets the target for control programs-the in- ness is less than 100 percent (that is, h < I ) and Ro large, then tensity of control (whether by chemotherapy or other meth- it is possible that the value of p will be greater than one. In ods) should aim to reduce reproductive success to less than these cases the units of the life expectancy term, L, must be unity in value (on average). If R0 is greater than one the reduced (that is, years to months) such that the proportion to parasite will persist. The magnitude of R,) for a given parasite be treated is calculated on a more frequent cycle (shorter time interval). In general, if we use the values of R. and L listed in table Table 7-10. Estimated Basic Reproductive Rate 7-11, for a drug that is 95 percent effective (h = 0.95) we can (R.) provide some rough guidelines on the coverage and frequency Parasite Ro Locaton Source of treatment for the main helminths in areas of moderate to high transmission intensity. These guidelines are summarized Ascaris lumbricoides 4-5 Iran Croll and others 1982 in table 7-11. Necatoransericanus 2-3 India Anderson 1980 In assessing the information presented in this table it must Tnichuns trchiura 4-6 St. Lucia Bundy and Cooper I sesn h nomto rsne nti al tms 1989 be noted that the parameter estimates (of Ro and L) are rather Schistosomamansoni 1-2 Egypt Hairston 1965 crude at present. As a consequence of this, the estimates of Schistosoma frequency and coverage are very approximate and should only hematobium 2-3 Egypt Hairston 1965 be viewed as rough guidelines. Schistosomajaponicum 1-4 Philippines Barbour 1982 The data in table 7-11 indicate that when consideringsingle Onchocerca volvulus >9 C6te d'lvoire Dietz 1982 (or short course) treatment for multiple infections the interval Source: See last column. between treatments will depend on the precise mix of parasite 150 Kenneth S. Warren and others Table 7-1 1. Estimated Ideal Coverage and combinations of three anthelmintics. More precise estimation Frequency of Drug Treatment to Reduce Parasite of the logistic implications of this strategy-in particular, Transmission to Very Low Levels finely stratified data on the global prevalence and intensity of Frequency of infection-should be possible from existing data sources, but L Coverage treatment further studies are necessary to determine the effectiveness and Parasite R0a (years)b (percent) (years) pharmacodynamic properties of these drugs when used Tnchuris trichiura 5 1 80-90 0.5-1 concomitantly. Ascaris lumbricoides 5 1 80-90 0.5-1 Second, the analyses suggest that recent advances in the Necator americanus 2.5 2.5 70-80 2 theory of population dynamics permit the estimation of "opti- Ancylostoraduodenale 2.5 2.5 70-80 2 mal" chemotherapeutic interventions (in frequency and cov- Schistosoma nansoni 2 4 70-80 4 erage of treatment) for a given parasite species in a given Schistosorma locality. These estimates currently lack precision because of hematobium 2 4 70-80 4 the absence of adequate data, but they may serve as broad Onchocerca volvulus 10 8 70-80 1 Wuchereriabancvofvl 10 8 70-80 1 guidelines for the development of control strategy. Further work is required to improve procedures for parameter estima- a. Basic reproductive rate. tion (particularly RO), and to collect appropriate data for the b. Parasite life-expectancy. Source: Authors. range of parasite species in different epidemiological situa- tions. These conclusions suggest that the study of parasite population dynamics and epidemiology has an important, and species. In areas where filarial worms or geohelminths predom- increasing, role to play in the development of policies of inate, an ideal frequency is every year. For geohelminths, this parasite control. interval might be slightly lengthened if treatment is targeted at schoolchildren, since this age group tends to be heavily Cost and Effectiveness of Screening and Targeting infected. For hookworms the ideal interval is two years, whereas for schistosomes it is roughly four years. Many options are available for the delivery of chemotherapy These theoretical intervals are longer than those that are to the community. Both economic and epidemiological factors commonly applied in practice. There are two main reasons for determine which is the most cost-effective method in a given this. First, current estimates of frequency are often based on situation (Prescott 1987; Prescott and Bundy 1989: Prescott time taken for prevalence, rather than intensity, to rebound to and Jancloes 1984). At the broadest level, there is an obvious pretreatment levels. Because intensity is more closely corre- choice between the treatment of children only (targeted che- lated with transmission (and morbidity), it is the slow rebound motherapy) and the treatment of the whole population, in- in intensity which should more appropriately determine the cluding adults (population chemotherapy). For each of these frequency of treatment. Second, current estimates of the fre- options there is a choice between the treatment of all individ- quency of intervention are usually based on the rate of reinfec- uals, irrespective of infection status (mass chemotherapy), and tion observed after the treatment of a few individuals living in the treatment of individuals shown to be infected after diag- an area of otherwise uncontrolled transmission. Reinfection nostic testing (screening chemotherapy). In the latter case, rates under these conditions will be much higher than if the treatment can be given to all infected persons (prevalence population were treated as part of a community-wide control screening) or only to those individuals with "heavy" infections program, thus giving a misleading impression of the frequency (intensity screening) (terminology adapted from WHO 1985). of treatment required. Examples of situations under which Each of these six different options (table 7-12) is associated community-wide treatment results in slow rates of reinfection with different costs and different levels of effectiveness, de- with schistosomiasis are described by Wilkins (1989). fined as reduction in prevalence, intensity, morbidity, or some In all cases, the coverage of treatment should be high: other variable. Effectiveness should ideally also be defined per typically, 70 to 90 percent of the target community. In areas unit of time: because reinfection invariably occurs after treat- of high-intensity transmission, the intervals between treat- ment, the effective reduction in prevalence, intensity, and ments should be shorter and the population coverage greater, morbidity can be defined only for a particular observation or whereas in areas of low transmission, lower frequencies and intervention period. coverage would be appropriate. In this section we present an analysis of the cost-effective- These preliminary analyses suggest two important conclu- ness of the altematives outlined above in a standard popula- sions of relevance to the development of a strategy of helminth tion with defined epidemiological, economic, and behavioral control. First, they indicate that recent advances in anthel- parameters. The analyses are essentially static because, at this mintic development have made multispecies control of the stage, they do not reflect parasite population responses to the main helminthiases a plausible option. Theoretically, the various interventions. The more complex problem of dynamic broad spectra of modern anthelmintics should permit the economic analysis, taking account of differing reinfection pat- treatment of a complex range of infections using only five terns, is a subject for future research. Helminth infection 151 Table 7-12. Chemotherapeutic Strategies for Control of Helminthiasis No screening Screening for infection Screening for intensity of infection Targeted Population Control Population Control Population Control populaton tnreated strategy treated strategy treated strategy None All children and Population mass All infected Population screen- All heavily Population screen- adults (PM) children and adults ing for prevalence infected children ing for intensity (Psp) and adults (psi) Children All children Targeted mass (TM) All infected Targeted All heavily in- Targeted screen- children screening for fected children ing intensity (TsI) prevalence (TSP) Source: Authors. THE BASE CASE. Using representative data, we have simulated intensity options, which are more costly than the equally the costs and effectiveness of the chemotherapy options for a effective screening for prevalence options. They have the standardized population in the scenario for the base case shown highest screening costs because of lower technician productiv- in figure 7-3. The simulations model a chemotherapy program ity in assessing specimens for the intensity instead of the mere implemented for two years with six-month treatment cycles. presence of infection, which implies higher technician and Total costs per capita for the entire two-year program are related cost requirements. estimated from unit cost data derived from an actual control In examining the economic factors here and elsewhere in program (Bundy and others 1989; Terry, Bundy, and Horton this collection, two standardized populations are considered: 1989). Because costs vary between countries, these data should one has a low mortality rate and is at an intermediate level of be interpreted as comparative rather than absolute. Note that development, and the other has high mortality and birth rates. per capita costs refer to the cost of the entire program (two These demographic differences influence only the overall years) per person in the population (N = 1 million) whether costs of the various chemotherapy options, the populations or not infected. Effectiveness is measured by the proportion of with low mortality and low birth rate attracting lower costs total heavy infections (adults plus children) treated, assuming because there are proportionately fewer children. The rela- the epidemiological and behavioral parameter values summa- tive position of the different treatment options is unaf- rized in table 7-13. fected, and the comments in this section apply equally to In considering effectiveness only, without regard to budget both types of population. constraints, the main result in the base case is that the mass Increasing prevalence of infection does not change the main chemotherapy strategies are much more effective than any of results illustrated by the base case. Figure 7-3 also shows the the screening options. Population mass chemotherapy emerges effect of increasing prevalence to 100 percent in both children as the best option, treating 94 percent of heavily infected cases, and adults. The mass chemotherapy options continue to dom- compared with 80 percent for targeted mass chemotherapy. By inate both on effectiveness and on cost-effectiveness criteria. contrast the screening options are only about 60 percent The only consequence of higher prevalence is to lower the effective. This result reflects the importance of compliance effectiveness and raise the cost of all the screening options. In behavior: the effectiveness of mass interventions is modified the targeted case this results from the smaller share of children only by compliance with treatment, whereas the effectiveness in the total number of heavy infections. The graph in figure of screening-based interventions is also modified by compli- 7-3 showing 100 percent infection prevalence reflects the low ance with the diagnostic screening test. Because screening screening compliance of adults who, in this scenario, have a compliance rates are typically much lower than treatment larger share of the total heavy infections. Thus higher preva- compliance, especially for adults, any screening approach must lence tends to make mass chemotherapy more attractive. inevitably treat fewer infected individuals. When screening compliance is raised from the low levels In the presence of a budget constraint, the screening options assumed in the base case to the hypothetical maximum of 100 are not only less effective but also more costly than the mass percent both for adults and children, the screening options chemotherapy alternatives. At lower budget levels, targeted achieve the same effectiveness as the corresponding mass mass chemotherapy is both more effective and less costly than chemotherapy alternatives. The costs of screening also rise, targeted screening for prevalence. At higher budget levels, the however, because of the higher volume of postscreening treat- more effective population mass chemotherapy becomes afford- ment activity. As revealed in the graph showing 100 percent able and always dominates the more costly screening variants. screening compliance in figure 7-3, the implications for opti- This cost disadvantage of screening options occurs because the mal choice with a budget constraint are the same as in the base slightly lower treatment costs achieved by only treating in- case. Thus even when effectiveness is maximized by the im- fected cases are offset by the very high costs of the screening plausible assumption of 100 percent compliance with screen- itself. This disadvantage is most acute with the screening for ing, the cost criteria make screening options unattractive. 152 Kenneth S. Warren and others Figure 7-3. Cost and Effectiveness of Different Approaches to Community Treatment Base case High case: 100% infection prevalence Proportion of heavily infected treated Proportion of heavily infected treated 1- 1- PM PM 0.9 0.9 _ TM 0.8 _ 0.8 TM 0.7 -0.7 PSP PSI TSP TSI ~~~~~~~~~~~~~~PSP PSI 0.6 S 0.6 _ T TSP TSI 0.4 _ll0.4_lllllllllllllll 0 1 2 3 4 5 6 7 8 0 1 2 3 4 5 6 7 89101123145 Cost per capita (in U.S. dollars) Cost per capita (in u.s. dollars) High case: 100% screening compliance High case: drug cost =US$4.00 Proportion of heavily infected treated Proportion of heavily infected treated 1- 1- PM PSP PSI PM 0.9 0.9 - TM TSP TSI TM traten).Bae as asmpton araitdi al -3 0.8 0.8 - 0.7 -0.7- PSP PSP 00.6 .6 TSI TSP - 0.5 -0.5- 0.4 - 0.4- III 0 1 2 3 4 5 6 7 8 0 1 2 3 45 6 7 8 9 10 1112 1314 15 Cost per capita (in u.s. dollars) Cost per capita (in u.s. dollars) Note: Initials refer to the chemotherapy approaches detailed in table 7-12. Effectiveness is measured in terms of the percentage of heavy infections treated. Cost is given in U.S. dollars per person in a population of 1 million, whether or not infected, and are for the total program of two years (six-month cycles of treatment). Base case assumptions are listed in table 7-13. Source: Authors. Screening emerges as a preferred strategy only if drug costs intensity is less costly than screening for prevalence, although are high in relation to screening costs and only in the presence both options are equally effective. At higher budget levels, the of a budget constraint. The last graph in figure 7-3 illustrates more expensive but much more effective mass chemotherapy this result in that drug costs are increased to us$4, which alternatives become affordable and preferred. An important simulates the cost of praziquantel for the treatment of schisto- factor here is that it is only at higher budgetary levels, and only somiasis. In this scenario, only the screening options are afford- with mass chemotherapy, that an acceptable level of effective- able at lower budget levels, initially on a targeted basis and ness is achieved under the present assumptions. Note also that then on a population basis. At these levels, screening for this model assumes a six-month cycle of treatment, whereas, Helminth Infection 153 Table 7-13. Epidemiologic and Behavioral because it fails to treat the increased share of heavy infections Parameters of the Base Case in adults (figure 7-4). Even in this scenario, however, mass (percent) chemotherapy is more cost-effective than the screening equiv- Parameter Value' alents. Note also that the age distribution described here is an extreme case and that in reality even hookworm infections Prevalence of infection may have a high proportion of intense infections in children, Childuls 60 a situation that would favor targeted mass chemotherapy. Adults 30 The main implications of these simulations are the follow- Children with heavy infections 1 5 Percentage of all heavy infections that ing. First, if there is no budget constraint and the planning occur in children 90 objective is simply to maximize the proportion of heavy infec- Reduction in prevalence due to each cycle tions treated, then population mass chemotherapy will always of therapy 80 be the optimal intervention. Targeted mass chemotherapy will Compliance with treatment always be inferior because not all heavy infections occur in Children 5 children. The higher the proportion of heavy infections in Adults 75 Compliance with screening children, however, the greater will be the similarity in effec- Children 70 tiveness between population mass and targeted mass chemo- Adults 50 therapy. Screening options will generally be inferior because of attrition owing to imperfect screening compliance, but even a. vafues are derved from empical observasion and oroadly descibe the . pattems observed for Ascaris, Tnchuns, and the schistosomes. If compliance were complete, these screening options would Source: Authors. only be as good as, but never better than, mass chemotherapy. Second, if there is a budget constraint the mass chemother- in practice, chemotherapy against schistosomiasis is delivered apy options will generally tend to be preferred, with targeted on cycles of twelve months or more, which would substantially mass chemotherapy at lower budget levels and population mass reduce the costs of all the control programs. chemotherapy at higher levels. For infections which tend to In all the above scenarios we have assumed that intense be most intense in children (such as Ascaris, Trichuris, and the infection is concentrated in children, as is the case for Ascaris, schistosomes) the targeted option may provide similar levels Trichuns, and schistosome infections. When it is assumed that of effectiveness to the population option. For infections such intensity of infection rises to an asymptote in young adults, as as hookworm, where intensity is high in adults, the effective- appears to be the case for hookworm infection, the relative ness is reduced, although this may be outweighed by the cost-effectiveness of targeted mass chemotherapy is reduced beneficial reduction in morbidity in children. Third, only in the particular case of drug costs that are high in relation to screening costs, and in the presence of a budget Figure 7-4. Cost Effectiveness of Various constraint, will screening options be more cost-effective. In Approaches to Treatnent when Intense Infection these circumstances, higher drug costs will initially favor Occurs in Adults screening for prevalence, with screening for intensity taking over only when drug costs are high enough to offset the higher Proportion of heavy infections treated costs of screening. 0.9 Using the Educational System to Deliver Intervention 0.8 - PM The massive prevalence of helminthic infection and the need 0.7 to deliver treatment to a high proportion of infected individ- 0.6 uals impose significant logistic constraints on the design of drug delivery systems. Parasite control programs share with 0.5 * RSP * PSI vaccination programs the dubious distinction of being the 0.4 0 RSP 0 PSI most extensive health intervention programs ever attempted 0.3 by humankind. Anthelmintics offer significant logistic advan- 0.2 - * TM tages over vaccines, however: they are relatively thermostable * e and do not require a cold chain; they do not have specialized 0.1 - TSP TSI storage requirements; they have an extended shelf life; they do o I I I I I not require technical expertise for administration. Their dis- 0 2 4 6 8 advantage, when compared with most vaccines in current use, Cost per capita (in U.S. Dollars) is the necessity for repeated delivery, although as discussed above, the putative antihelminth vaccines may require re- peated administration and be mote costly to deliver than Note: Abbreviated treatment strategies are described in table 7-12. atedminis. Source: Authors. anthelmintics. 154 Kenneth S. Warren and others The main logistic concern, therefore, is the development of stages achieved by removing worms from the heavily infected methods to deliver anthelmintics to the community on a children, an effect that had been predicted on theoretical regular basis. One solution is to focus drug delivery on school- grounds (Anderson and May 1982). age children, using the existing educational infrastructure as a It is also now recognized that treatment of children has a delivery mechanism. Treating schoolchildren for worms is not long-term developmental effect on the whole community. a new idea; it appears to be part of the traditional perception Physical and intellectual retardation during childhood has of worm infection in many communities that it is the children consequences for the adult life of the affected individuals. who should be treated. Recent research, however, has shown Successful control of parasites in children would, therefore, thatschool-basedtreatmentoffersabroadrangeofadvantages, be expected to provide tangible benefits into the next including the potential to achieve community-wide control. generation. In this section we consider school-based treatment as a way of Targeting treatment at one specific age group obviously controlling geohelminthiasis and schistosomiasis. We take as offers significant cost advantages over universal treatment. given the finding of the preceding subsection that screening is When the target is schoolchildren these advantages are further likely to be economically unattractive and that the appropriate increased by the accessibility of the target group (King and strategy will involve mass application to groups targeted by others 1989; Wilkins 1989). Schools offer an existing infra- geography and age. structure for low-cost delivery of anthelmintics and often have On therapeutic grounds, children are the natural targets for established mechanisms for clinical surveillance and commu- treatment because they tend to have the more intense infec- nity mobilization. They also offer the opportunity for long- tions and hence are at greater immediate risk of morbidity. It term surveillance of randomly selected children with respect is now also recognized, however, that disease in childhood has to reinfection rates and the effects of treatment of selected potentially long-term consequences because the peaks of in- parasites on such indexes as height- and weight-for-age and tensity, and thus morbidity, occur at an age which is crucial for hemoglobin level. Poor school attendance either generally or physical and intellectual development. Recent studies have seasonally may vitiate the effectiveness of this approach. shown that the growth of children with intense worm infec- tions may be seriously impaired; furthermore, they have shown Cost-Effectiveness of Mass Chemoprophylaxis that these effects are primarily due to worm infection, because they can be reversed by deworming (Stephenson 1987; Cooper Earlier we assessed the economics of screening prior to treat- and others 1990). It is particularly significant that deworming ment for various helminthiases and concluded that, for a robust alone is sufficient to improve the nutritional status of the range of assumptions conceming the costs of screening and children and that great restoration of growth can be achieved treatment, mass treatment was a preferable population-based without nutritional supplementation. This was graphically approach to use of diagnostic screening to ascertain which illustrated in a recent study showing a remarkable increase in individuals required treatment. Mass treatment, however, height and weight in children treated for S. hematobium infec- could (and should) be targeted to specific geographical regions tion in Kenya (Stephenson and others 1989a). There is similar evidence that worm infection affects intellectual development Table 7-14. Costs of a Ten-Year School-Based (Nokes and others 1992), and the common physical conse- Program Covering 1,000 Children with quences of infection, such as anemia and stunting, are signifi- Chenwprophylaxis for Intestinal Helminthiasis cantly correlated with deficits in cognition and educational and Schistosomiasis measures (Pollitt 1990). The potential effect of helminthiasis - - - - - - - on intellectual development is a great cause for concern, Comonent Cost (U.S. dollars) because this may prevent children in many developing coun- Praziquantel (4,000 doses) 2,000 tries from benefiting from basic education-the only educa- Albendazole (8,000 doses) 1,000 tion they may ever receive. The United Nations Educational, Delivery system 5,000-15,000 Scientific, and Cultural Organization (UNEsco) has under- Total 8,000-18,000 taken the important new initiative of assessing the effect of helminthiases on the intake of basic education. Cost per child 8-18 Clearly the treatment of school-age children is beneficial in Note: Prcigram seeks to reach all children age five to fourteen through an health and developmental terms. Repeated treatment of chil- annuial school visit by a health team. In the ten years of the program, each dren also appears to offer.benefits to the wider community by child should receive praziquantel five times and albendazole ten times. It is dren also appears to offer benefits to the wider community by assumed, however, that the program is 80 percent effective; on average each reducing the rate of transmission of infection to adults. A child will receive four dosesof praziquantel and eight doses of albendazole. school-based chemotherapeutic program in the Caribbean not The cost per child contact in WHO's Expanded Programme of Immunization averages about $2.50. That is an upper bound (probably a multiple) of what only achieved a substantial reduction in childhood infection school-based systems are likely to cost. with enteric nematodes, as might be expected, but also reduced The present value of costs would be slightly less than these estimates if costs infection in adults, less than 4 percent of whom were treated over rhe ten-year period were discounted back to the initial year (at 3 percent), and ohers1989. It s prbabl tha thiseffet wa due as shiould be done in principle. The effect is small, however, compared with (Bundy and others 1989). It is probable that this effect was due the uncertainty in delivery system costs. to the significant reduction in contamination by infective Source: UNICEF 1989. Helminth Infection 155 Table 7 15. Estimated Gains in Healthy Life-Years achieved through sanitation and hygiene but far less costly. As from a School-Based Program of Chemoprophylcxis an interim strategy, therefore, mass chemoprophylaxis is highly for Intestinal Helminthiasis and Schistosomiasis cost-effective and, at present, the case is strong for rapid and (years) widespread adoption. Maderate Low Because of the requirement for repeated drug administra- Mechanism est-rnate" estimatea tion, intervention is best considered as a long-term program. Reduction in mild-to-moderate The example we take here assumes a ten-year intervention Reduchistionsmild-to-moderate designed to control schistosomiasis and the intestinal helmin- Reduction inmild-to-moderate thiases in the school-age population (age five to fourteen intestinal helminthiasish 225 90 years). This combination would be appropriate for populations Reduction in heavy in Northern, Western, and Southem Africa, where a combi- schistosomiasisc 300 150 nation of albendazole and praziquantel is used. At low marginal Reductlon in heavy Intestinal cost, ivermectin could be added to the albendazole and prazi- Reduction In child morralityd 125 65 quantel used in northern and eastern South America, Central Post-intervention health benefits Africa, and Southeast Asia; indeed, given that delivery costs for target group' 150 0 are preponderant among all costs, the cost of serving any of the During-intervention benefits for five groups depicted on the map would vary little from our cost families of target groupf 150 0 estimates in table 7-14. In the table, we assume a school-based Total 1,475 545 delivery system (probably provided by an annual visit to the Total1,45 5 4 - school by a mobile team) that serves the entire school-age Note: Gains in DALY over a ten-year period in a cohort of 1,000 school-age population. A more detailed discussion of school-based deliv- children, estimated from the program assumptions of table 7-14. ery systems for health intervention may be found in Jamison a. Inrervenrion is assumed to result in a gain of 1.2 healthy life-vears per year in moderate estimate and 0. I life-vear in low estimate for heavily infected and Leshe (1990). The estimates In table 7-14 suggest that it children. Comparable figuresfor mild to mnoderate infection are 0.05 and 0.02 would be likely to cost between $8.00 and $18.00 per child for ifte-years. a ten-year program (that is, between $0.80 and $1.80 per year) h. Assumes that 45 tercent of children are affected before interxvention. c. Assumes that 15 percent of children are affected before intervention. that provided, In that period, four administrations of d. Asstimes 200,000 deaths in all age groups per year from schistosomiasis praziquantel and eight of albendazole. This should control the and all intestinal helminthiases for the low estintare and 400,000 for the schistosomes and hookworm very well and Trichunis and Asca- moderate; 50 percent of these deaths occur in school-age children in the 40 percent most at risk (in which the cohort of 1,000 students occurs.) These ris moderately well (table 7-1 1). assumptions posit the occurrence of an average of 2.3 deaths over ten years Table 7-15 indicates both moderate and low estimates of the among 1,000 children in the low estinmate and 4.6 deaths in the moderate benefits of such an intervention for a population of 1,000 estimate. l)iscounting at 3 percent from a life expectancy of sixty-five gives schol-ag chid n tereni on ih a a ran of the estimates of life-years gained by averring these deaths. school-age children. There is, as shown in the table, a range of e. In the moderate case, it is assumed that health benefirs for each child benefits; explicit assumptions are made about how these can from the ten-year intervention in the postintervention years would equal 15 be expressed in disability-adjusted life-years (DALYs) gained. percent of the benefits accruing during intervention. f. In the moderate case, it is assumed rhat reducing infection in school- These assumptions are, in our view, conservative; further, the children will partially internipt transmission, thereby redticing morbidity in results (at a broad level) are relatively insensitive to the other members of the children's families. The effect is conservatively assumed assumptions. The implications of alternative assumptions, in- to be 15 percent as important as the direct effect on schoolchildren. Source: Authors. cluding assumptions about local epidemiology, can easily be assessed. The cost estimates in table 7-14 and the effectiveness esti- and age groups. Given the substantially greater intensity of mates in table 7-15 yield a range ofcost-effectiveness estimates infection among school-age groups than others (figure 7-2), from $6 to $33 per DALYgained (table 7-16). Even the high end targeting this group becomes a natural priority; the attractive- ness of this strategy is further increased by the (relative) logistic Table 7-16. Cost-Effectiveness of a School-Based and cost attractiveness of intervention through the schools, as T a m of C o phylaxi s or Stinal was discussed above. Our purpose in this section is to provide Program of Chemoprophylaxis for Intestinal an extended hypothetical example of the cost-effectiveness of H delminthiasis and Schistosomi)sis school-based intervention. Although hypothetical, the cost _ _ _ __dollars) and effectiveness parameters used are intended to span a rea- Cost per child for ten- Cost per disability-adjusted life-year gained sonable range of probable values; the range of cost-effective- year intertentiona Moderate effectb Low effectb ness estimates is likely, therefore, to provide a realistic sense of 8 6 15 what can be achieved. 18 12 33 Given the inevitability of reinfection until high levels of - - - hygiene are ultimately achieved, a strategy based on mass a. These costs are the low and high ends of the range reported in table chemoprophylaxis requires periodic, repeated administration b. The program effect cites the moderate and low estimates for table 7-15. of anthelmintics. This is less desirable than prevention Source: Authors. 156 Kenneth S. Warren and others of this range of costs is relatively attractive-comparable to provide for large-scale treatment without costly and time-con- diphtheria-pertussis-tetanus (DPT)-plus-polio immunization suming diagnostic procedures. The administration of lower (Jamison and others, chapter 6, this collection). At the low dosages of certain of the drugs might reduce or eliminate end of the range, the intervention becomes one of the most symptoms without necessarily drastically reducing efficacy. cost-effective means of promoting child health. Use of lower dosages might also be less costly. It is, in this context, worth recalling the estimates of cost- The design, implementation, and evaluation of treatment effectiveness of several preventive interventions that were regimens are complicated by the fact that most individuals in previously presented. Although applicable in only limited foci, the tropics are infected with not one but several helminths. improved water supply to control dracunculiasis is also highly Symptomatic manifestations, not surprisingly, usually repre- cost-effective-perhaps $25 per healthy life-year gained. Vec- sent a cumulative burden of the effects of multiple infections. tor control for prevention of onchocerciasis is far less attractive Studies which dissect out the proportionate contributions to at perhaps $300 per healthy life-year gained. illness of the different helminths are few indeed and often difficult to interpret. Even if done, it would be impossible to Priorities and Conclusion generalize results from one geographic area to another, given the array of variables which are present. The availability of a series of broad-spectrum anthelmintics Given this quandary, and the immense global burden of which have minimal side effects and are administered orally in multiple helminthic infection, it would be worthwhile to a single dosage has begun to be widely appreciated. Coupled consider pilot programs in which two or even three chemo- with the unique biology, ecology, and epidemiology of the therapeutic agents are administered concomitantly at periodic macroparasites (helminths) as compared with microparasites intervals and on a large scale to selected population groups to (viruses, bacteria, protozoa), these new anthelmintics allow a evaluate the cost and effectiveness of such programs in dimin- strategy of controlling disease, rather than infection, by main- ishing symptomatic disease. Available data suggest that such taining worm burdens at low levels via chemoprophylaxis. programs should be cost-effective, but confirmation is re- How best the new drugs can be employed, with what effect, quired. Such a strategy would be consonant with a transforma- and at what cost are all questions which are beginning to be tion which has begun to occur in the provision of a number of addressed. An important question is that, although the health services indevelopingcountries. Community-wide pro- helminthiases are the most widespread infections in the devel- grams for vaccination, oral rehydration therapy, vitamin A oping world, is their control warranted in so many countries supplementation, and family planning have been growing in which today are operating under severe budget constraints and number and extent, proving to be far more effective than in an environment where prospects for significant assistance programs directed solely to individuals who present themselves from donors for any health program are discouraging? The at health centers or hospitals. Such community-based programs, introduction of any new health intervention has to be weighed directed to the control of helminthic diseases, have only begun carefully, as to both the benefits and the costs, as well as to the to be explored but, where conducted, they have proved to be prospects for obtaining needed resources. most successful. Community-wide programs for anthelmintic Available data, as reviewed in this chapter, document the drug administration might be targeted at school-age children, prevalence of helminthic infections, usually with several being for example, and be conducted simultaneously or at least present simultaneously. Substantial improvements in control in concert with other community-based programs for vita- through prevention is not an option in the short to medium min A administration or vaccination, at marginal additional term without unrealistically large investments in sanitation or cost. vector control. (Dracunculiasis provides an important excep- Programs are needed to determine with certainty that two tion to this generalization; preventive measures appear both or three anthelmintic agents can be given concomitantly or cost-effective and capable of eradicating the disease during the within a circumscribed time period with safety and without loss 1990s.) For the foreseeable future, therefore, chemotherapy is of efficacy. Mechanisms for involving, motivating, and educat- the dominant option in most countries. Traditional practice ing effectively the populations concerned are required, as they would call for evaluating the new chemotherapeutic agents by are with other community-based programs. High-risk popula- their ability to cure diagnosed infections in individuals. Diag- tions requiring treatment would need to be identified. This nosis and treatment of infection in individuals, however, are could be done inexpensively by sample surveys rather than by not only costly procedures, but their application is beyond the a total screening of the population. Evaluation of selected capacity of most health services. Moreover, where infection is regimens-for example, mass treatment of school-age children rife, amelioration of disease symptoms is apt to be of short yearly-could be conducted in a similar manner. duration because of rapid reinfection. The important departure, however, is to conceptualize the Given the fact that infections in many areas are so wide- potential use and evaluation for these new chemotherapeutic spread, although not infrequently asymptomatic, alternative agents in a new framework. Given their cost, particularly practices should be considered with two objectives in mind: where therapeutic doses that are lower than usual could be to diminish the incidence of symptomatic infection and to employed, and given the burden ofhelminthic infection, large- Helmmth Infection 157 scale mass treatment could represent an important and cost- Behm, C. A., and C. Bryant. 1985. "The Modes of Action of Some Modem effective health strategy in many parts of the world where Anthelmintics." In Anderson and Waller, eds. Resistance in Nematodes to improved sanitation and vector control are not presently fea- Anthelminmc Drugs. Glebe, Australia: CSIRO Publication. Borgers, M., S. De Nollin, A. Verheyen, M. De Brabander, and D. Thienpont. sible. Estimates of cost-effectiveness suggest that measures of 1975. "Influence of the Anthelmintic Mebendazole on Microtubules and this sort are likely to prove attractive within the present range Intracellular Organelle Movement in Nematode Intestinal Cells." Aiencan of options for improving child health. Journal of Veterinary Research 36:11 53-66. Bradley, D. J., and F. S. McCullough. 1974. "Egg Output Stability and the Epidemiology of Schistosoma haermatobium." Transactions of the Royal Society Notes of Tropical Medicine and Hygiene 80:706-18. Bundy. D. A. P. 1986. "Epidemiological Aspects of Trichuris and Trichuriasis Preparation of this chapter was supported financially by the World Bank in Caribbean Communities." Transactions of the Royal Society of Tropical and by Global Health Partners, Inc. The views expressed do not, of course, Medicine and Hygiene 80:706-18. necessarily reflect those of the sponsoring organizations. The authors wish to . 1988. "Population Ecology of Intestinal Helminth Infections in express their gratitude to the following individuals for valuable advice and Human Communities." Philosophical Transactions of the Royal Society of comments on earlier drafts: J. Cook, C. Cruz, D. Evans, T. Evans, R. Feachem, London 321:405-20. D. Hopkins, J.-L. Lamboray, W. H. Mosley, M. Percopo, A. Seim, G. Tor- Bundy, D. A. P., and U. Blumenthal. 1989. "The Impact of Human Behaviour rigiani, and S. Van der Vynckt. on the Epidemiology of Schistosomiasis and Geohelminthiasis." In G. Bernard and J. Behnke, eds., Behaviour and Parasinsm. London: Taylor and Francis. References Bundy, D. A. P., and E. S. Cooper. 1989. "Trichuris and Trichuriasis in Humans." Advances in Parasitology, 28:107-73. Albers-Schoenberg, G., B. H. Arison,J. C. Chabala, A. W. Douglas, P. Eskola, Bundy, D. A. P., E. S. Cooper, D. E. Thompson, R. M. Anderson, and J. M. M. H. Fisher, A. Lusi, H. Mrozik, F. L. Smith, and R. L. Tolman. 1981. Didier. 1987. "Age-related Prevalence and Intensity of Tnchuns mchiura "Avermectins: Structure Determination."Journal of the American Chemical Infection in a St. Lucian Community." Transactions of the Royal Society of Association 102:4216-21. Tropical Medicine and Hygiene 81:85-94. Albiez, E. J., G. Walter, A. Kaiser, P. Ranque, H. S. Newland. A. T. White, Bundy, D. A. P., S. P. Kan, and R. Rose, and others. 1988a. "Age-related B. M. Greene, H. R. Taylor, and D. W. Buttner. 1988. "Histological Prevalence, Intensity and Frequency of Distribution of Gastrointestinal Examination of Onchocercomata after Therapy with Ivermectin." Tropical Helminth Infections in Urban Slum Children from Kuala Lumpur, Malay- Medicine Parasitology 39:93-99. sia." Transactions of the Royal Societyt of Tropicaf Medicine and Hygiene Anderson, R. M. 1982. Populaton Dynaamics of Infectious Diseases: Theory and 82:289-94. Applications. London: Chapman and Hall. Bundy, D. A. P., E. S. Cooper, D. E. Thompson, J. M. Didier, and 1. Simmons. - 1986. "The Population Dynamics and Epidemiology of Intestinal 1988b."EffectofAgeandInitiallnfectionStatusontheRateofReinfection Nematode Infections." Transactions of the Royal Society of Tropical Medicine with T. trichiura after Treatment." Parasitology 97:469-76. and Hygiene 686-96. Bundy, D. A. P., E. S. Cooper, D. E. Thompson, ]. M. Didier, R. M. Anderson, - 1987. "Determinants of Infection in Schistosomiasis." Clinical Tropi- and 1. Simmons. 1988c. "Predisposition to Trichuris mrichiura Infection in cal Medicine and Communicable Diseases 2:279-300. Humans." Epidemiology and Infectiont 98:65-71. -. 1988. "DeterminantsofInfection inHumanSchistosomiasis."Clnical Bundy, D. A. P., D. E. Thompson, E. S. Cooper, M. H. N. Golden, and R. M. Tropical Medicine and Communicable Diseases 2:279-300. Anderson. 1985. "Population Dynamics and Chemotherapeutic Control .1989 "Transmission Dynamics of Ascaris lumbricoides and the Impact of Trichuris trichiura Infection of Children in Jamaica and St. Lucia." of Chemotherapy." In D. W. T. Crompton, M. C. Nesheim & Z. S. Transactions of the Royal Society of Tropical Medicine and Hygiene 69: Pawlowski, eds., Ascanasis and Its Prevention and Control. London: Taylor 759-64. and Francis. Bundy, D. A. P., M. S. Wong, L. L. Lewis and J. Horton. 1989. "Control of Anderson, R.M. and R.NI. May. 1982. "Population Dynamics of Human Gastro-intestinalHelminthsby,Age-GroupTargetedChemotherapyDeli'.- Helminth Infections: Control by Chemotherapy." Nature 297:557-63. ered through Schools." Transactions of the Royal Society of Tropical Medicine - 1985. "Helminth Infections of Humans: Mathematical Models, Pop- and Hygiene 8:115-20. ulation Dynamics and Control." Advances in Parasitology 1-101. Cabrera, B. D., B. Caballero, L. Rampal, and W. de Leon. 1989. "The Anderson, R. M., R. M. May, and S. Gupta. 1989. "Non-linear Phenomena Philippines." In D. W. T. Crompton, M. C. Nesheim, and Z. S. Pawlowski, indHost-Parasitenteractions."oParasitolog99:S59-S79. eds., Ascariasis and Its Prevention and Control. London: Taylor and in HostParasie Inteaction." Parsiroloy 99:S9-S79.Francis. Anderson, R. M. and G. F. Medley. 1985. "Community Control of Helminth Infections of Man bv Mass and Selective Chemotherapy." Parastofogv Caimcross, A. C., and R. G. Feachem.1983. EnvironmenIHealehEngsneenng 90:629-60. mi the Tropics. Chichester, U.K.: John Wiley and Sons. Andreano, R., and T. Helminiak. 1988. "Economics, Health and Tropical Campbell, W. C., M. H. Fisher, E. 0. Stapley, G. Albers-Schoenberg, and T. Disease: A Review." In H. N. Herrin, and P. Rosenfield, eds., Economics, A. Jacob. 1983. "Ivermectin: A Potent New Antiparasitic Agent." Science Health and Tropical Diseases. Manila: University of the Philippines, School 221:823-28. of Economics. Chandler, A. C. 1925. "The Rate of Loss of Hookworms in the Absence of Arfaa, F. 1986. "Ascariasis and Trichunasis." In J. A. Walsh and K. S. Warren, Reinfections." IndianJournal of Medical Research 13:625-34. eds., Strategiesfor Pnrmary Health Care. Chicago: University of Chicago Press. CIBA Foundation. 1987. Filariasis, ed. 1). Everard and S. Cook. Chichester, Barbour, A. D. 1982. "Schistosomiasis." In R.M. Anderson, ed., Population U.K.: John Wiley and Sons. Dynamics of Infectious Diseases: Theory and Appbcations. London: Chapman Cook, J. A. 1987. "Strategies for Control of Human Schistosomiasis." Clinical and Hall. Tropical Medicine and Communicable Diseases 2:449-63. 158 Kenneth S. Warren and others Cook, J. A., P. Jordan, and P. Armitage. 1976. "Hycanthone Dose-Response Ascaris lumbricoides Infection." Transactons of the Royal Society of Tropical in Treatment of Schistosomiasis Mansoni in St. Lucia." AmericanJournal of Medicine and Hygiene 80:774-92. Tropical Medicine and Hygiene 25:602-7. Evans, T. G., and C. J. L. Murray. 1987. "A Critical Re-examination of Cooper, E. S., and D. A. P. Bundy. 1986. "Trichuriasis." In A. S. McNeish and Blindness Prevention under the Onchocerciasis Control Programme." So- J. A. Walker-Smith, eds., Diarrhoea and Malnutrition in Childhood. London: cial Science and Medicine 25:241-49. Butrersworth. Feachem, R. G. A., D. J. Bradley, H. Garelick, and D. D. Mara. 1983a. 1987. "Trichuriasis: In Intestinal Helminthic Infections." Bailliere's Sanitaaon and Disease. Chichester, U.K.: John Wiley and Sons. Clinical Tropical Medicine and Communicable Diseases 2:629-43. Feachem, R. G. A., M. W. Guy, S. Harrison, K. 0. Wugo, T. Marshall, N. 1988. "Trichuris Is Not Trivial." Parasitology Today 4:301-6. Mbere, R. Muller, and A. M. Wright. 1983b. "Excreta Disposal Facilities Cooper, E. S., D. A. P. Bundy, and F. S. Henry. 1986. "Chronic Dysentery, andIntestinalParasitisminUrbanAfrica:PreliminaryStudiesinBotswana, Stunting, and Whipworm Infestation." Lancet (August 2):2808I. Ghana and Zambia." Transactions of the Royal Society of Tropical Medicine Cooper, E. S., D. A. P. Bundy, T. T. Macdonald, and M. H. N. Golden. 1990. and Hygiene 77:515-21. "Growth Suppression in the Tnchuns Dysentery Syndrome." European Fisher, M. H., and H. Mrozik. 1989. "Chemistry." In W. C. Campbell, ed., Journal of Clinical Nutrition 44:138-47. Ivermectin and Abamectin. New York: Springer Verlag. Cort. W. W. 1922. "A Graphic Analysis of Certain Factors in Hookworm Fosdick, R. 1952. The Story of the Rockefeller Foundation. New York: Harper & Control." American Journal of Tropical Medicine 2:449-63. Brothers. Croll, N. A., R. M. Anderson, T. W. Gyorkos, and E. Ghadirian. 1982. "The Gemmell, M. A., J. R. Lawson, and M. G. Roberts. 1986. "Control of Echino- Population Biology and Control of Ascans lumbricoides in a Rural Commu- coccosis[Hydatidosis: Present Status of World-Wide Progress." Bullenn of the nity in Iran." Transactions of the Royal Society of TropicalMedicine and Hygiene World Health Organization 64:333-39. 76:187-97. Greene, B. M., K. R. Brown, and H. R. Taylor. 1989. "Use of Ivermectin in Crompton,D.W.T. 1985."ChronicAscariasisandMalnutrition."Parasitologv Humans." In W. C. Canspbell, ed.. Ivermectin and Abamectnn. New York: Today 1:47-52. Springer Verlag. - . 1989. "The Prevalence of Ascariasis." In D. W. T. Crompton, M. C. Grove, D. 1989. StrongyLoidiasis. London: Taylor and Francis. Nesheim, and Z. S. Pawlowski, eds., Ascariasis and Its Prevention and Control. Hairston, N. G. 1965. "On the Mathematical Analysis of Schistosome Popu- London: Taylor and Francis. lations." Bullenn of the World Health Organization 33:45-62. Cupp, E. W., A. 0. Ochoa, R. C. Collins, F. R. Ramberg, and G. Zea. 1989. Harpham, T., T. Lusty, and P. Vaughan. 1987. In the Shadow of the City: "The Effect of Multiple Ivermectin Treatments on Infection of Simulium Community Health and the Urban Poor. Oxford: Oxford University Press. ochraceum wvith Onchocerca voldulus." American Journal of Tropical Medicine and Hygiene 40:501-6. Hayashi, S. 1980. "Economic Loss from Parasites." Collected Papers on the Control of Soil-TranLsmitted Helminthiases. Tokyo: Asian Parasite Control Dadzie, K. Y., A. C. Bird, K. Awadzi, H. Schulz-Key. H. M. Gilles, and M. A. Organization. Aziz. 1987. "Ocular Findings in a Double-Blind Study of Ivermectin versus Diethylcarbamazine versus Placebo in the Treatment of Onchocerciasis." Hill, R. B. 1926. "The Estimation of the Number of Hookworms Harboured, Bntish Journal of Ophthalnology 71:78-85. by the Use of the Egg Dilution Method." American Journal of Tropical Davis, Andrew. 1986. "Available Anthelmintics and Future Needs." In H. MedicmeandHygiene6:19-41. Machleidt, ed., Contributions of Chemistry to Health: Proceedings of the Fifth Holland, C. V. 1 987a. "Hookworm Infection." In L. S. Stephenson, ed., Impact CHEMRAWN Conference, vol. 2. Wemheim, Fed. Rep. Germany: VCH of Helhninth Infections on Human Nutrition. London: Taylor and Francis. Verlagsgesellischaft mbH. - 1987b. "Neglected Infections-Trichuriasis and Strongyloidiasis." In - 1989. "Chemotherapy: Options for Delivery Systems." In Max J. L.S. Stephenson, ed., Impact of Helminth Infections on Human Nutrition. Miller and E. J. Love, eds., Parasitic Diseases: Treannent and Control. Boca London: Taylor and Francis. Raton, Fla.: CRC Press. Hopkins, D. R., and E. Ruiz-Tiben. 1990. Dracunculiasis Eradication: Target Degremont, A., G. K. Lwihula, C. Mayombana, E. Bumier, D. de Savigny, and 1995. Atlanta: Carter Center. American Journal of Tropical Medicine and M. Tanner. 1987. "Longitudinal Stidy on the Health Status of Children in Hygiene. 43:296-300 a Rural Tanzanian Community: Parasitoses and Nutrition Following Con- llegbodu, V. A., 0. 0. Kale, R. A. Wise, B. L. Christensen, J. H. Steele, and trol Measure against Intestinal Parasites." Acta Tropica 44:175-90. L. A. Chambers. 1986. "Impact of Guinea Worm Disease on Children in Dietz, Klaus. 1982. "The Population Dynamics of Onchocerciasis." In R. M. Nigeria." AmericanJournal of Tropical Medicine and Hygiene 35:962-94. Anderson, ed.. Population Dynamics of Infectous Diseases: Theory and Appli- Ismid, 1. S., and B. Rukmono. 1980. "The Effect of Latrine Provision and cantons. London: Chapman and Hall. Health Education on Soil Pollution." In Collected Papers on the Control of Edungbola, L. D., S. J. Watts, T. 0. Alabi, and A. B. Bello. 1988. "The Inipact Soil-TransmittedHelminthiases, vol. 1. Tokyo: Asian Parasite Control Orga- of a UNICEF-Assisted Rural Water Project on the Prevalence of Guinea nization. Worm Disease in Asa, Kwara State, Nigeria." Amencan Journal of Tropical Jamison, D. T., and J. Leslie. 1990. "Health and Nutrition Considerations in Medicine and Hygiene 39:79-85. Educational Planning: The Cost and Effectiveness of School-based Inter- Egerton, J. R., D. Suhayda, and C.H. Eary. 1988. "Laboratory Selection of vention." Food and Nutriton Bulletin 12:204-14. Haemonchus contortus for Resistance to Ivermectin."Journal of Parasitology Kan, S. P., H. L. Guyatt, and D. A. P. Bundy. 1989. "Geohelminth Infection 74:614-17. of Children from Rural Populations and Urban Slums in Malaysia." Trans- el Karim, M. A., K. J. Collins, J. R. Brotherhood, C. Dore, J. S. Weiner, M. Y. acutons of the Royal Society of Tropical Medicine and Hygiene 83:817-21. Sukkar, A. H. Omer, and M. A. Amin. 1980. "Quantitative Egg Excretion Kass, 1. S., A. 0. W. Stretton, and C. C. Wang. 1984. "The Effects of and Work Capacity in a Gezira Population Infected with Schistosoma Avermectin and Drugs Related to Acetylcholine and 4-aminobutyric Acid mansoni." American Journal of Tropical Medicine and Hygiene 29:54-61. on Neurotransmission in Ascans suum." Molecular and Biochemical Parasitol- Elkins, D. B., M. Haswell-Elkins, and R. M. Anderson. 1986. "The Epidemi- ogy 13:213-25. ology and Control of Intestinal Helminths in the Pulicat Lake Region of Kass, 1. S., C. C. Wang, J. P. Walrond, and A. 0. W. Stretton. 1980. Southem India: Study Design and Pre- and Post-treatment Observations on "Avermectin Bia, a Paralyzing Anthelminthic that Affects Intemeurons Helminth Infection 159 and Inhibitory Motor Neurons i Ascaris." Proceedings of the National Acad- Nokes, C., S. M. Grantham-McGregor, A. W. Sawyer, E. S. Cooper, and D. emy of Science 77:6211-15. A. P. Bundy. 1992. "Helminth Infection and Cognitive Function." Proceed- Kilama, W. L. 1989. "Sanitation in the ColItrol Of Ascariasis." In D. W. T. ings of the Royal Society (London) 247:77-8 1. Crompton, M. C. Nesheim, and Z. S. Pawlowski, eds., Ascariasis and Its Ottesen, E. A., V. Vijayasekaran,V. Kumaraswami, S. V. Perumal Pillai, A. Prevention and Control. London: Taylor and Francis. Sadanandam, S. Frederick, R. Prabbakar, and S. P. Tripathy. 1990. "A King, C. A., D. W. Wiper, K. V. Destigter, P. A. S. Peters, D. Koech, J. H. Controlled Trial of Ivermectin and Diethylcarbamazine in Lymphatic Fil- Ouma,T. K. A. Siongok, and A. A. F. Mahmoud. 1989. "Dose-findingStudy ariasis." New EnglandJournal of Medicine 322:1113-18. for Praziquantel Therapy of Schistosoma hematobium in Coast Province, Otto, G. F., W. W. Cort, and A. E. Keller. 1931. "Environmental Studies of Kenya." AmericanJournalof TropicalMedicine and Hygiene 40:507-13. Families in Tennessee Infested with Ascaris, Trichuris, and Hookworm." Kumaraswami, V., E. A. Ottesen, V. Vijayasekaran, S. Uma Devi, M. AmericanJournal of Hygiene 14:156-93. Swaminathan, M. A. Aziz, G. R. Sarma, R. Prabhakar, and S. P. Tripathy. Partono, F. 1985. "Diagnosis and Treatment of Lymphatic Filariasis." Parasi- 1988. "Ivermectin for the Treatment of Wuchereria bancrofti Filariasis: tologs Today 1:52-57. Efficacy and Adverse Reactions."JAMA 259:3150-53. Partono, F., Pumomo, and A. Soewarta. 1979. "A Simple Method to Control Kvalsvig, J. D. 1988. "The Effects of Parasitic Infection on Cognitive Perfor- Brugia timori by Diethylcarbamazine Administation." Transacuons of the mance." Parasitology Today 4:206-8. Royal Society of Tropical Medicine and Hygiene 73:536-42. Lacey, E. 1986. "The Biochemistry of Anthelmintic Resistance." In Anderson Paul. J. E., R. B. Isley, and G. M. Ginsberg. 1986. "Cost-Effective Approaches and Wallet, eds., Resistance in Nematodes to Anthelmintic Dr-ugs. Glebe, to the Control of Dracunculiasis." WASH Technical Report 38. USAID Office Australia: Commonwealth Scientific and Industrial Research Organization. of Health, Washington, D.C. Pawlowski, Z. S. 1984. "Ascariasis." Annales de la Societe' BeIge de M&licine Lemma, A. 1987. "Overview of Endod Studies." In Endod 11: Report of the Tropicale64 125-34. Second International Workshop on Endod. New York: uLJNICEF ClEPA. Le Rickhc W. H. 1967. "World Incidence and Prevalence of the Major Pawlowski, Z. S., and A. Davis. 1989. "Morbidity and Mortality in Ascariasis." LRChe,unicabH. Di967ses."Wo In cidWoenceoand Pealence O'Cofn Th eds In D. W. T. Crompton, M. C. Nesheim, and Z. S. Pawlowski, eds., Ascanasis Communicable Diseases." In 0. Wolstenholme and Mi. O'Connor, eds., n t rvninadCnrl odn alradFacs Health of Mankind. London: J. and A. Churchill. and Its Prevennon and Control. London: Taylor and Francis. Peters, W. 1978. "The Relevance of Parasitology to Human Welfare Todav." Levin, N. 1988. "Controlling Iron-Binding Deficiency Anemia: A Cost-Ben- Symposium of the British Society of Parasitolopy 16:25-40. efit Analysis." World Health 27-29. Lok, J. B., T. Harpaz, and D. H. Knight. 1988. "Abnormal Patterrms of Polderman, A. M., B. Gryseels, and P. DeCaluwe. 1988. "Cure Rates and Egg k bryo, Tnesis In D' D.f H TKngted . i9 h . " rmal" JPate of Reduction in Treatment of Intestinal Schistosomiasis with Oxainniquine Helmbnthologesi i r i tT d ee " n and Praziquantel in Maniema, Zaire." Transacuons of the Royal Society of Hebninthology' 62:175-80. ~~~~~~~Tropical Medicine and Hygiene 82:115-16. McGarvev, S. T., B. L. Daniel, M. Tso, G. Wu, S. Zhong, R. Olveda, P. M. Wiest, and G. R. Olds. 1990. "Child Growth and Schistosomiasis Japonica in the Philippines and China." Paper presented at the meeting of the Pollitt, E., P. Hathirat, P., and N. J. Kotchabharkdi. 1989. "Iron Deficiency American Society for Clinical Investigation,Washington, D.C. Clnical and Educational Achievement in Thailand." Amerncan Journal of Clinical Research 38. Nutrinon 50:687-97. Mligasena, S., and H. M. Gilles. 1987. "Hookworm Infections." Cinical and Prescott, N. M. 1979. "SchistosomiasisandDevelopment." WorldDevelopment Tropical Medicine and Communicable Diseases 2:617-2 7. 7:1-14. Miller, T. A. 1978. "Industrial Development and Field Use of the Canine . 1987. "'The Econoimics of Schistosomiasis Chemotherapy." Parasitol- Hookworm Vaccine." Advances in Parasitology 16:33 3-42. ogy Today 3:21-24. Morishita, K. 1980. "Japanese Literatures Conceming Influence of Parasites, . 1989. "Economic Analysis of Schistosomiasis Control Projects." In Especially of Soil-Transmitted Helminths upon the Psychosomatic Condi- M. W. Service, ed., Demography and Vector-borne Diseases. Boca Raton, Fla.: tion in Matemity and Childhood." In Collected Papers Otn the Control of CRC Press. Soil-TransmittedHelminthiases. Tokvo: Asian Parasite Control Organization. Prescott, N. M., and M. F. Jancloes. 1984. "The Analysis and Assessment of Morrow, R. H. 1984. "The Application of a Quantitative Approach to the Health Programns." Social ScieTice and Medicine 19:1057-60. Assessment of the Relative Importance of Vector and Soil-Transmitted Prichard, R. K. 1970. "Mode of Action ofthe Anthelmintic Thiabenda-zole in Diseases in Ghana." Social Sciences and Medicine 19:1039-49. Haemonchus contortus." Nature 228:584-85. Mossinger, J., H. Schulz-Key, and K. Dietz. 1988. "Emergence of Onchocerca Prost, A. and N. Prescott. 1984. "Cost Effectiveness of Blindness Prevention volvulus Microfilaria from Skin Snips before and after Treatment of Patients by the Onchocerciasis Control Programme in Upper Volta." Bulletn of the with Ivermectin." Tropical Medicine and Parasitology 39:313-16. World Health Organization 62:795-802. Mott, K. E. 1988. "Schistosomiasis Control." In D. Rollinson and A. Simpson, Rajagopalan, P. K., P. K. Das, S. Subramanian, P. Vanamail, and K. D. Ramaia. eds., The Biology of Schistosomes. London: Academic Press. 1989. "Control of Bancroftian Filariasis in Pondicherrv, South India. Pre- Mrozik, H., P. Eskola, M. H. Fisher,J. R. EgerLon, S. Cifelli, and D. A. Ostlind. control Epidemiological Observations." Epidemiology and Infecton 103:685- 1982. "Avermectin Acyl Derivatives with Anthelminthic Activity." J. 92. Med. Chem. 2 5:658-63. Richard-Lenoble, D., M. Kombila, E. A. Rupp, E. S. Papayliuo, P. Gaxotte, C. Naquira, C., G. Jiminez, J. G. Guerra, R. Bernal, D. R. Nalin, D. Neu, and M. Nguiri, and M. A. A.iz. 1988. "Ivermectin in Loiasis and Concomitant 0. Aziz. 1989. "Ivermectin for Human Strongyloidiasis and Other Intestinal volvulus and M. perstans Infections." Amencan Journal of Tropical Medicine Helminths." American Journal of Tropical Medicine and Hygiene 40:304-9. and Hygiene 39:480-83. Newland, H. S., A. T. White, B. M. Greene, S. A. D'Anna, E. Keyvan- Rim, H. J. 1986. "The Current Pathobiology and Chemotherapy of Clonor- Larijani, M. A. Aziz, P. N. Williams, and H. R. Taylor. 1988. "Effect of chiasis." KoreaniournalofParasitology 24:1-141. Single-dose Ivermectin Therapy on Human Onchocerciasis volvulus Infec- Rosenfield, P. L., F. Galladay, and R. K. Davidson. 1984. "The Economics of tion with Onchocercal Ocular Involvement." BntishJournal of Ophtiilmol- Parasitic Diseases: Research Priorities." Social Science and Medicine 19:117- ogy 72:561-69. 126. 1 60 Kenneth S. Warren and others Sasa, M. 1977. Human Filariasis: A Global Survey of Epidemzolog's and Control. Taylor, P., H. M. Murure, K. Manomano. 1988. "Efficacy of Low Doses of Tokyo: University of Tokyo Press. Praziquantel for Schzstososa mansonm and S. haematobium."Journalof Tropical Schad, G. A., and K. S. Warren. 1989. Hookworm Infecton: Current Status and Medicine and Hygiene 91:13-17. New Direcuons. London: Taylor and Francis. Terry, C.. D. A. P. Bundy, and J. Horton. 1989. "Relative Costs of Targeted, Seim, A. R. 1990. Guinea Worm Disease Eradicaaon Project. Fagerstrand, Selective, and Selected Chemotherapeutic Control of Geohelminths." Norway: Health and Development Intemational. PERG/SKF. Mimeo. Sl WG1u Alabama." Tumer, M. J., and ]. M. Schaeffer. 1989. "Mode of Action of Ivermectin." In Smillie, W. G. 1924. "Control of H4 okwor9 Disease in SouthA W.C.Campbell, ed., IvermectinandAvermectin.New York:SpringerVerlag. Southens Medical Journal 17:494-99. Smith, G. S., D. Blum, S. R. A. Huttly, N. Okeke, B. R. Kirkwood, and R. G- UNICEF (United Nations Childrens Fund). 1989. Essential Drugs Pnce List. A. Feachem. 1989. "Disability from Dracunculiasis: Effect on Mobility." Copenhagen: Supply D tvision. Annals of Tropical Medicine and Parasitology 83:151-58 Wakelin, D. 1978. "Immunitv to Intestinal Parasites." Nature 273:617-20. Soboslay, P. T., H. S. Newland, A. T. White, K. D. Er-ttmann, E. J. Albie:, H. Walsh, J. A. 1990. "Estimating the Burden of Illness in the Tropics." In K. S. Warren and A. A. F. Mahmoud, eds., Tropical and Geographical Medicine. R. Taylor, P. N. Williams, and B. M. Greene. 1987. "Ivermectin Effect on New A. F. Microfilaria of Onchocerca volvulus after a Single Oral Dose in Humans." New York: McGraw-Hill. Trop.Med.ParasitoL 39:8-10. Walsh, J. A., and K. S. Warren. 1979. "Selective Primary Health Care: An Interim Strategy for Disease Control in Developing Countries." New Eng- Soderlund, D. M., P.M. Adams, andJ. R. Bloonsquist. 1987. "Difference in landJournalofMedicine 301:967-74. the Action of Avermectin Bla oBn the GABA Receptor Complex of Mouse Aciond Rat." Bvermeticand BJa ionhesc A Researc r Communications Warren, K. S. 1973. "Regulation of the Prevalence and Intensity of Schisto- 146:692-98. somiasis in Man: Immunology or Ecology?" Journal of Infectious Diseases Srephenson9 L.2S.9198. ImactoJ elrmthInecuns n HmanNumon 127:595-609. Stephenson, L. S. 1987. Impact of Helinrith Infections on Humian Nutntson. -___ . 1989. "Selective Primary Health Care and Parasitic Diseases in London: Tay'lor and Francis. McAdam, KPR." In K. P. R. McAdam, ed., New Strategies in Parasitology Stephenson, L. S., M. C. Latham, K. M. Kurz, and S. N. Kinotti. 1989a. London: Churchill Livingstone. "Single Dose Metrifonate or Praziquantel Treatment in Kenyan Chil- Warren, K.S., and A. A. F. Mahmoud. 1990. Tropical and Geographical Medi- dren. Effects on Growth in Relation to Schistosoma hematobium and cine. 2d ed. New York: McGraw Hill. Hookworm Egg Counts." American Journal of Tropical Medicine and Hygiene 41: 445-53. Warren, K. S., ]. H. Ouma, T. Arap Siongok, and H. B. Houser. 1978. "Hycanthone Dose-Response in Schistosoma mansoni Infection in Kenya." Stephenson, L. S., M. C. Latham, K. M. Kurtz, S. N. Kinoti, and H. Brigham. Lancet 1:352-54. 1989b. "Treatment with a Single Dose of Albendazole Improves Growth of WHO (World Health Organization). 1980. Sxth Report on the World Health Kenyan Schoolchildren with Hookworm, Trichuris trichiura, and AscaTis (o e Oation..P190. SxthoeporAon hesWrldnealt lumbricoides Infections." Arnericatn Journal of Tropical Medicine and Hygiene Situaton. Part 1, Global Analysis. Geneva. 41:78-87. . 1985. "The Control of Schistosomiasis." Technical Report Series 728. Stephenson, L. S., S. N. Kinoti, K. M. Kurz, and H. Brigham. 1990. "Im- Geneva. provements in Physical Fitness of Kenyan Schoolboys Infected with Hook- . 1987a. Atlas of the Global Distribution of Schistosomes. Geneva. worm, Trichuris trichuria and Ascaris lumnbicoides Following a Single Dose of . 1 987b. "Prevention and Control of Intestinal Parasitic Infections." Albendazole." Transactions of the Royal Society for Tropical Medicine and Technical Report Series 749. Geneva. Hygiene 84:277-82. Wilkins, H. A. 1989. "Reinfection after Treatment of Schistosome Infec- Stoll. N. R. 1947. "This Wormy World."Journalof Parasitology 33:1-18. tions." Parasitology Today 3:83-88. Taylor, H. R., R. D. Semba, H. S. Newland, E. Keyvan-Lanjani, A. White, Z. Yekutiel, P. 1980. "Eradication of Infectious Diseases." In M. A. Klinsberg, ed., Dukuly, and B. M. Greene. 1989. "Ivermectin Treatment of Patients with Contnbutions to Epidemiology and Biostatistics, vol. 2. Basel: Karger. Severe Ocular Onchocerciasis." Amnencan Journal of Tropical Medicine and Young, M. E., and A. Prost. 1985. "Child Health in China." World Bank Staff Hygiene 40:494- 500. Working Paper. Washington, D.C. 8 Measles Stanley 0. Foster, Deborah A. McFarland, and A. Meredith John Measles is a highly infectious disease transmitted person-to- tacts are with an infectious person. Together, these factors person by way of the respiratory route; the severity of measles determine not only the likelihood that an individual will be (morbidity, disability, and mortality) is affected by a wide infected but also the age pattern of infection and whether range of epidemiologic, demographic, physiologic, socioeco- measles is maintained endemically in the population or occurs nomic, and behavioral determinants. Almost all children un- only in sporadic outbreaks or epidemics. protected by immunization will be infected with measles; in the developing world, 1 to 5 percent will die of measles and its Susceptibility to Infection complications. In 1954, Enders and Peebles isolated the measles virus, Measles transmission results from the exposure of a susceptible which paved the way for the development of an effective person to respiratory droplets or aerosolized droplet nuclei from vaccine. Measles vaccine provides long-term protection to a measles-infected person (Black 1982; Bloch and others susceptible persons. Use of this vaccine has proved effective in 1985). The probability that an exposed, susceptible person in reducing measles cases in both industrial and developing coun- a household will be infected by the measles virus is 90 percent tries. With technical direction and leadership from the World or higher. Health Organization's (WHO's) Expanded Programme on Im- Most infants are protected from measles at birth by passively munization (EPI), advocacy and financial support from the acquired transplacental maternal antibodies. Breastfeeding United Nations Children's Fund (UNICEF), and bilateral tech- practices affect neither the level nor the persistence of measles nical cooperation, national immunization programs have in- antibodies. Studies of 2,917 matemal blood samples from creased the global coverage of measles vaccine for children twenty different populations in thirteen countries demon- twelve months of age or younger from 20 percent in 1974 to strated measles antibodies in 99.2 percent of samples (Black 78 percent in 1990 (WHO 1991). Although global efforts effec- 1989), suggesting that a corresponding proportion of newborn tively prevented an estimated 2.12 million measles deaths in infants have some degree of passively acquired antibody im- 1990, an estimated 880,000 deaths were not prevented. In munity. The mean duration of this protection varies consider- 1989 the WHO World Health Assembly established measles ably, however, ranging from three to six months in some targets for achievement by the year 1995: a 95 percent reduc- populations to twelve months or more in others. Black identi- tion in measles mortality, a 90 percent reduction in measles fied three factors contributing to these interpopulation differ- incidence, and a measles vaccine coverage of 90 percent in the ences: (a) "the women of different countries have different first year of life. In this chapter we review what is currently amounts of measles antibody to pass to their children"-ma- known about measles and its control and identify policies, ternal titers as measured by hemagglutination inhibition in strategies, and practices that will enable the achievement of Gazankulu, South Africa, were eightfold higher than those the global objectives. from Taiwan; (b) "there are genetic or environmentally deter- mined differences in the efficiency of the placenta in transport- Epidemiology ing IgG [immunoglobin G]"-that is, simultaneous collections of matemal and cord blood have shown matemal-infant dif- Measles transmission occurs when an infectious individual ferences in hemagglutination inhibition titer ranging from comes into close contact with a susceptible individual. The +0.86 log? in New Haven, Connecticut, to -0.97 1og2 in Kuala probability that an individual will contract measles in a given Lumpur, Malaysia; and (c) "there are differences in the rate at time period depends upon his or her immune status (suscepti- which children lose passively acquired antibody immunity"- bility to infection), the population size and density of the for example, differences in the half-life of maternally acquired community, the frequency of the individual's contact with antibody (Black 1989, p. 19). These differences may be related other population members, and the probability that such con- to the rate of infection with other infectious agents, leading to 161 162 Stanley 0. Foster, Deborah A. McFarland, and A. Meredith John increased catabolism of IgG, and higher rates of diarrhea, casesoccurcontinuously. Dataon 10,078measlescasesfrom leading to increased loss of IgG into the interior lumen of the the preimmunization era were collected from the Lagos gut (Black 1989, p. 19). Geographic variations in the age at Infectious Disease Hospital in Nigeria; they showed that 36 which infants lose passive immunity, as estimated by serologic percent of hospitalized cases occurred in infants twelve studies, are consistent with observed pattems of immunologic months of age or younger (the median age of infection was response to vaccine or disease on exposure to infection, fifteen months) and that 85 percent of admissions were of children less than thirty-six months old (Smith and Foster Age of Infection in Four Populations 1970). Moderate levels of vaccine coverage have not always changed this urban pattern of early infection. In Kinshasa, Individuals exposed to measles virus and not protected by Zaire, where measles vaccine coverage during 1983 in chil- either matemal or vaccine-acquired immunity usually develop dren of twelve through fifty-nine months was 62 percent, a infection. Age-specific rates of measles infection are deter- community survey showed high rates of measles transmis- mined by the number of infective measles cases in the popula- sion in infants and young children as indicated by the tion, the size and age composition of the pool of susceptibles, following age distribution of cases: six through eight and the rate and age pattern of contact. Widely divergent age months, 18 percent; nine through eleven months, 19 per- patterns of measles transmission are seen in the following four cent; one year, 40 percent; and two years, 10 percent (figure populations: (a) high-density urban populations in developing 8-1) (Taylor and others 1988). An estimated 77 percent of countries (for example, Kinshasa, Zaire); (b) rural populations cases occurred prior to age three. in developing countries with low vaccine coverage (for exam- ple, Matlab, Bangladesh); (c) rural populations in developing RURAL-IN DEVELOPING COUNTRIES WITH LOW VACCINE COVERAGE. countries with high vaccine coverage (for example, Lesotho); In rural areas, where contact between young, susceptible chil- and (d) populations in industrial countries with high vac- dren and infectious individuals is less frequent than in urban cine coverage such as the United States of America (figures areas, measles is primarily a disease of childhood. Community 8-1-8-4). surveillance data from Matlab, Bangladesh, collected prior to the introduction of measles vaccine, showed 23 percent of HIGH-DENSITY URBAN-IN DEVELOPING COUNTRIES. In large cit- cases occurring in children under two years, 34 percent in ies in westem and central Africa, an urban transmission pat- children two and three years old, 22 percent in children four tem prevails; measles occurs primarily in the first two years of and five years old, and 22 percent in children six to ten years life. The early age of infection can be attributed to the high old (figure 8-2) (Koster and others 1981). Although the pop- population density, the early exposure of infants to infectious ulation density in Bangladesh isone ofthe highest in the world, individuals as mothers carry their babies on their backs on the relative isolation of rural enclaves, the riverine geography, crowded public transport and in urban markets, and the early and the limited social mobility of the traditional Moslem loss of matemally acquired antibody in relation to such loss in culture result in a low probability that the measles virus will be industrial countries. Measles is endemic in the population and introduced into a village and thus a low probability of exposure Figure 8-1. Age Distribution of Measles, Figure 8-2. Age Distribution of Measles, Kinshasa, Zaire, 1983 Matlab, Bangladesh, 1975-76 Age distribution (percent) Cumulative distribution (percent) Age distribution (percent) Cumulative distribution (percent) 50 100 50 100 40 - - 80 40 - - 80 30 - - 60 30 - - 60 20 - - 40 20 - 40 10 - 20 10 - 20 0 0 0 0 0-5 6-8 9-11 1 2 3 4 months months months year years years years 0-1 2-3 4-5 6-10 Age (years) Age (years) Source: Taylor and others 1988. Source: Koster and others 1981. Measles 163 Figure 8-3. Age Distribution of Measles, Figure 8-4. Age Distribution of Measles, Lesotho, 1988 United States, 1989 Age distribution (percent) Cumulative distribution (percent) Age distribution (percent) Cumulative distribution (percent) 50 100 50 100 40 80 40 - 80 30 - - 60 30 - - 60 20 -40 20 -40 10 20 10 - 20 0 0 0 0 <1 1 2 3 4 5-9 10 <1 1-4 5-9 10-14 15-19 20-24 25-29 30 Age (years) Age (years) Source: Lesotho 1990. Source: MMWR 1990. of susceptibles to measles infection. In such areas, measles Clinical Illness occurs in sporadic epidemics and vanishes between outbreaks. Measles is a clinical illness easily recognized both by health RURAL-IN DEVELOPING COUNTRIES WITH HIGH VACCINE COVERAGE. workers and by experienced family members, and it frequently Many developing countries are achieving 80 percent measles has a distinct name in the local language. The disease has been immunization during the first year of life, the target established well described by Preblud and Katz: after an incubation period for 1990. In countries with low population density-for exam- of ten to twelve days, "the prodromal stage is heralded by the ple, Lesotho-significant reductions in measles incidence and onset of fever, malaise, conjunctivitis, coryza, and tracheo- a corresponding change in the age pattem of disease have been bronchitis manifesting as cough, and it lasts for 2-4 days.... documented. The age at onset of measles has increased; 60 The temperature rises during the ensuing 4 days and may be as percent of cases occur in children over five years of age high as 40.6" C.... The rash is an erythematous maculopapular (Lesotho 1990) (figure 8-3). This change is not being seen in eruption that usually appears 14 days after exposure and the more densely populated developing countries with similar spreads from the head to the extremities over 3-4 days. Over levels of vaccine coverage, such as Rwanda, Burundi, and the next 3-4 days, the rash fades in the order of appearance. Malawi. Desquamation can be detected in areas of greatest involve- ment" (Preblud and Katz 1988, p. 183). INDUSTRIAL COUNTRIES WITH HIGH VACCINE COVERAGE. In the Naturally occurring measles infection provides lifetime pro- United States, where, in the prevaccine era, measles was tection against reinfection. This was clearly demonstrated in primarily a disease of children, childhood immunization cou- the 1846 outbreak in the Faroe Islands, where infection was pled with mandatory school immunization has reduced measles limited to those under age sixty-five, individuals bom after the incidence by 98 percent (Markowitz and Orenstein 1990). The last measles epidemic in 1781 (Panum 1939). national goal of measles elimination has, however, been frus- In severe disease, a frequent occurrence in developing coun- trated by outbreaks affecting urban preschoolers, who are tries, the manifestations of clinical illness reflect the epithelial primarily unvaccinated, and high school and college students, loci of infection as illustrated in Morley's classic diagram of who, as a group, are highly vaccinated. The latter outbreaks severe measles: eyes (conjunctivitis), larynx (laryngitis), lungs represent transmission among the 2 to 5 percent who are not (pneumonia), and gastrointestinal tract (diarrhea [Morley 1973, protected by a single dose of measles vaccine (United States, p. 2141; figure 8-5). National Vaccine Advisory Committee 1991). Complications Measles Infection and Its Cost Most measles deaths are attributed to complications, which While measles is recognized as an acute childhood infection, may be acute (within one month) or delayed (one month to the long-term costs in terms of morbidity, disability, and mor- one year). In industrial countries-for example, the United tality are less well understood. States-the most commonly cited complications are otitis 164 Stanley 0. Foster, Deborah A. McFarland. and A. Meredith John Figure 8-5. Clinical Manifestations of Severe diarrhea and measles come from Bangladesh, where a large Measles outbreak of measles occurred in twelve villages among 5,775 children undergoing prospective surveillance for nutrition and diarrhea (Koster and others 198 1). The frequency and duration of diarrheal episodes increased beginning one week before and lasting four weeks after the onset of rash. Fifty-one percent of Conjunctivitis the diarrheal episodes lasted longer than seven days compared with 25 percent of the diarrheal episodes of those who did not Sore mouth have measles. The case-fatality rate (CFR) for those with mea- Laryngitis sles who had diarrhea episodes longer than seven days (11.9 Bronchopneumonia percent) was significantly greater than the cFR for those with measles but without diarrhea (4.0 percent). Children with postmeasles diarrhea had a significant and prolonged (10 per- cent) deficit in weight-for-height. In a WHO-sponsored review assessing potential interventions to reduce diarrhea morbidity and mortality, it was projected that measles immunization could prevent 0.6 to 3.8 percent of all diarrheal episodes and 6 to 26 percent of all diarrheal deaths (Feachem and Koblinsky 1983). Enteritis Postmeasles pneumonia is the main cause of measles- associated mortality in the developing countries; 56 percent of measles-associated deaths in a community outbreak in Uttar Pradesh in India and 92.8 percent of measles-associated deaths in a hospital in Ilorin, Nigeria, were attributed to pneumonia (Fagbule and Orifunmishe 1988; Narain and others 1989). In the Sri Lanka community survey (CFR, 1.3 percent), 44 percent of measles-associated mortality was attributed to pneumonia, 25 percent to diarrhea, 19 percent to convulsions, and 9 percent to coma (Bloch, de Silva, and de Sylva 1983). Measles is responsible for a significant proportion of acute respiratory morbidity and mortality, 6 to 21 percent of the morbidity and 8 to 93 percent of the mortality (Markowitz and Nieburg 1991). Source: Morley 1979. Studies have documented transient post-measles immuno- suppression (Whittle and others 1973; Dossetor, Whittle, and Greenwood 1977; Kaschula, Druker, and Kipps 1983). Hussey media (6 percent), diarrhea (6 percent), pneumonia (4 per- and Simpson (1991) have identified immunosuppression as a cent), measles encephalitis (0.2 percent), subacute sclerosing probable cause of increased risk of nosocomial bacteremia in encephalitis (I in 100,000 measles cases), and death (0.1-0.2 measles cases (3.37 per 100 hospital admissions), as compared percent) (Preblud and Katz 1988; Atkinson and Markowitz with nonmeasles pediatric admissions (0.57 per 100 admis- 1991). Fifteen percent of the reported cases in the United sions). General immunosuppression following measles is an States required hospitalization. important factor in measles-associated mortality. Autopsy The distribution of complications in developing countries studies, which revealed serious nonbacterial bronchiolar and is somewhat different. Using active surveillance to identify interstitial necrosis caused by adenovirus, measles virus, and measles cases in the community, investigators of 2,386 cases of herpes virus, indicate a failure of the immune mechanism in measles in Sri Lanka documented complication frequencies as the postmeasles child (Kipps and Kaschula 1976). follows: diarrhea, 37 percent; respiratory infections, 30 per- The main causes of long-term disability following measles cent; ear infections, 7 percent; and convulsions 2 percent have been identified as blindness and malnutrition. In Africa, (WHOEP[ Sri Lanka 1985). Fifty-seven percent of cases had where the prevalence of blindness in preschool children is medical care (Bloch, de Silva, and de Sylva 1983). estimated at I in 1,000, measles has been identified as respon- Morley's classic study from Imesi-lle, Nigeria, was the first sible for half of childhood blindness (Foster and Johnson to document the long-term effect of measles on child health 1988). in developing countries. In that study, 25 percent of the children with measles lost 10 percent or more of body weight Mortality (Morley 1973). The time required to regain that weight ranged from 4.5 weeks for children with no diarrhea to 8.1 weeks for In some developing countries, measles cFRs are 300 times those those with diarrhea. Further data on the relation between currently reported in industrial countries. Such rates were Measles 165 common in Europe in the 1800s. In the well-documented forty-eight months to seventy-one months, 4.2 percent (Koster outbreak in Sunderland, England, in 1885, 25 of 311 measles and others 1981). patients died, a CFR of 8 percent (Drinkwater 1885). Commu- nity studies of measles outbreaks between 1961 and 1978 show GENDER. In two studies from Asia, cFRs were higher for cFRs from 1.5 to 15 percent (Walsh 1983; Cutts 1990; Cutts females than for males. In Bangladesh, the cFR among males and others 1991). High mortality from measles was initially (0.98 percent) was significantly less than for females (2.64 thought to occur only in Africa, but high cFRs from measles percent) (Bhuiya and others 1987). Similar results were re- have also been reported in Asia and Latin America, for exam- ported from Varanasi in India: 1.3 percent for males and 3.3 ple, 7 percent in Tamil Nadu, 5.7 percent in Uttar Pradesh, percent for females (Chand and others 1989). Such differences and 4.5 percent in Guatemala (Gordon 1965; John and others have not been documented in African studies, suggesting that 1980; Narain and others 1989). differences arise from sex-specific patterns of child care and Although measles outbreaks that result in high mortality response to illness rather than to biological differences be- are more likely to be investigated and reported, data from tween the sexes. prospective surveillance among populations that have been studied in developing countries have also revealed high SOCIOECONOMIC STATUS. Using a multivariate logistic re- measles CFRs: 6.5 percent in Kenya; 6.1 percent in Zaire; gression analysis of community data collected in the Matlab 6.5 percent in Senegal; and 3.7 percent in Bangladesh Bangladesh field area, Bhuiya and others (1987) identified (Voorhoeve and others 1977; Kasongo Project Team 1981; low number of household articles owned (a proxy indicator Koster and others 1981; Garenne and Aaby 1990). The of poverty) as a significant risk factor for measles mortality. World Health Organization estimates that 880,000 measles deaths occurred in 1990 (WHO 1991). INTENSITY OF EXPOSURE: ACQUISITION OF MEASLES IN THE HOME. Aaby, in his studies in Guinea-Bissau, documented increased Mortality Risk Factors rates of measles mortality in secondary cases acquired in the home (Aaby 1988). Reexamination of the Machakos data High infant and child mortality in the developing world is from Kenya and ORSTOM data from Senegal has shown similar usually attributed to the complex interaction of poverty, un- findings (Aaby and Leeuwenburg 1988; Garenne and Aaby dernutrition, and infection. Mosley and Chen have proposed 1990). In the Senegal data, odds ratios (OR) on mortality risk a child mortality proximate determinant model which empha- were related to the probable intensity of exposure: same hut sizes the importance of the complex interactions of maternal (OR 3.8, 95 percent confidence interval ci, 1.7-8.4), same behavior, environmental contamination, nutritional deficiency, household (OR 2.3, 95 percent [Ci], 1.0-5.7), and same com- injury, and personal illness control action including preven- pound (OR 1.9, 95 percent ci, 0.6-6.0). tive measures and seeking treatment at time of illness (Mosley and Chen 1984). Epidemiologic studies have documented a NUTRITIONAL STATUS. Although high mortality from mea- number of risk factors which explain, in part, the high rates of sles is seen in undernourished populations, individual nutri- measles-associated mortality in the developing world. tional status has not proved to be a reliable predictor of mortality in most studies (Aaby 1988; Koster and others AGE. Case-fatality rates exhibit substantial variations by age, 1981). During the last decade, vitamin A deficiency has the highercFRs occurring in the younger age groups, usually six been increasingly linked to higher child mortality and to to eighteen months of age. The age pattern of mortality varies high measles-associated mortality (Sommer 1990). In Tan- in different populations, reflecting both the epidemiology of zania, a clinical trial showed decreased measles mortality in the disease and the general health environment in which the hospitalized patients who received vitamin A (Barclay, Fos- child exists. In a recent outbreak in Rwanda, reported age- ter, and Sommer 1987). In Zaire, a multivariate logistic specific cFRs decreased with age: zero through eight months regression model of 283 measles patients admitted to two (3.0 percent); nine months through twenty-three months (1.4 Kinshasa hospitals identified an increased mortality risk in percent); twenty-four months through fifty-nine months children less than two years of age with a vitamin A level (1.0 percent); and sixty months and older (0.5 percent) of less than 5 micrograms per deciliter (relative risk [RR] (Weierbach 1989). In contrast, the Kasongo study from Zaire 2.9-cl 1.3-6.8 [Markowitz and others 1989]). In South Af- documented the highest rates in children of thirteen months rica, a randomized double-blind trial using vitamin A in the to twenty-four months: one month to six months, 0 percent; treatment of 189 measles cases reduced measles mortality by seven months to twelve months, 6.2 percent; thirteen months half; the durations of pneumonia, diarrhea, croup, and hos- to twenty-four months, 9.8 percent; twenty-five months to pitalization were shortened (Hussey and Klein 1990). Of thirty-six months, 4.3 percent; and thirty-seven months to historical note is Hussey's reference to a preantibiotic-era sixty months, 2.8 percent (Kasongo Project Team 1981). Data paper by Ellison (1932), in which intensive vitamin therapy from Bangladesh show similar cFRs from measles in all age reduced measles CFR from 8.7 percent to 3.7 percent and groups: one month to twenty-three months, 4.4 percent; measles pneumonia CFR from 67.7 percent to 31.3 percent. twenty-four months to forty-seven months, 4.2 percent; and Measles and its sequelae have also been identified as signif- 166 Stanle-y 0. Foster, Deborah A. McFarland, and A. Meredith John icant risk factors contributing to the development of protein- habilitation) for measles patients; they also estimated the calorie malnutrition and kwashiorkor. number of deaths that would have occurred in the absence of a measles immunization program. The analysis demonstrated ABSENT OR DELAYED MEDICAL CARE. Most of the mortality large societal benefits of measles immunization with a benefit- associated with measles can be prevented through timely and to-cost ratio of 100 to 1. Indirect costs were estimated to be appropriate medical care. In Senegal, early treatment of mea- approximately 77 times the direct costs, but the authors ac- sles has resulted in the near elimination of measles mortality knowledge that the direct-cost measure represents a very low (Garenne and others 1992). Timely appropriate case manage- level of access to care and a severe underreporting of measles ment is rare in many developing countries because of lack of cases. A previous study (Verduzco, Calderon, and Velazquez- access to care, delayed seeking of care, lack of trained person- Franco 1974) of measles immunization in Mexico calculated a nel, or lack of appropriate drugs. Contact with health facilities benefit-to-cost ratio of 27 to 1, although indirect costs (as usually occurs late in the illness. In a hospital-based study from measured by lost earnings) were not estimated. Burkina Faso, 55 percent of measles deaths occurred within Estimates of both direct and indirect costs attributable to twenty-four hours of admission (Sahuguede and others 1989). measles were made in a study comparing Israel, the West Bank, The symptoms upon admission to the hospital of 714 measles and Gaza (Ginsberg and Tulchinsky 1990). Total costs for patients illustrate the seriousness of illness at the time of entry patients with simple cases of measles, patients requiring hospi- into the hospital: dehydration (91 percent), diarrhea (64 per- talization, and patients with complications were estimated in cent), conjunctivitis (56 percent), fever above 39.5 C (50 each of the three regions. Estimates for a simple case of measles percent), respiratory infection (46 percent), and cardiovascu- (treated in the outpatient setting) ranged from $13 in Gaza to lar collapse (34.5 percent). $141 in Israel.' Costs of early mortality due to measles ranged from $11,628 in Gaza to $76,518 in Israel. The wide range in LOCAL TREATMENT. In many traditional cultures, measles is the latter estimates highlights one of the most difficult meth- considered a normal event. In others, it has been attributed to odological issues in making estimates of premature mortality- the work of witches or sorcerers (Imperato and Traore 1969). that of placing a monetary value on life. Any of the available Withholding food, especially protein, and fluids from measles methods relates valuation of life to social productivity and cases has been reported (Morley 1973). In the Machakos data, is usually measured in discounted future expected earnings. measles was identified among "God's diseases" (Maina- Thus, the value of a life in Gaza, a region with low earnings Ahlberg 1979). Withholding of water and milk from children and income, is valued below a life in Israel, an area with higher sick with measles was documented in 62 percent of 242 cases. per capita income. Of cases reported, 50 percent received indigenous medicines The economic cost of measles in industrial countries has and 98 percent also received Western medicine. Local treat- been measured in more detail. Using 1983 data in the United ments, restriction offluidsandfood, delay in access to effective States, White, Koplan, and Orenstein (1985) estimated a chemotherapy,anduseofpotentialtoxicsubstanceshavebeen benefit-to-cost ratio for measles immunization of 11.9 to 1. identified as potential contributors to increased measles- Both direct and indirect costs were estimated with and without associated mortality. a vaccination program. Indirect costs were 3.2 times direct costs. The cost per measles case was estimated to be $209 (in NONVACCINATION. The single most important determinant 1983 dollars). Cost-benefit studies of measles immunization of measles morbidity and mortality is vaccination status. Al- haveconsistentlydemonstrated largesocial benefits (Mast and most all children unprotected by measles vaccine will eventu- others 1990) because of the high direct cost of treating com- ally be infected with measles and I to 5 percent will die. plications of measles cases, the attendant indirect costs of work and productivity loss, and the relatively low cost of immuniza- Economic Cost tion programs. The economic burden of measles can also be measured by The economic cost of a disease can be divided into direct and days of healthy life lost as a result of premature mortality and indirect costs. Direct costsare thosebome by the health system disability. In Mali (Duflo and others 1986), measles ranked in the prevention and treatment of the disease and by house- fifth among diseases in terms of days of healthy life lost, with holds or individuals in seeking preventive services or treat- 94.7 percent of the days lost because of premature death (as ment. Indirect costs are usually measured in terms of lost opposed to illness or disability). Losses resulting from measles productivity of workers as a result of their premature death or accounted for 6.4 percent of the total days of healthy life lost disability. in Mali. An earlier study in Ghana (Ghana Health Assessment Few estimates of the costs for treatment of measles have been Project Team 1981) ranked measles second, accounting for 7.3 made in developing countries. Because care is generally de- percent of the total days of healthy life lost in the population. layed or absent, reliable data on costs of treatment for measles Barnum (1989) notes the importance of applying a discount cases is not routinely available. In a study in Mexico, Cardenas- factor in order to account for the fact that the number of Ayala and others (1989) estimated the costs of medical care healthy life-years lost which are attributed to a disease each (hospitalization, physician's visits, medical treatment, and re- year do not actually occur in that year. The choice of discount Measles 167 factor is thus critically important when ranking diseases vaccinated at nine to eleven months and subsequently in those by productive life lost. Using the Ghana data, Barnum vaccinated at twelve to fourteen months, the minimum age of shows that the relative ranking of measles among diseases immunization was increased to twelve months in 1965 and is second when the discount rate is zero but is fifteenth then to fifteen months in 1976 (Orenstein and others 1986). when the discount rate is 0.20. When the discount rate is In 1989, a two-dose measles vaccine schedule was recom- zero, the diseases with the greatest cost in lost productivity mended (ACIP 1989). The United States experience empha- are the diseases of childhood, such as measles, but as the sizes the importance of epidemiology, disease surveillance, and discount rate rises, adult problems increase in importance outbreak investigation in setting and amending national vac- and childhood diseases fall in significance. cine policies. In Lesotho, where the age for measles immunization is nine Prevention of Measles months and coverage has reached 80 percent, 60 percent of cases occur in school-age children. Serologic studies using Measles can be prevented through immunization of the suscep- enzyme-linked immunosorbent assay have documented 13.6 tible child with a potent live virus vaccine. percent seronegativity in six- and seven-year-old children entering school (Lesotho 1990). When a policy of vaccinating Vaccine all schoolchildren for the first time, regardless of vaccine history, proved difficult to maintain, a second dose of measles The history, uses, and effectiveness of measles vaccines are vaccine was added to the routine scheduled booster dose of the discussed below. diphtheria-pertussis-tetanus (DvT) at fifteen months. Immunization with potent vaccine administered at the rec- HISTORY. Measles virus was first isolated in the 1950s by ommended age does not ensure seroconversion or protection. Enders and Peebles (1954) from a child infected with measles Primary vaccine failures (the lack of a serologic and immuno- and was attenuated through passage in tissue culture. Most logic response to initial immunization) do occur. Secondary of the vaccine strains used today (Schwarz, Moraten, Becken- vaccine failures (the occurrence of disease in previously suc- ham, Edmonston-Zagreb, EKC, and AIK-C) were developed from cessfully immunized children) have been reported but are the original Edmonston isolation (Preblud and Katz 1988). thought to be rare. In a vaccine study population in British The Leningrad-16 strain used in the former U.S.S.R. and Columbia, 93 percent of 188 children responded serologically eastern Europe, the strains used in China, and the CAM-70 to immunization. This percentage corresponds to a primary strain were derived independently of the Edmonston isolation vaccine failure rate of 7 percent. In 1985-86 an outbreak of (Clements and others 1988). Heat stability of most strains has measles occurred in the same British Columbia study popula- been increased through the addition of stabilizers. The mini- tion; 9 of the 175 original seroconverters developed measles, mum recommended dose of current standard measles vaccine corresponding to a secondary vaccine failure rate of 5 percent applied at or after the age of nine months is 1,000 median tissue (Mathias and others 1989). Low rates (2 to 5 percent) of culture infective dose given subcutaneously in the arm. secondary vaccine failure have also been documented by other authors (Edmonson and others 1990; Markowitz, Preblud, AGE OF IMMUNIZATION. Because of differences in the duration Fine, and Orenstein 1990). of protection from passive maternal protection and differences in risk of exposure, selection of the age of immunization VACCINE EFFICACY. Orenstein and colleagues (1985) have requires a balancing of two factors: "the earliest age at which outlined a range of methods for the field evaluation of vaccine high rates of seroconversion can be obtained, and the age group efficacy (VE), including screening methods that can be used at with the greatest risk of infection" (Orenstein and others health facilities, outbreak investigations, and case control 1986). On the basis of epidemiologic data on age-specific studies. Several factors limit vaccine efficacy. They include the measles incidence and age-specific seroconversion data, WHO following: has recommended nine months as the optimal age for measles withvaccinevirusreplicationbyprenatally * Interference thvcm ru elctobyrnaly immunization in most developing countries (Kenyan Ministry of Health and WHO 1977; WHO/EPI Kenya 1979). In Haiti, acqure ternal aibody seroconversion to a standard dose of Schwarz vaccine ranged * Exposure to wild virus infection prior to the recom- from 45 percent in six-month-old children to 100 percent in mended or actual time of immunization children at twelve months of age (Halsey and others 1985). In * Impotent vaccine resultingfrom failure ofthe cold chain some industrial countries, 100 percent seroconversion is not (the system designed to ensure vaccine potency from site of obtained until children are fifteen months of age. In the manufacture, through shipment, to central storage, to dis- United States, policy recommendations regarding age for mea- tribution, to peripheral storage, to dilution with cold dilu- sles immunization have changed three times in response to ent, to vaccine delivery) field data on vaccine efficacy. Measles immunization was ini- * Incorrect administration of measles vaccine, for exam- tially introduced in children at nine months of age. When ple, administration of less than the required 0.5 cubic cen- challenge with wild virus identified vaccine failures in children timeters or immunization at an inappropriate age 168 Stanley 0. Foster, Deborah A. McFarland, and A. Meredith John * Failure of immunologic response of a susceptible person WHO and endorsed at the 1990 World Summit for Children (90 to potent vaccine for unknown reasons percent reduction in measles morbidity and 95 percent reduc- tion in measles mortality) were estimated for the Task Force In industrial countries, vaccine efficacy in children vacci- on Child Survival at $15 per child for up to 80 percent nated at twelve months to fifteen months of age is high. In coverage and then an additional $1 for each 1 percent increase Poland, measles vaccine efficacy has been estimated at 97 incoverageupto$30for95 percentcoverage (Forgyandothers percent (WHO/EPI Poland 1986). In a measles outbreak investi- 1990). The authors of two studies conducted in Swaziland five gated in Browning, Montana, vaccine efficacy was estimated years apart, 1984 and 1989, calculated a cost per fully im- at 96.9 percent (95 percent ci 89.5-98.2) (Davis and others munized child of approximately $55 for coverage rates of 70 1987). Evidence to date indicates that live virus measles percent and 71 percent, respectively (Robertson 1985; McFar- immunizationalso induces life-long immunity in most individ- land and Kraushaar 1990). Although the costs are high in uals (Markowitz, Preblud, Fine, and Orenstein 1990). In de- relation to other countries, they reflect the tradeoffs of achiev- veloping countries, where logistics and maintenance of the ing high coverage rates in a small population (approximately cold chain are difficult, seroconversion studies have occasion- 700,000) with excellent access to health services and extensive ally documented low rates of seroconversion: 40 percent in surveillance and outbreak control activities. The Swaziland Yaounde, Cameroon, and 0 percent in Guinea-Bissau studies emphasize the importance of understanding the con- (McBean and others 1976; Aaby and others 1989). Most text in which health services are delivered and the fallacy of outbreak investigations in developing countries, however, applying average cost figures for across all countries and all document rates of vaccine efficacy in the range of 70 to 90 settings. percent. Examples include community and hospital studies of The cost of the measles component of the EPI total cost can urban measles in Point Noire, Congo, which reported vaccine be determined either as an incremental cost to the total or as efficacies of 78 percent and 87 percent, respectively; a Tanza- the cost base of EPI to which other antigens are added. Using nian case-control study in which card-documented records the first method, Phillips, Feachem, and Mills (1987) calcu- revealed a vaccine efficacy of 96 percent (95 percent confi- lated the incremental cost of adding measles immunization to dence level 83-99 percent); and arecent study in rural Burundi an existing EPI program as $1.35 (in 1982 dollars). Shepherd, that reported a vaccine efficacy of 72.4 percent (Dabis and Sanoh, and Coffi (1986) used the second method in C6te others 1988; Chen 1990; Killewo and others 1991). Even d'Ivoire, allocating 75 percent of EPI costs to the measles under good cold-chain conditions, vaccine efficacy in devel- component. The cost per child vaccinated against measles was oping countries is lower than in industrial countries because $12.30 (in 1980 dollars). The authors of an earlier study in vaccine is applied at an age when 10 to 20 percent of children Zambia, using slightly different methods, derived a cost of still have maternal antibody. Because of the risk of infection $8.00 to $14.00 (in 1982 dollars) per child vaccinated at an early age, vaccination cannot be delayed. against measles in the rural areas and $2.00 to $5.00 in urban areas (Ponninghaus 1980). When estimating the cost of COST-EFFECTIVENESS: EPI. Total costs of the Expanded Pro- achieving the 1995 measles targets, Forgy and others (1990) gramme on Immunization have been estimated in a number attributed the entire cost of EPI ($15.00 per vaccinated of countries (Brenzel 1990, 1991). Although the range is child) to the measles component. Those who used UNICEF quite wide, the average cost of approximately $15.00 per mortality figures estimated total worldwide costs for achiev- fully immunized child (BCG vaccine, oral polio vaccine [opv; ing the targets to be $5.707 billion; those who used World four times], DPT [three times], and measles) appears to be Bank mortality estimates arrived at a total worldwide cost indicative of true program costs. Comparisons of the cost of $8.517 billion. per fully immunized child for alternative immunization Besides variation in methods used to assign costs to the strategies are approximately $11.00 for facility-based pro- measles component of EPI, several factors influence cost esti- grams, $10.60 for mobile programs, and $15.60 for acceler- mates, including the level of immunization activity (volume), ated strategies. the ratio of fixed costs to variable costs, prices of key inputs, Several crucial questions remain regarding the cost of im- the type of technology used, and the productivity of personnel munization programs, including the relationship of costs to providing services (Brenzel and Claquin 1991). Understand- coverage, the effect of technology on costs, the current cost ing cost behavior can assist program managers and donor levels and the ability of countries to meet stated immunization agencies in controlling these factors and thus influencing the targets (Rosenthal 1990), the distribution of costs between costs of measles immunizations. countries and donor groups, and the relative costs of alterna- Most cost studies of EPI and measles immunization focus on tive immunization strategies (for example, campaigns) and the direct cost to the system of providing the service rather sustainable increases in coverage. Furthermore, all the EPI cost than the cost to the family or household in seeking immuniza- studies have been at one point in time; they have not been tion services. Thus, cost figures routinely underestimate the conducted in conjunction with coverage surveys over time. It full societal cost of an immunization program. is thus difficult to predict changes in cost per fully immunized Given the range of cost and effectiveness estimates, it is not child as coverage increases. The 1995 targets established by surprising that cost-effectiveness measures of measles immuni- Measles 169 zation vary considerably. Several studies have attempted to EFFECT ON CHILD SURVIVAL. In 1981, Lancet published an measure cost-effectiveness of measles immunization by the article on measles in Zaire which questioned the effect of number of measles cases prevented and measles deaths pre- measles immunization on child survival: "In a zone with high vented. These estimates are compared in table 8- 1. measles case-fatality, the risk of dying between the ages of 7 Since measles contributes to morbidity and mortality from and 35 months for a vaccinated population was compared with diarrhea, the same measures of cost-effectiveness can be calcu- an unvaccinated control group. Life-table analysis for both lated for diarrheal cases and deaths prevented as a result of groups showed that measles vaccination reduced the risk of measles immunization. Phillips, Feachem, and Mills (1987) dying at the age of maximum exposure to measles. The gain in estimate the cost per diarrheal case prevented at $7 (in 1982 survival probability, however, tended to diminish afterwards dollars) and the cost per diarrheal death prevented at $143. to approach that of the unvaccinated group" (Kasongo Project Another measure of the cost-effectiveness of different diseases Team 1981, p. 33). and interventions is the number of disability-adjusted life- Although the interpretation of these data was questioned years (DALYs) added for each intervention. Using data from (Aaby and others 1981), the issue ofreplacement mortality has Cote d'lvoire and Zambia, Prescott, Prost, and Le Berre (1984) not, until recently, been adequately addressed. Several recent compared the cost-effectiveness of measles immunization with studies, however, have assessed the effect of measles immuniza- an onchocerciasis program in Upper Volta (now Burkina Faso) tion on child survival. with regard to disability-adjusted life-years added. For measles Because a definitive double-blind placebo control study immunization, the cost per DALY is $49 (in 1977 dollars) in would not be ethical, a variety of methods have been used in C6te d'lvoire and $56 in Zambia, compared with $150 for the the following epidemiologic analyses to assess the effect of onchocerciasis program. measles vaccine on child survival. In developing countries Great care should be exercised in interpreting and extend- with high mortality in children under five, measles vaccine ing the results of cost-effectiveness studies. As Brenzel and increases child survival. Claquin (1991) note, the most cost-effective intervention is not necessarily the most efficient; future costs of programs e Bangladers Using a cas trol mhodology, Clem- , ,, . , , . ,, ,. . ,. ~~~~~ens and others matched 536 deaths of children ten to sixty should cautiously be projected from cost-effectiveness studies months of age with two age and gender matched neighbor- because average costs do not remain the same over time; and hood controls. Measles immunization was associated with a overall cost savings do not necessarily accrue when the most 36 percent (95 percent ci 21-48 percent) proportionate cost-effective interventions are implemented, because re- reduction in overall mortality rate. For deaths plausibly source allocation decisions are not made solely on the basis of associated with measles (measles, pneumonia, diarrhea, and cost-effectiveness results. Findings from cost-effectiveness malnutrition), vaccine effectiveness was estimated at 57 studies are but one type of information for decisionmaking and prnt. must be weighed alongside political, ethical, organizational, percent. managerial, and other factors. * Bangladesh: Using a cohort methodology and the same Immunization programs have been the subject of many population described above, but with an additional year of cost-effectiveness analyses, perhaps because of the large follow-up,Koenigand others (1990) matched 8,135 vacci- donor investment in such programs and because of the narted-unvaccinated pairs by month and year of irth. The relatively straightforward measure of effectiveness em- mortality rate for the immunized children was 45 percent ployed. But immunization programs and other preventive less than that of the controls (P < .0001, Gehan-Wilcoxon interventions should not be subject to a standard which testX2 = 4.18). Differences were significant for all children exceeds that of other services, in particular treatment and immunized under three years of age. curative services. When cost-effectiveness analysis is em- * Guinea-Bissau: Aaby and others (1989) found that, in a ployed, it should be applied to the whole range of services population in which serological data identified a subgroup and interventions available in order to obtain a fairer assess- of children not responding to measles vaccine, subsequent ment of how all resources are used and how such resources mortality among responders to vaccine (4.8 percent) was might be more effectively allocated. significantly less than among nonresponders (13.2 percent), a threefold difference in mortality. Table 8-1. Costs per Measles Case and Death * Haiti: Using a logistic regression model, Holt and col- Prevented leagues (1990) followed up 1,381 children vaccinated to (1980 U.S. dollars) measure seroconversion rates; two and one half years later, infants who were seronegative at the time of vaccination Coun-y Caseprevented Death prevented had significantly lower mortality (1.27 percent) than that The Gambia 1.96 41 of nonvaccinated infants (6.62 percent). The adjusted odds Cote d'Ivoire 14.00 480 ratio in a multivariate stepwise logistic regression associat- Cameroon 3.30 30-60 ing measles vaccine with survival was 6.5 (95 percent ci Source: Makinen 1980; Shepherd, Sanoh, and Coffi 1986; Robertson and 1.6-27.1). Estimates of measles vaccine effectiveness in others 1987. prevention of mortality in children from nine months to 170 Stanley 0. Foster, Deborah A. McFarland, and A. Meredith John thirty-nine months of age ranged from 84.7 percent to 90 1989b). Lack of availability of large quantities of the vaccine percent. has limited the implementation of the WHO recommendation. * Senegal: Data from the Khombole study area in Senegal In Senegal, prospective follow-up of children immunized with (Garenne and Cantrelle 1986) showed that children six to high-titer EZ and Schwarz vaccines has shown increased child thirty-six months of age who had been immunized against mortality (Garenne and others 1991) . Increased mortality after measles had an overall mortality risk 31 percent lower than immunization with high-titer vaccines has also been reported nonimmunized controls (P = .028). from Guinea-Bissau and Haiti (WHO, personal communication, June 1992). The World Health Organization no longer recom- These data strongly suggest that the survival benefit of mends use of high-titer EZ vaccine (WHO/EPI 1992). Although measles vaccine is significantly greater than that predicted by there is a clear need for an effective measles vaccine for measles-specific mortality. Longitudinal data from the Medical children at six months of age, alteratives to the current EZ Research Council in the Gambia suggest a modest effect on vaccine will need to be developed. infant mortality but a more marked effect on child mortality when compared with preimmunization data from an adjoining area (Greenwood and others 1987). Management of Immunization Activities The importance of measles in decreasing child survival can also be estimated from retrospective (verbal autopsy) studies Many of the obstacles to the reduction of measles morbidity of child mortality by cause. Such studies have been used to and mortality stem from suboptimal management. Listed below assess the relative contribution of measles to overall mortality. are ten critical areas which determine, in large part, the effec- Using the criteria of age greater than 120 days and rash with tiveness of immunization programs. Each of these areas should fever for at least 3 days, Kalter and others (1990) estimated be reviewed at least annually at the national and subnational sensitivity and specificity of a diagnosis of measles as cause of level to assess the appropriateness and effectiveness of policy, death at 98 percent and 90 percent, respectively. In Sri Lanka, strategy, and implementation. Although some factors can be in an analysis of reported deaths among children six months monitored through analyses of routine data (coverage and to thirteen years, it was estimated that measles or a measles disease incidence), others require collection of data through complication was associated with 53 percent of 122 deaths supervision, surveys, outbreak investigations, or management (WHO/EPI Sri Lanka 1985). In Rangoon, Burma. verbal autopsy audits. follow-up reports of 249 deaths of children age six months to ten years, identified from death certificates, attributed 35 percent * Policy. Is the current immunization policy-for example, of the deaths to measles and its complications (WHo/EPI Burma schedule-consistent with current technical knowledge (the 1985). Reported causes of mortality attributed to measles itn- guidelines of the WHO/EPI Global Advisory Group) and the cluded pneumonia, chronic diarrhea, and malnutrition. In a in-country epidemiology ofthe Epl diseases? Within-country prospective study in Senegal during a period of eight years. policy variations may he required to meet different epide- measles accounted for 31 percent of deaths of children six months miologic situations (for example, urban slums, rural nomad to nine years of age (Pison and Bonneuil 1988). population). * Targets. Are there national targets for coverage and dis- ALTERNATIVE MEASLES VACCINES. Seroconversion rates to the ease reduction? Are these targets understandable, realistic, standard Schwarz strain are low when it is administered prior and measurable? Do local areas have the responsibility and to nine months of age, and the risk of measles infection is great authority to set local targets, to measure their achievements, in high-density areas, where as many as 30 percent of reported and to alter program implementation? measles cases may occur prior to the age of nine months. * Strategies. Are national and local strategies designed to Because of these factors, improved measles control in areas of high population density requires a vaccine which can be effec- ensure the achievement of targets? If not, what alterations tively administered before the earliest age of infection, four are needed? months to six months. Studies in Mexico by Sabin and others * Logistics. Is the national system of vaccine and equipment (1983), who used the human diploid Edmonston-Zagreb (EZ) procurement and distribution adequate to ensure the avail- vaccine strain, showed that administration of this vaccine ability of essential commodities (cold chain, potent vac- could produce seroconversion in the presence of maternal cines, sterilizers, needles and syringes, vaccination cards) at antibody. Field trials in the Gambia, Guinea-Bissau, Mexico, every immunization delivery point? and Haiti have documented the effectiveness of high-titer EZ * Training and supervision. Is there a central authority re- vaccine when administered at six months and, in some cases, sponsible for ensuring that preservice and inservice educa- four monthsofage(Whittle and others 1984; Aabyand others tion are providing current knowledge on policy, strategy, 1988; Markowitz and others 1990; Job and others 1991). In and practice? Is there a national system of performance 1989 the WHO/EPI Global Advisory Group recommended the assessment, supervision, or surveys which provides data on introduction ofhigh-titer EZ vaccine in areas where measles is the quality of immunization delivery (Foster and others a significant cause of mortality in the first year of life (WHO/EPI 1990; Heiby 1990)? Measles 171 * Access. Do local health jurisdictions have maps of their Table 8-2. Surveillance Methods Used in Measles service areas and a sense of responsibility for the people Control living within those areas? What percentage of the popula- Surveillance method UIse of data tion has access to immunization services? How can access be increased? Routine reporting' Monitor trends in incidence over time Identify foci of measles for case * Coverage. What percentage of the at-risk population investigation has been immunized with measles vaccine by twelve . , 7 . 8 * . . ~~~~~~~~Sentinel surveillance Monitor trends in incidence over time months of age? Two methods are used to assess immuniza- Monitor demographic and epide- tion coverage: (a) dividing the numtber of immunizations iologic characteristics of cases reportedly administered under one year of age by the Identify high-risk populations number of surviving infants and (b) completing coverage surveys as recommended by WHO (1988). If local or na- investigationsh bmortality, and disability tional targets have not been achieved, what can be done Estimate vaccine efficacy to achieve these targets? Identify populations at high risk * Morbidirv and mortality reduction. Is there a routine or Identify risk factors for vaccine failure sentinel reporting system to monitor trends in measles inci- Special studies Assess susceptibility to infection and dence? Are morbidity reduction targets being achieved? If vaccine seroconversion by serologic not, what changes in policy or strategy are needed? surveys Test alter-ative vaccine strains and * High-risk strategy. Are epidemiologic data available to delivery schedules identify populations at increased risk of dying when in- Evaluate impact of measles immuni- fected with measles (high case-fatality rates)? If so, how zation on survival can strategies be altered to ensure high coverage in those a. Useful only if reporting is constant over time. populations? h. Includes cohort, case-control, and cross-sectional studies. * Community partcipation. Does the local community par- Source: Authors. ticipate in immunization through identification of individ- uals in need of vaccination, publicity of time and place for Effective surveillance requires timely and effective use of vaccine delivery, and in disease surveillance' data at each level of the health system: local, district, national. At the local level, every measles case should be considered for Su-veilLance its epidemiologic and management relevance. Each case should be assessed as preventable or nonpreventable. Identification of preventability is not a method of faultfinding but a source Achievement of the 1995vmoity an moalit reductin of information for problem identification and problem solu- targets will require improvements i measles surveillance i tion. Early identification of cases is the necessary first step in effective outbreak control. Case data, together with locally * Documentation of morbidity and mortalitv associated available coverage data, can also be used to assess vaccine with measles infection efficacy (Orenstein and others 1985). At district and national wslevels, subunit coverage and incidence data can be used to * Identification of population groups at high risk of moIdentificationyof popula tion groups at high risk ot assess individual area performance and to identify high-risk mortalityf areas for supervisory attention. Epidemiologic analyses of na- * The monitoring of trends in measles incidence tional data provide important programmatic data for assessing * Assessmentoftheeffectivenessofprograminterventions program status, establishing targets, monitoring performance. * Identification and targeting of program failures in order and providing information for feedback. to reformulate, where necessary, policies and strategies and to define research priorities. Measles Strategies for the 1 990s Traditionally, disease surveillance is understood as the routine Countries, regions and WHO are currently in dialogues on the reporting of morbidity and mortality from health facilities, appropriate goal for measles to be achieved over the next through intermediary levels, for collation, analysis, and report- decade. ing at the national level. This traditional approach to surveil- lance is flawed on two accounts: (a) measles surveillance data Control, Elimination, Eradication are most useful at the level of collection and (b) achievement of the surveillance objectives listed above requires the use of At the global level, there is considerable debate as to the multiple surveillance methodologies. In table 8-2 we summa- appropriate long-term measles objective: control, elimination, rize types of surveillance methods useful in effectively manag- or eradication. Understanding of the terminology is essential ing measles immunization programs. to this dialogue. Control means the reduction of measles mor- 172 Stanley 0. Foster, Deborah A. McFarland, and A. MeredithJohn bidity and morality to a level that it is no longer a public health years (the risk of importation was small); conversely, measles problem. Elimination implies the interription of measles trans- eradication, because of the ease of importations, would require mission in a geographically defined area, island, nation, or work on a global scale. continent. E-radication is the interruption of person-to-person The mathematical models of measles transmission for indus- transmission, the elimination of the virus reservoir, and the trial and developing countries both predict that if more than termination of prevention procedures. The current WHO/1NJICEF 98 percent of young, susceptible children are protected against goals of 90 percent reduction in morbidity and 95 percent reduc- measles, the disease can be eradicated in large populations. It tion in mortality by 1995 are consistent with measles control. is important to note, however, that this prediction is based on Measles elimination has been targeted for the United States, the assumptions that the population is homogeneous (there are Europe, and the Caribbean (PsAHO 1990). Although measles no isolated subpopulations and everyone is equally likely to elimination has been achieved in certain populations (the mix with infected individuals and be vaccinated) and that Gambia; Sao Paulo, Brazil; and Cuba), the goal of sustained vaccination failures are rare. These assumptions clearly do not measles elimination has been more difficult. In the United hold in large urban populations. When these assumptions are States, the measles elimination target of October 1, 1982, was relaxed, allowing, for example, for variations in susceptibility not achieved. Although the program was successful in achiev- to infection, in-home exposure, and access to vaccination, the ing a remarkable 98 percent reduction in measles incidence, critical level of protection necessary for eradication rises to persistent transmission has continued primarily in two popu- nearly 100 percent. lation groups: urban infants, and older high school and college- These differences and the data presented in this chapter on age stLidents. In urban- areas, the problem has been one of disease epidemiology and vaccine efficacy would seem to indi- progratn implementation, the failure to achieve high coverage cate that measles eradication is not achievable with the cur- in infants from poor families. Intensified efforts are being rent vaccine and the current or projected levels of coverage. carried out to increase timely immunization of infants in urban This is not to say that measles eradication is not a desirable communities. Infection in the older age group reflects the long-term objective. New, improved vaccines and possibly accumulation of susceptibles caused by nonimmunization and alternative strategies are needed. However, in a world of lim- vaccine failures, many of which relate to immunizations given ited health resources, careful attention must be given to the prior to the currently recommended age of fifteen months. The opportunity cost of allocating funds to measles eradication addition of a second dose of measles vaccine will, in time, instead of to other priority health and development needs. eliminate most of the susceptibles among the older age group. The high cost of achieving and sustaining measles control Mathematical Models has prompted some individuals to propose the global eradica- tion of measles (Hopkins and others 1982; Foege 1984). Much In the past sixty years, many mathematical models of measles of the advocacy for eradication arises out of the successful transmission and of measles control by vaccination have been smallpox eradication program. Hopkins and colleagues have developed (for example, Kermack and McKendrick 1927; identified similarities and dissimilarities of measles to small- Dietz 1976; Hethcote 1976; Fine and Clarkson 1982a, 1982b; pox: "Both viruses cause infections which are accompanied by Schenzle 1989; Anderson and May 1985). These models are typical rashes and which confer life-long immunity; and both based upon the demography and epidemiology of industrial viruses have no animal reservoir and do not produce a chronic countries and thus describe fairly well the measles patterns carrier state in man" (Hopkins and others 1982, p. 1396). in these settings. They do not, however, always accurately Dissimilarities, they report, include "the highly contagious describe measles transmission pattems typical of developing nature of measles" (70 percent attack rate for measles com- countries. pared with 33 percent for smallpox), the average age of infec- tion (twelve months to eighteen months for measles as opposed TRANSMISSION. The simplified characteristics of measles to four to five years for smallpox), the age at which a vaccine transmission at the population level are the same in both is effective (six months to nine months for measles as opposed industrial and developing countries: infants born to mothers to at birth for smallpox), and the difficulty in diagnosing mild who are immune to measles are protected from infection for measles as opposed to the ease of diagnosis of both the acute several months by transplacental matemal antibodies; the and the recovered case of smallpox (diagnostic pox and scars infectivity of measles is high; an individual is both infected and for smallpox) (Hopkins and others 1982, p. 1396). Other infectious at roughly the same time; case-fatality rates are differences include vaccine effectiveness (99 percent forsmall- higher in infancy than in childhood; and recovery from mea- pox as against 80 to 90 percent for measles), the stability of sles results in subsequent long-lasting immunity. Thus, in the vaccine (one year at ambient temperatLire for smallpox vaccine simplest model, there are four epidemiologic classes of people: as opposed to the cold chain required for measles vaccine), and (a) those protected by maternally derived antibodies, (b) those the effectiveness of outbreak control (achievable within one susceptible to infection, (c) those who are infected and infec- incubation period for smallpox but difficult beyond the first tious, and (d) those who have recovered from measles and are generation for measles). It should also he noted that smallpox therefore immune. Figure 8-6 traces the progression of individ- eradicationi required activities in thirty countries for twelve uals among the epidemiologic classes. Measles 173 Figure 8-6. Stages in the Measles Infection Cycle decreases the probability that a susceptible individual will be of an Individual from Birth through Sequence exposed to measles. of Epidemiologic Classes to Death TRANSMISSION MODELS FOR INDUSTRIAL COUNTRIES. In the models of measles transmission and control for industrial coun- < Blrth < tries, the principal epidemiologic assumption is that the rate of measles spread is independent of the spatial density of the host population; the principal demographic assumption is that Protected by the host population is not growing. It is generally assumed that maternai vaccination takes place at a precisely targeted age and that all I vaccinations are effective. The measles transmission model for Loss of protection industrial countries yields several predictions: * The number of susceptibles in the population remains Susceptible + fl r the same in the presence and absence of immunization. * The median age at infection in the population increases Non-casrelated after vaccination. ea~t~h v * In the presence of even modest levels of vaccination, the \ Infection f(t) A L ' period between epidemic peaks (interepidemic period) will A I lengthen. * At any given level of vaccination coverage, the percent- Infected/ age drop in the incidence of measles should be greater than aerat infectious the level of vaccination coverage. * The proportion of each cohort that must be immunized Recovery to interrupt measles transmission is less than 1.0. The predictions of this model correspond well to observed Immune _ pre- and postvaccination measles transmission pattems in the United States and in many European countries. TRANSMISSION MODELS FOR DEVELOPING COUNTRIES. In con- Source: Authors. trast, measles transmission models for developing countries Initially, the population consists of infants protected by explicitly account for the demographic and epidemiologic matemal antibodies and of susceptible individuals who mix structure typical of the populations of such countries (John randomly. A measles outbreak begins when infected and in- 1990a, 1990b; John and Tuljapurkar 1990; Tuljapurkar and fectious individuals have contact with a sufficient density of John 1990; Nokes and others 1990), particularly the high rates susceptibles. Each time a susceptible individual is encountered, of population growth. In these models the demographic struc- the latter may be infected with a probability proportional to ture of the host population is determined by the population's the intensity of exposure. At the earliest stage of the outbreak fertility and mortality. The distribution of individuals among most encounters by infectious individuals are with susceptible the four epidemiologic classes at each age is governed by individuals; therefore, measles spreads quickly. When the ill- epidemiologic parameters: age pattern of loss of matemal an- ness runs its course in the infected individual, he or she is then tibodies, age-spec,fu c immunizatnon coverage, age-specific cs- immune. As measles transmission progresses through the pOp- fectlion rate, duration of g ifectrvutu, and age-specific case- ulation, the number of susceptibles decreases whereas the fatalityrate Thedemographicstructureofthepopulationand number of immune individuals increases; therefore, it becomes the epidemiologic behavior of the infectious disease are linked less likely that an infected individual will encounter suscepti- by the infection rate, which depends on the population's ble individuals and create new infections. If the number of demographic and epidemiologic structure and on the spatial immune individuals is high enough, measles will die out, even densitv of the population. More complex developing-country though there are still some susceptible individuals in the models allow for spatial heterogeneity in infection rates, population: this is the phenomenon of "herd immunity." If, urban-rural migration, and seasonality in birth deh d however, susceptibles enter the population (by birth or migra- migration. tion) at a sufficiently high rate, measles may not die out but The predictions of even the simplest model for developing may instead become endemic. ~~~countries are strikingly different from those of the model for may instead become endemic.inutalcnres Immunization programs thus exert their effect at both the industrial countries individual and the population level: vaccination changes the * The equilibrium proportion of infected individuals in immune status of the individual and, within the population, it the population (the equilibrium measles prevalence) in- 174 Stanlev 0. Foster, Deborah A. McFarland, and A. Meredith John Figure 8-7. Interepidemic Interval as Function in Zaire and Cameroon. Between 1980 and 1985 an intensive of Population Growth Rate at Different Levels measles vaccination program in Kinshasa, Zaire, resulted in the of Immunization Coverage vaccination of almost 60 percent of the children who were twelve months through twenty-three months old, yet "two Interepidemic period (years) results expected from [measles transmission models]-a reduc- tion in measles incidence greater than the level of vaccina- 5.0 tion coverage and a shift in the age distribution of measles 4.5 Immunization coverage to older children-have not occurred in this African urban 40 \------ High population" (Taylor and others 1988, p. 792). In addition, 4.0 ------------ Low the predicted increase in the interval between epidemic 3.5 ' Zero outbreaks of measles was not observed: epidemics continued 3.0 _ -. a to occur biennially. In Yaounde, Cameroon, the results of a .5 .. -- measles vaccination program showed a slight shift upward 2.5- - - - in the mean age of infection but no corresponding length- 2.0 - ening of the interepidemic interval (Guyer and McBean 1981). In both cases, the observed results are consistent with 1.5 the prediction of the simplest transmission model for devel- 1.0 l l l l oping countries. 0.000 0.005 0.010 0.015 0.020 0.025 0.030 0.035 0.040 APPLICATION OF MEASLES MODELS: MEASLES INCIDENCE DYINAM4ICS. Annual population growth rate When designing vaccination programs for developing coun- tries, one is rarely faced with the task of fine-tuning the details of vaccination delivery, such as deciding whether the optimum Note: Interepidemic interval (years) plotted as a function of population growth rate, r, for different levels of immunization coverage (low and high) and for no age for vaccination is eight months or nine months. Rather, immunization (i=0.0). In the absence of immunization, the inter-epidemic interval oeweighs the merits of substantial program mdfctos shortens as r increases (3.0 years at r=0.0 to 1.5 years at r=4). At r=0.0, one modifications: immunization sharply increases the periods between epidemics from 3.0 to decreasing missed opportunities, starting vaccination at six 4.6 years, while at r=4.0, immunization increases the inter-epidemic period by i only 0.4 years. months rather than nine months, instituting annual or semi- Source: Authors, annual vaccination days, or changing to two-dose schedules. Mathematical models of measles transmission and control are creases as the growth rate of the population increases, both useful tools for vaccination program design and evaluation. when there is no vaccination in the population and when there is an ongoing vaccination program. Figure 8-8. Combined Effect of Population Growth * The mean age at infection in the population need not Rate and Immunization Coverage on Interepidemic increase after vaccination, because the remaining post- Level vaccination cases may be concentrated at the extremes in the youngest and the oldest children; however, the age Interepidemic period (years) distribution of cases may change substantially. * The interepidemic period does not necessarily increase 10 after implementation of a vaccination program in a growing 9 - Annual population growth rate / ' population: when the level of vaccination is a small fraction 8 - .o/ of the critical level of vaccination required to stop transmis- 7 - --- - 0.02 sion, changes in the interepidemic period are quite stnall, -- 0.04 6 but when the level of vaccination nears the critical level, ' the interepidemic period shows a substantial increase (fig- 5 ures 8-7 and 8-8). 4 * The percentage drop in the incidence of measles, for any 3 - - - - , ' given level of vaccination, will be smaller in a growing 2 population than in a nongrowing population: for example, 1 vaccination of 50 percent of the children might induce a 0 l l l l drop in measles incidence of 60 percent in an industrial 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 country but of only 52 percent in a iapidly growing popula- tion in a developing country. Immunization level/critical level In contrast to the model for industrial countries, the predictions of the model for developing countries are con- Note: Horizontal axis shows the ratio of achieved level of immunization and the crtical level of immunization required to stop measles transmission. sistent with the observed effects of immunization programs Source: Authors. Measles 175 Figure 8-9. Simulation Model of Measles Incidence during the twenty-five years projected. Thus, ascertaining after Immunization whether or not an immunization program has achieved its goals, in the short run, is in fact quite difficult; the "success" of Annual incidence (per 1,000 population) the program depends very much on the relative timing of 40 intervention and evaluation. begin 35 - immunization DISEASE REDUCTrON. We have used a simple measles model 30 _ for developing countries to estimate measles cases and deaths that would occur under the following five different levels of 25 - vaccinations in two settings-high-density urban and low-den- 20 - sity rural: no vaccination, measles vaccination at nine months, measles vaccination at nine months and decreased missed 15 - opportunities, high-titer Edmonston-Zagreb (Ez) or equivalent 10 _ measles vaccine at six months, high-titer Edmonston-Zagreb or equivalent measles vaccine at six and twelve months. Be- 5 - cause an increase in mortality has recently been observed in 0 2 - l l populations receiving high-titer vaccine (Garenne and others 5 10 15 20 25 1991), development of an altemative safe effective vaccine providing 85 percent or higher seroconversion at six months Time (years) will be needed. To facilitate a comparison among the two populations, a standard population was used with the following Source: Authors. characteristics: a birth rate of 48 per 1,000, an infant mortality rate of 100 per 1,000 live births, a population growth rate of Even though one must recognize the importance of parameter 3.5 percent, and a population under ten years of age of 34 assumptions, the models represent powerful tools for evaluat- percent. The age distribution of cases and age-specific case- ing the potential mortality and morbidity effects of different fatality rates for each of the two scenarios are listed in table vaccination programs, for anticipating the dynamic behavior 8-3. For each scenario, estimates were also made of cases of measles in the population, and for examining the influence occurring prior to vaccination, of vaccine coverage, and of of demographic variation on measles transmission patterns. For vaccine efficacy, as shown in table 8-4. example, for any given specification of the demographic and Using the measles model for developing countries and the epidemiologic parameters, the potential effect of a vaccination assumptions listed above, we summarize in table 8-5, for both program can be estimated and alternative vaccination delivery urban and rural settings, the estimates of the number of cases, strategies compared; the effects of changes in the host popula- the number and percentage of cases prevented, and the number tion due to child survival programs, family planning programs, and percentage of deaths prevented. and urbanization can be studied; and the short-run fluctuations Several important observations can be made from the in measles incidence can be predicted. simulation model: measles mortality is higher in urban areas The nature of the short-run fluctuations in measles inci- than in rural areas because of the younger age of infection, dence foillowing an immunization program is crucial in ascer- an age at which case-fatality rates are higher; at equivalent taining whether the WHO goal of a 90 percent reduction of measles incidence can be achieved by 1995. The introduction of any vaccination program reduces the long-run (equilibrium) Table 8-3. Mathematical Model Assumptions incidence of measles in the population. The short-run effect for Unvaccinated Urban and Rural Scenarios: on measles incidence is, however, dominated by the fluctua- Age-Specific Case-Fatality Rates and Age tions in annual incidence (figure 8-9). In the simulation model Distribution of Cases presented here, immunization results in the desired 90 percent (Age distribution in 1990) reduction in measles incidence within two years (that is, by Age distributon year seven), but by year nine, the annual incidence is half of Age (months) Case-fatality rate Urban Rural the preimmunization level, which would suggest that this <6 - 0 0 immunization program had not achieved the desired goals, 6-8 4 18 3 despite the evidence from two years earlier. Yet in year ten, 9-11 4 19 6 there appears to have been an 80 percent reduction from 12-23 5 40 14 preimmunization levels. In this model, with this set of param- 24-35 3 10 16 eters, the steady estimate of reduction of disease incidence is 36-47 2 8 18 from 38 to 14 cases per 1,000 population, a reduction of 63 48-59 1 5 20 percent. This model, so configured, assumes that fertility, 60-119 .05 23 mortality, and immunization parameters remain unchanged Source: Authors. 176 Stanley 0. Foster, Deborah A. McFarland, and A. Meredith John Table 8-4. Mathematical Model Assumptions for currently achieved levels of coverage and vaccine efficacy, Measles: Subjects Immune from Infection at Time projected reductions in morbidity and mortality significantly of Vaccination, Coverage, and Efficacy of Vaccine below the 90 percent morbidity and 95 percent mortality Urban Rural reduction targets established by WHO and affirmed at the 1990 Cover- Cover- World Summit for Children. According to the model, a Scenario Immune age Efficacy Immune age Efficacy two-dose schedule with a vaccine effective at six months of age has the greatest potential for moving operational pro- Measles vaccmne at9 months 30 60 85 10 60 85 grams toward the global targets. Further improvements in Eliminate missed vaccine coverage and effectiveness will be required to en- opportunities 30 80 85 10 80 85 sure the elimination of measles as a significant cause of EY at 6 months 10 80 85 2 80 85 childhood morbidity and mortality. Ez' at 6 months 80 85 80 85 and 12 months 10 60 95 2 60 95 COST-EFFECTIVENESS OF ALTERNATIVE STRATEGIES. Given the a. Edmonston-Zagreb or equivalent measles vaccine providing 85 percent above results, we can now consider the costs of the altemative seroconversion and protection when given at six months of age. strategies. Costs are predicated on an average cost of $15 per Source: Authors. fully immunized child with 40 percent of the cost, or $6, allocated to measles. The cost profile (the contribution of each levels of coverage, rural strategies are more effective in reduc- cost component to total cost) is based on the delivery modality ing morbidity and mortality; use of an effective vaccine at six for immunizations in fixed sites (Brenzel 1990). For each months significantly increases the effectiveness of urban im- altemative strategy, assumptions were made which would munization; and two-dose schedules further increase program change original cost estimates, that is, the incremental costs effectiveness. For simplicity, these calculations do not take attributable to each strategy. These assumptions are enumer- into account herd immunity; they do, however, provide an ated below. Costs are assumed to be the same in both urban estimate of the long-term effect of alternative strategies for the and rural settings with the exceptions noted. delivery of measles vaccine. Maximum reductions in morbidity * Measles immunization at nine months. and mortality are obtained with the two-dose vaccine schedule * Measles immunization at nine months but a 50 percent in which a vaccine is used that is effective when administered at six months of age. Coverage levels used in these models increase In vaccine use so that vaccines represent 10 percent represent those currently being achieved by well-managed oftotalcostsanda10percentincreaseinsupervisioncosts. immunization programs in developing countries. Few coun- * Edmonston-Zagreb or equivalent vaccine given at six tries have been able to achieve and maintain the 95 percent months are $6 (the same as measles immunization at nine levels achieved in industrial countries. months). Cost of vaccine is same as currently used measles Simulations are only as accurate as the assumptions and vaccine. Delivery pattern and sites remain the same. Al- the model used. The assumptions used above reflect data though one might expect declining average costs because of collected from the city of Lagos, Nigeria, Kinshasa, Zaire, increasingvolume, the effect isprobably quite small and thus and the rural areas of Matlab, Bangladesh. It is our expecta- negligible to average cost. tion that these data will contribute to the dialogue on * Edmonston-Zagreb or equivalent vaccine given at six alternative measles strategies. The model, reflecting the months and twelve months: cost for second dose is the same Table 8-5. Simulation Estimates of Measles Cases and Deaths in High-Density Urban and Low-Density Rural Scenarios: Developing Country Model Urban Rural Cases Deaths Cases Deaths Cases prevented Deaths prevented Cases prevented Deaths prevented Scenario Cases prevented (percent) Deaths prevented (percent) Cases prevented (percent) Deaths prevented (percent) No vaccination 36,400 n.a. n.a. 1,452 n.a. n.a. 36,400 n.a. n.a. 806 n.a. n.a. Measles vaccine at 9 months 25,744 10,656 29 1,025 425 29 20,193 16,207 45 456 350 45 Vaccine at 9 months and eliminates missed opportunities 22,192 14,208 39 885 567 39 14,791 21,609 59 339 467 58 Eza at 6 months 14,123 22,277 61 563 889 61 12,165 24,235 67 269 537 67 Ez aat 6 and 12 months 9,051 27,349 75 361 1,091 75 5,844 30,556 84 140 660 83 n.a. Not applicable. a. Edmonston-Zagreb or other vaccine that achieves 85 percent seroconversion and protection when administered at six months of age. Source: Authors' calculations. Measles 177 Table 8-6. Efficacy and Cost of Alternative Strategies to Increase Measles Coverage in Urban Areas Incre- Incre- Total Total Total mental mental Incre- Cases Deaths Unit Cover- Doses Total incre- costper costper cost per cost per Total mental pre- pre- cost age adminis- annual mental case death case death cost per cost per Strateg-y vented vented $ (percent) tered cost cost prevented prevented prevented prevented DALY DALY Measles vaccine at 9 months (baseline) 10,656 425 6.00 60 30,600 183,600 n.a. 17.23 432.00 n.a. n.a. 14.90 n.a. Vaccine at 9 months and missed opportunities 14,208 567 6.42 80 40,800 261,936 78,336 18.44 461.97 22.05 551.66 15.93 19.02 Ez' at 6 months 22,277 889 6.00 80 40,800 244,800 61,200 10.99 275.37 5.27 131.90 9.50 4.55 Eza at 6 months 6.00 80 40,800 and 12 months 27,349 1,091 6.00 60 27,000 406,800 223,200 14.87 372.87 13.37 335.14 12.86 11.56 n.a. Not applicable; increment is determined in relation to this baseline. a. Edmonston-Zagreb or equivalent vaccine providing 80 percent seroconversion when given at six months of age. Source: Authors' calculations. as first dose at six months. Assume that second dose is The cost per disability-adjusted life-year was calculated administered at routine vaccination session or child health under the assumption that a death averted "buys" about sixty visit and therefore does not require additional personnel or years of life or, if one discounts future life-years gained at 3 outreach. A decrease in volume may predictably increase percent, the annuity stream reveals that a prevented death of the average cost of the second dose only if the average cost a child from measles buys about 29 disability-adjusted life- curve is quite steep. years. The calculation does not account for DALYS lost to disability caused by measles, because it is estimated that more Using the measles model for developing countries pre- than 95 percent of years of life lost from measles are due to sented earlier, the annual number of cases and deaths in premature mortality and not to disability (Duflo and others children under ten that would be prevented by each alter- 1986). More refined estimates of DALYs would need to take native strategy in each scenario was tabulated for use in the into account disability caused by measles complications and cost-effectiveness calculations (tables 8-6 and 8-7). In these the concomitant cost in healthy life-years lost. Results of calculations, cases and deaths prevented represent the an- the cost-effectiveness studies are summarized in tables 8-6 nual number of cases and deaths prevented for the entire and 8-7. cohort of children under ten in any given year. Total annual The tables give relative estimates of the cost-effectiveness costs, however, only reflect costs incurred in a single year to of alternative strategies to increase measles coverage, notwith- immunize the currently eligible children (those under standing all the caveats and assumptions built into the analysis. twelve months of age). Costs are expressed in 1990 dollars. For urban populations, the most cost-effective strategy appears In order to compare these results with previous studies, the to be administering Edmonston-Zagreb or equivalent vaccine costs must be converted to the relevant year for which the to children at six months of age. This strategy is also the most study data were reported. cost-effective for rural populations, although use of the current Table 8-7. Efficacy and Cost of Alternative Strategies to Increase Measles Coverage in Rural Areas Incre- Incre- Total Total Total mental mental Incre- Cases Deaths Unit Cover- Doses Total incre- cost per cost per cost per cost per Total mental pre- pre- cost age adminis- annual mental case death case death cost per cost per Strategy vented vented $ (percent) tered cost cost prevented prevented prevented prevented DALY DALY Measles vaccine at 9 months (baseline) 16,207 350 6.00 60 30.600 183,600 n.a. 77.33 524.57 n.a. n.a. 18.09 n.a. Vaccine at 9 months and missed opportunities 21,609 467 6.42 80 40,800 261,936 78,336 12.12 560.89 14.50 669.54 19.34 23.09 Eza at 6 months 24,235 537 6.00 80 40,800 244,800 61,200 10.10 455.87 7.62 327.27 15.72 11.29 Eza at 6 months 6.00 80 40,800 and 12 months 30,556 666 6.00 60 27,000 406,800 223,200 13.31 10.81 15.56 706.33 21.06 24.36 n.a. Not applicable; increment is determined in relation to this baseline. a. Edmonston-Zagreb or equivalent vaccine providing 80 percent seroconversion when given at six months of age. Source: Authors' calculations. 178 Stanlev 0. Foster, Deborah A. McFarland, and A. Meredith John measles vaccine at nine months and use of every missed oppor- Figure 8-10. Age of Measles Vaccination tunity along with the current measles vaccine appear to be of 48-59-Month-Old Children in Six Countries almost as effective. At least for rural populations, there is no significant difference in the cost-effectiveness of the first three Age (months) srrategies, given the limits of the analysis. 60 F c O~~~~~~~~~~~~ Achieving the 1995 Measles Targets _ 0 0 48 8 0 In 1989, WHO established global EFI targets for the decade of o 8 the 1990s: that coverage levels will surpass 80 percent in all _ 0 countries or areas by the end of 1990 and that levels of 90 36 - _ percent, in the context of comprehensive maternal and $ child health services, can be achieved by the year 2000. At o the September 1990 World Summit for Children, the WHO 24 - 1995 targets for morbidity and mortality reduction were affirmed, 90 percent and 95 percent, respectively. Although 4 global levels of immunization coverage have increased 12 - dramatically during the last decade, representing a major achievement of national governments and their collaborat- ing partners, there is still a significant gap between the 0 _ current levels of coverage and disease reduction and the Colombia Guatemala Sri Lanka Senegal Thailand Zimbabwe 1990 targets, as shown in table 8-8. 1986 1987 1987 1986 1987 1988-89 Achievement of the 1995 and 2000 targets will require increases in both coverage and vaccine efficacy. Eleven strat- Note: For each population, the median age at immunization is indicated by line inside box. The median age of immunizations administered at ages egies, some already a part of the EPI program, hold the potential greater than the median age (approximately the 75th percentile) are graphed as the upper bar of the box. The line extended from the top of the box ends at to increase levels of coverage, increase vaccine efficacy, de- upper boundary: the observation closest to but less than the sum of the 75th crease measles incidence, and decrease measles mortality: (a) percentile plus 1.5 times the difference between the 25th and 75th percentile. Any observations greater than the boundary were deemed outliers and are vaccination in the first year of life, (b) reduction of missed plotted individually. The lower boundary was constructed in an analogous manner. opportunities, (c) increase in community partnership, (d) reg- Source: Authors. istration and follow-up of newborns, (e) use of accelerated immunization strategies, (f) vaccination of hiigh-risk groups, VACCINATION IN THE FIRST YEAR OF LIFE. Measles vaccine is (g) adoption of two-dose measles vaccine schedules, (h) pro- effective when administered to a susceptible individual prior vision of vitamin A supplementation in vitamin A-deficient to or at the time of exposure to measles. The World Health areas, (i) treatment of severe cases of measles with vitamin A, Organization and UNICEF have emphasized the importance of (j) effective treatment of measles complications, and (k) ex- vaccination in the first year of life in developing countries. pansion of the infrastructure. The first six of these strategies, Vaccination of older children is less effective in developing in part developed from experience in the developing world, are countries, because many children (the number increasing with important components of the current United States initiative age) may have already become immune through infection with to achieve measles control (United States, National Vaccine wild virus. Using data from demographic and health surveys Advisory Committee 1991). (Boerma and others 1990), we provide in figure 8-10 a boxplot of the age distributions of card-documented immunizations Table 8-8. Measles Coverage and Estimated from six countries. Except for Zimbabwe, the boxplots show a Disease Rediction, by WHo Region, 1989 pattern of delayed measles vaccination. (percent) Increased attention to vaccination in the first year of life, at nine months for the Schwarz strain and six months for alter- Measles Estimated reduction native vaccines that provide high rates of seroconversion and Regioni coverage in morbiditN' .. -gncegi o protection, will increase the probability that a dose of vaccine Africa 47 40 will be administered to a susceptible infant and thus, depend- Americas 73 62 ing on age-specific rates of vaccine efficacy, increase program Mediterranean 70 60 effectiveness in achieving disease reduction. Europe 85 72 Southeast Asia 58 49 Western Pacific 90 77 REDUCTION OF MISSED OPPORTUNITIES. The term missed op- portunities" is defined as contacts of a target-age individual Note: 1995 target is 90 percent for both measles coverage and morbidity (infant, child, or reproductive-age woman) in need of one or a. Conerage times Xaccine efficacv (85 percent). more vaccines with a health facility capable of providing that Source: WHO intemal data. vaccine and a failure of that contact to provide the needed Measles 179 vaccine(s). There are two types of missed opportunities for too ill to require hospitalization should be deferred for immunization: missed vaccination opportunities and missed decision of hospital authorities" (wHo/EPI 1984, p. 15). health facility opportunities. Although recent data suggest decreased rates of sero- Missed vaccination opportunities occur when an individual conversion in children with respiratory infection (Krober attending a vaccination session fails to receive a needed vac- 1991), an overall assessment of risks and benefits mandates cine or receives one inappropriately timed by age or inter- vaccination of both sick and well children. Because of the vaccination interval. A review of fifteen published articles has risk of nosocomial spread of measles, all children from six documented missed vaccination rates ranging from 8 percent months through five years of age who are admitted to in Mozambique to 76 percent in Indonesia; the median for the a hospital should receive measles vaccine on admission fifteen countries was 49 percent (Hirschom 1990; Grabowsky if there is no documentation of age-appropriate measles 1991). immunization. Missed opportunities are also being identified in industrial * Incorrect screening by the health worker. (Screening countries. Investigation of an urban outbreak of measles in the errors relate to a lack of understanding of national vaccina- United States identified missed opportunities for measles im- tion schedule, difficulties in calculating the interval be- munization in fifteen of twenty-six measles cases in an urban tween recorded date of birth and current date, failure to outbreak among preschool children (Hutchins and others check all antigens, clerical error because of fatigue, or lack 1989). Missed opportunities were also documented as a signif- of motivation.) icant factor in the 1990 measles epidemic in the United States (United States, National Vaccine Advisory Committee 1991). t rowded Four methods have been used to assess the rate of missed deClnir opportunities: * Absence of staff, vaccine, or transport resulting in the * Record reviews. Facility-held records, vaccination regis- . o ters, or individual patient immunization cards are reviewed cancellation of a scheduled immunization session. to assess whether all indicated antigens were administered * Mothers too busy to wait, not informed that they should on recorded dates of vaccination contact. wait, or dissatisfied. * Exit interviews. Child caretakers and reproductive-age * Healthworkers'fearofwastingmeaslesvaccine,resulting women leaving a clinic during a time at which immuni- in refusal to open a multidose vial for one child (WHo and zations are being administered are interviewed by health UNICEF recommend opening a vial of vaccine even for one staff. Immunization records are examined for missed or child). inappropriately timed immunizations (early for age or too * Acceptance by health worker of an oral history of mea- short an interval between doses). sles or measles vaccination as a reason for nonimmunization. * Clinic observation. Supervisory staff observe immuniza- (Serologic studies have documented the unreliability of tion sessions to identify errors in screening, referral, or histories of measles and measles immunization; all eligible immunization. children without written proof of measles vaccination * Coverage surveys. As part of surveys to assess imnmuniza- should be immunized with measles vaccine.) The failure of tion coverage, data from individual client-held record cards child caretakers to bring the immunization cards to clinic are reviewed for missed opportunities for immunization on also contributes to this problem. days of recorded attendance at an immunization session and * Unwillingness of health workers to administer more than for inappropriately timed immunizations (Cutts, Glik, Gor- one antigen at a time. (Studies have documented the safety don, and others 1990). Analyses of these data are facilitated and efficacy of simultaneous separate multiple antigen vac- by use of COSAs, a software package which analyzes coverage cine administration IFoege and Foster 1974]). survey data for coverage, age, intervals between immuni- * Nonimmunizationofiindividual identified for immuniza- zations, and missed opportunities (Boyd 1991). tion. This occurs when immunization cards are returned to Ten major causes for missed opportunities for immunization parents prior to the completion of all immunizations. Such have been identified (Hirschorn 1990). situations may arise when children have been referred for multiple immunizations, for example, DPT, OPV, and measles. * Coexistent illness as determined by health worker or Delaying the return of the vaccination card until all anti- child caretaker. In contrast, note that WHO policy calls for gens required for that vaccination session are administered immunization at every opportunity. "It is particularly im- can eliminate this problem. portant to immunize children suffering from malnutrition. Low-gradefever, mild respiratory infectionsordiarrhea, and When persons eligible for immunizations visit health other mild illnesses should not be considered contraindica- facilities that are capable of delivering vaccine and vaccines tions to immunization. The decision to withhold immuniza- are not given, a missed health facility contact has occurred. tion should be taken only after serious consideration of the Such missed contacts happen in clinics in which adminis- individual child and community. Immunization of children tration of vaccines is limited to certain days (infant welfare) 180 Stanle-y 0. Foster, Deborah A. McFarland, and A. Meredith John or groups (well children). And all too commonly they occur around the world, both in the provision of funds (over $200 because curative-oriented health workers fail to assess im- million) and in the active involvement of local Rotarians in munization status. Although these missed opportunities may social mobilization and direct assistance in vaccine delivery be the most common ones, they are less well documented. activities. Prime emphasis is being given to the training of Identification of such missed opportunities is facilitated by a community volunteers to identify and refer eligible children unified clinic- or patient-held record system in which im- for immunization. In areas where such programs are opera- munization and all health contacts are entered on the same tional (for example, ljeru-Ekiti, Nigeria), coverage rates are record. In the Central African Republic, a comparison of dates over 90 percent and drop-out rates are near zero. of all health facility contacts with opportunities for measles immunization showed that use of all opportunities would have REGISTRATION AND FOLLOW-UP OF NEWBORNS. In rural areas achieved a measles vaccine coverage of 76 percent, rather than where population movement is limited, the enumeration of the actual 54 percent (Roungou 1991). In Guinea, a commu- births and the monitoring of immunization status through the nity survey estimated that 30 percent of the opportunities for first year of life has proved effective in increasing coverage. In measles immunization had been missed (Cutts, Glik, Gordon Oman, health facility records of immunization are maintained and others 1990). by month of birth (health facility usage is over 95 percent); a Integrated service delivery in which all health facility con- monthly review of immunization records of one-year-old chil- tacts are used to screen and immunize all eligible persons can dren provides a mechanism for assessing coverage and identi- reduce this problem. Although the transition from specialized fying defaulters for follow-up. Coverage in this population is immunization clinics to routine immunization is initially dif- over 98 percent (Foster 1989). In other places, maintenance ficult, the shift to a "comprehensive" approach to vaccine of village registers serves the same purpose. Door-to-door visits delivery has been effective in reducing the missed health have also been used to register the at-risk population, identify facility opportunity. Pioneered by Shanti Ghosh in Delhi, susceptibles, and refer eligibles for vaccination. India, the practice of screening and immunizing all emergency department and outpatient cases prior to their contact with a USE OF ACCELERATED IMMUNIZATION STRATEGIES. During the physician or nurse has been effective in reducing missed op- last decade, WHO and UNICEF have promoted accelerated im- portunities, increasing coverage, and reducing disease morbid- munization activities as a mechanism to increase coverage ity in Zimbabwe and Mozambique (Ekunwe 1984; WHO/EPI (WHO/UNICEF 1985). Historically, immunization days date back Zimbabwe 1989; Hirschom 1990). Yach and others (1991) to the 1950s, when Sunday vaccination days were instituted estimated 240,000 missed opportunities for measles vaccina- to increase polio coverage in the United States. Biennial opv tion per year at two tertiary-level referral hospitals in South polio campaigns have been used widely in Latin America, Africa. In an investigation of an inner-city measles outbreak especially in Brazil, and have been credited with the near in the United States, 38 percent of cases had received DPT or elimination of that disease from the Western hemisphere (de diphtheria-tetanus vaccine at an age when they were eligible Quadros, Andrus, and Olive 1991). to receive measles vaccine (Hutchins and others 1989). In the Largely through the personal advocacy of the executive United States, extension of immunization to contacts with director of UNICEF, national immunization days have been public assistance could significantly increase coverage (United established to increase vaccination coverage. Countries using States, National Vaccine Advisory Committee 1991). this strategy have included Colombia, Turkey, Senegal, Nige- ria, and C6te d'lvoire. These accelerated strategies have been INCREASE IN COMMUNITY PARTNERSHIP. Community partner- promoted to achieve several important objectives: ship in immunization is important to the achievement and * To elevate the health sector in general, and immuniza- maintenance of high levels of vaccine coverage. This is well non programs in particular, to the national political agenda. demonstrated in Indonesia, where the PKK, a national organi- tica progrs m provided to the plannida. zation of women, has become a major partner in childhood Political leaders have provided leadership in the planning, immunization. Vaccination coverage provided by outreach promotion, and implementation of immunization days. vaccinators that had been in the range of 15 to 30 percent *hTo change the public perception of immunization from increased to 70 to 90 percent in villages where the PKK was that of an intermittently available service to that of a basic active. The PKK organizes the clinics and participates in clinic human right. activities, including the preparation of advance publicity, the * To increase business, voluntary organization, and public weighing of children, the recording of weights, nutrition edu- support for immunization programs. cation, and the distribution of contraceptives. In Liberia and * To increase access to immunization services. Mozambique, the participation of local chiefs, traditional birth * To increase immunization coverage and reduce morbid- attendants, and village health committees has been effective ity and mortality. in increasing coverage (Cutts and others 1988; Bender and Macauley 1990). Immunization days in many countries, for example, Colombia, Polio eradication is providing many new strategies to in- Turkey, and C6te d'lvoire, have been spectacularly successful crease partnerships. This is best exemplified by the Rotarians in achieving their political and coverage targets. Immuniza- Measles 181 tion coverage rates have increased to levels in excess of 90 * Frequency and repeatability. As needed, accelerated strat- percent; rates of disease incidence have been dramatically egies should be conducted annually, as in the Indian exam- reduced. ple above, or twice a year, as in the polio campaigns in Latin Fromtheperspectiveofmaintaininghighlevelsofcoverage, America. For measles, timing the activity to the pre- however, the value of these accelerated strategies has been epidemic season maximizes effect and cost-effectiveness. questioned on four accounts: * Decentralization. As sustainability and effectiveness are very dependent on local participation, responsibility for * Opportunity cost-the diversion of limited health re- planning, vaccine delivery, supervision, and evaluation sources from essential preventive and curative health ser- should be decentralized to the level of implementation, for vices to immunization activities, example, district, sector, and so on. * Quality and safety-the inability to provide the quantity * Targeting. Target age groups for immunization, selection and quality of supervision required to ensure compliance of antigens, and timing of accelerated activities need to be with basic technical guidelines (Bryce, Cutts, and Saba based on relevant local data about the availability of the 1990). population, physical access to that population, and disease * Cost-effectiveness-the high cost required for a major epidemiology. campaign (vaccine, supplies, cold-chain equipment, person- * Safety. Because immunizations (perhaps the most cost- nel, transport, and publicity) and the relative inefficiency of effective of all health interventions) are not totally safe-for campaigns in providing vaccines to those at greatest risk, example, in the transmission of pathogens through use of children in the first year of life. nonsterile procedures-systems to ensure quality must be * Sustainability-the value of campaigns in promoting developed and sustained. Systems of training and supervi- and maintaining high rates of vaccine coverage and disease sion need to ensure the quality of vaccine delivery as part of reduction over time. accelerated strategies. This includes maintenance of the cold chain and sterilization, appropriate age and intervals The expanding experience with accelerated strategies Is for immunization, and instructions to the mother about the prompting a shift of policy dialogue from the question of need for return visits to complete immunization. their appropriateness to the question of where, when, and under what conditions accelerated strategies are useful. The poliomyelitis experience in the Americas, especially Brazil, VACCINATION OF HIGH-RISK GROUPS. Four groups of children has shown that immunization days are effective in increas- are particularly at risk from measles and should be vaccinated ing coverage and decreasing disease incidence and can be or, in certain cases, revaccinated. Among refugees, measles has sustained over time (de Quadros, Andrus, and Olive 1991). been identified as the main cause of mortality in new refugee For measles, experience in India demonstrated that annual populations (Toole and Waldman 1988). Ethiopian refugees single-day measles campaigns in a village without access to in Sudan show a measles cFR of 33 percent (Shears and others routine vaccine services was effective in achieving and 1987). Measles immunization has been identified as a "high maintaining measles control (John, Ray, and Steinhoff 1984). priority in emergency relief programs, second only in import- In Liberia, a country in which only 40 percent of the ance to the provision of adequate food rations" (Toole and population had access to health facilities, annual immuniza- others 1989, p. 381). tion weeks for five consecutive years, epidemiologically Hospitalized children, especially those who are severely timed to precede the measles season, have succeeded in malnourished, are, if infected, at high risk of measles-associ- increasing immunization coverage from 15 to 60 percent ated mortality. Mortality in malnourished children infected (CCCD 1990). During 1989, approximately 40 percent of with measles in a hospital setting is frequently above 50 per- annual immunizations were administered during this vacci- cent. Among sixty nosocomial infections requiring admission nation week. Of special importance to the success of these to a South African hospital, measles and its complications campaigns was the local partnership in the planning, fund- accounted for twenty-eight (47 percent) of readmissions and ing of local costs, and implementation of the vaccination seven deaths (Cotton and others 1989). All pediatric admis- weeks. This system, a viable model for many parts of Africa, sions without written documentation of measles immunization has unfortunately been destroyed by civil war. at an appropriate age should be given measles vaccine on Five conditions are suggested as criteria for the appropriate admission. Children vaccinated prior to twelve months of age use of accelerated strategies in achieving local and national EPI should be reimmunized. targets: Nosocomial transmission of measles is common in the de- veloping world. In a study in Cote d'Ivoire, 69 percent of * Low access ( 50 percent access of target population to a measles patients seen at an urban health facility had attended facility regularly providing vaccines). In areas of low access a health facility eight to twenty-one days prior to onset of to health facilities and where the potential for outreach is measles (Foulon and others 1983; Klein-Zabban and others limited, accelerated strategies provide an attractive option 1987). Nosocomial transmission has also been reported from in achieving coverage and disease reduction targets. Taiwan (Gao and Malison 1988); severity of illness wasgreater 182 Stanlrv 0. Foster, Deborah A. McFarland, and A. MeredithJohn in children who had attended the clinic for illness than in Murmansk and Pskov areas in the former U.S.S.R. and in those who had attended for well baby care (P < .01). Immuniza- Czechoslovakia (Davis 1991). tion at every opportunity, as advocated by WHO, would have prevented most of these cases. VITAMIN A SUPPLEMENTATION IN DEFICIENT AREAS. The World In urban populations in westem and central Africa, measles Health Organization has recommended that vitamin A sup- is primarily a disease of the first two years of life, an age at which plementation become a routine part of immunization pro- cFRs are highest (Taylor and others 1988). Children in urban grams. Specifically, it recommends the administration of slums have been identified as at increased risk for high measles 200,000 international units (Iu) to mothers at time of delivery mortality (low coverage and low age of infection), and thus orduring the next four weeks, and 25,000 iu at each imrnuniza- such slums are a priority area for targeted immunization tion contact beginning at six weeks of age and with at least (Coetzee, Berry, and Jacobs 1991). Lot quality assessment four-week intervals (WHO/EPI 1989a, 1992). This recommenda- sampling has been used to identify low coverage areas in tion is based on data from Indonesia and India showing that Kinshasa, Zaire. Priority attention to these high-risk groups vitamin A supplementation to children in an area of vitamin will have maximum effect on measles-associated mortality. A deficiency reduced overall mortality (Rahmathullah and Three guides to improve urban immunization have recently others 1990; Sommer 1990). Although Rahmathullah and been published (UNICEF 1989; Claquin 1991; Cutts 1991). colleagues (and other researchers) did not show a clear reduc- Targeting vaccine to places and groups for which epidemiologic tion in measles or other specific cause of mortality, there is a data document increased mortality risk (for example, supplemen- growing consensus that vitamin A supplementation in defi- taryfeedingcentersandgirlsinpoorhomesinBangladesh[Bhuiya cient areas will reduce measles case-fatality rates and increase and others 1987]), will increase the efficiency of immunization in child survival. achieving the mortality reduction goal. TREATMENT OF SEVERE MEASLES WITH VITAMIN A. Therapeutic ADOPTION OF TWO-DOSE MEASLES VACCINE SCHEDULES. Ninety doses of vitamin A are now recommended for children with percent coverage with a vaccine producing 90 percent efficacy severe cases of measles. In a placebo-control double-blind will provide protection to 81 percent ofvaccinees, significantly study in South Africa, the risk of death or severe measles less than the 1995 disease reduction target of 90 percent. complication was reduced by half (RR 0.52 95 percent ci Two-dose schedules have the potential to facilitate the 0.35-0.74) through administration of 400,000 Iu of retinyl achievement of the 90 percent measles reduction target by palmitate (Hussey and Klein 1990). reducing the number of primary vaccine failures. Two-dose schedules can use the currently available Schwarz vaccine or EFFECTIVE TREATMENT OF MEASLES COMPLICATIONS. Most mea- an improved vaccine providing 85 percent seroconversion for sles deaths are due to complications, a high proportion of those vaccinated at six months of age. Three two-dose sched- which can be effectively treated through standard treatment ules are provided as examples: practices. Data from Senegal suggest that treatment in the first a In dense urban areas where risk of infection is highest in few days of illness can reduce measles CFRs by 78 percent * I dese rba araswhee rsk f ifetio ishigestin (Garenne 1992). the first two years of life, two doses of EZ or equivalent vaccine need to be given as early as possible, for example, at EXPANSION OF THE INFRASTRUCTURE. In many areas, at-risk six months and twelve months of age. infants have limited access to vaccination services. Develop- * In rural areas with low coverage, immunization at nine ment of new vaccine delivery points in such places has to be a and fifteen months of age with Schwarz or equivalent vac- long-term priority. cines would be appropriate. * In countries where high immunization coverage has Research Priorities shifted the age distribution of measles cases to school- children, a second dose of measles vaccine at school entry Improved vaccines and implementation strategy will be re- should be considered. School immunization not only de- quired to achieve the 1995 targets. Research is a continuing creases infection in susceptible older children but also de- priority. creases the risk of morbidity and mortality in their preschool siblings (measles transmission in schoolchildren has been OPERATIONAL RESEARCH. The currently available tools (mea- identified as a source of infection for their high-risk pre- sles vaccines, cold chain, disposable and reusable needles and school siblings). In Burundi, twenty-five out of twenty-eight syringes) have the potential of significantly aiding the effort to cases of measles in school-age children were index cases in reduce measles and measles-associated morbidity and mortal- their households and the source of infection for thirty-one ity. Operational research is needed to identify the optimal use secondary cases, twenty-eight of whom were younger of these tools to achieve the maximum effect, for example, use siblings (Chen 1990). Introduction of two-dose schedules of two-dose schedules, targeting of high-risk groups, and accel- has reduced measles transmission to very low levels in erated vaccination strategies in urban areas. Measles 183 VACCINE DEVELOPMENT. Although the current more heat sta- * Operational research to ensure maximum effective use of ble vaccine is a highly effective vaccine, further improvements available technologies within the epidemiologic and re- in measles vaccine could significantly increase the effective- source realities of the local environmcnt. ness of efforts to control measles. The ideal criteria for a * Strengtheneddecentralizedmanagementandownership measles vaccine in the developing world, based on experience in the planning, implementation, and evaluation of im- acquired in the 1980s and the expected improvements to be munization program. gained by the introduction of vaccine capable of providing protection at six months of age, have not yet been met. Listed * Development and use of management and disease infor- below are suggested criteria for such a vaccine: mation systems to strengthen decisionmaking, implementa- tion, and evaluation. * Heat stable at 370C for twelve months. In the developing * Continued awareness and commitment of bilateral and world, the areas with highest measles-associated mortality, international technical assistance agencies on the need of lack of a reliable cold chain limits many health workers, developing countries for continuing foreign-exchange sup- especially private practitioners, from providing immuniza- port for vaccines and cold-chain equipment. tion. Resources for fuel and refrigerator maintenance, repair, and replacement are expected to shrink during the next decade. Notes * Ability to achieve 95 percent seroconversion and life-long protection when administered at three months of age or earlier. We appreciate and acknowledge Felicity Cutts, Michael Deming, Michelle Access to health facilities is inversely related to age-the Garenne, Mark Grabowsky, Rafe Hlenderson, Bert Ilirschorn, Laurie younger the age at immunization, the greater the probable Markowitz, Walter Orenstein. and Akanne Sorungbe for their thoughtful contact of that child with health facilities and the opport,, review of sections of this chapter in manuscript form. The expert clerical contact of that child with health facilities and the opportu- support of Pat Jennings, Judith Clark, Quin Long, and Arvis McCormick is nity for immunization. An effective vaccine for three- also recognized. month-old infants would prevent almost all the measles I. Except whereTnoted otherwise, alildollaramounts are current U.S. dollars. cases that occur before nine months of age. As increasing 2. Missed opportunities: contacts between a child needing vaccination and numbers of infants in the developing world are born to a health facility with vaccine delivery capability at which needed vaccinations mothers whose antibodies resulted from immunization are not provided. rather than wild virus infection, infection in the first six months of life may increase in frequency. Immunization in References the first few months of life will be needed to address this problem. A major initiative to develop such a vaccine is Aaby, Peter. 1988. Malnourished or Overinfected: An Analysis of the Detertmi- under way (Bart and Lin 1990). nants of Acute Measles Mortality. Copenhagen: Laegeforeningens Forlag. * Prepackaged in a single-dose non-reusable syringe. Single- Aaby, Peter, Jette Bukh, 1. M. Lisse, and A. J. Smits. 1981. "Measles Vaccina- unit packaging would facilitate expanding vaccine delivery tion and Child Mortality." Lancet 2:93. to private practitioners and nurse-midwives operating out- Aaby, Peter, T. G. Jensen, H. L. Hansen, Hans Knstiansen, Jesper Tharup, side of the health facilities. Use of self-destruct syringes Anja Poulsen, Morten Sodemann, Marianne Jakobsen, Kim Knudsen, M. would eliminate sterilization costs and the risks of human C. da Silva, Hilton Whittle. 1988. "Trial of High-Dose Edmonston-Zagreb immunodeficient virus Hivcosts andhepatnetns otransmisson Measles Vaccine in Guinea-Bissau: Protective Efficacy." Lancet 2:809-14. immunodeficient virus (HIv) and hepatitis B transmission Aaby, Peter, and). Leeuwenburg. 1988. "Pattems of Transmission and Sever- through reuse of syringes. ity of Measles Infection: A Reanalysis of Data from the Machakos Area." * Affordable. Vaccine cost should be in the range of the Kenya Journal of lnfectous Disease 161:171-74. current $0.10 to $0.15 per dose. Aaby, Peter, 1. R. Pederson. Kim Knudsen, M. C. da Silva, C. H. Mordhorst, N. C. Helm-Petersen, B. S. Hansen, Jesper Tharup, Anja Poulsen, Morten Conclusions Sodemann, Marianne Jakobsen. 1989. "Child Mortality Related to Sero- conversion or Lack of Seroconversion after Measles Vaccination." Pediatric Measles immunization is a proven, cost-effective primary health Infectious Disease joural 8:197-200. ACIP. 1989. "Measles Prevention. Recommendations of the Immunization care intervention capable of reducing morbidity and mortality Practices Advisory Committee." Morbidity and Mortalty Weekly Report and increasing child survival. The use of current vaccines and 38:1-18. strategies will not, however, achieve the targets (90 percent Anderson, R. M., and R. M. May. 1985. "Age-Related Changes in the Rate of coverage, 90 percent morbidity reduction, and 95 percent mor- Disease Transmission: Implications for the Design of Vaccination Pro- tality reduction) endorsed at the 1990 World Summit for Chil- grams."]ournalofHygiene 94:365-436. dren. Five priorities have been identified for the 1990s: Atkinson, W. L., and L. E. Markowitz. 1991. "Measles and Measles Vaccine." PediatTic Infectious Disease Journal 2:100-7. * Development of a heat-stable vaccine providing 85 per- Barclay A. J. G., A. Foster, and A. Sommer. 1987. "Vitamin A Supplements cent or higher protection when administered at six months and Mortality Related to Measles: A Randomized Clinical Trail." British of age or earlier. MedicalJournal 294:294-96. 184 Stanley 0. Foster, Deborah A. McFarland, and A. Meredith John Barnum, H. 1989. "Evaluating Healthv Days of Life Gained from Health Cutts, F. T. 1990. Measles Control in the 1990s: Principles for the Next Decade. Projects." Social Science and Medicine 24:833-41. Geneva: World Health Organization. Bart, K. S., and K. F. Lin. 1990. "Vaccine Preventable Disease and Immuniza- . 1991. "Strategies to Improve Immunization Services in Urban Af- tion in the Developing World." Pediatric Clinics of North America 37: rica." Bulletin of the World lealth Organization 69:407-14. 735-56. Cutts, F. T., D. C. Glik, A. Gordon, K. A. Parker, S. Diallo, F. Haba, and R. Bender, D., and R. J. Macauley. 1990. "Immunization Drop-Outs and Maternal Stone. 1990. "Application of Multiple Methods to Study the Immunization Behavior: Evaluation of Reason Given and Strategies for Maintaining Programme in an Urban Area of Guinea." Bulletin of the World Health Gains Made in the National Vaccination Campaign in liberia." Interna- Organization 68:769-76. tional Quarterty of Health Education 9:283-88. Cutts, F. T., R. H. Henderson, C.]. Clements, R. T. Chen, P. A. Patriarca. Bhuiva, Abbas, Bogdan Wojtyniak, Stan D'Souza, Lutfun Nahar, Kashem 1991. "Principles of Measles Conro." Bulletin of the World Health Organi- Skaikh. 1987. "Measles Case Fatality among the Under-Fives: A Multivar- zation 69:1-7. iate Analysis of Risk Factors in a Rural Area of Bangladesh." Social Science Cutts, F. T., Kortbeeks, R. Malalane, P. Penicele, K. Gingell. 1988. "The and Medicine 24:439-43. Development of Appropriate Strategies for EPI: A Case Study from Mozam- Black. F. L. 1982. "Measles." In Evans, A. S., ed., Viral Infectons of Humans. bique." Health PolicN and Planning 3:291-301. Epidemiology and Control, 2d. New York: Plenum Medical Book Companv'. Dabis, F., A. R. Sow, R. J. Waldman. P. Bikakouri. J. Senga, G. Madzou, and - 1989. "Measles Active and Passive Immunity in a Worldwide Per- T. S. Jones. 1988. "The Epidemiology of Measles in a Partially Vaccinated spective." Prog Med Virol 36:1-33. Population in an African City: Implications for Immunization Programs." Bloch, A. B., A. V. K. V. de Silva, R. L. de Sylva. "The Public Health AmencanJournalof Epidemiology 127:171-78. Importance of Measles in Sri Lanka." Typescript. South-East Asia Regional Davis, R. M. 1991. "Revised Measles Agenda for the 1990s: From Control to Office, WHO, New Delhi, India. Pre-eradication." International Chuld Health 11 3:45-50. Bloch, A. B., W. A. Orenstein, W. M. Ewing, W. H. Spain, G. F. Mallison, Davis, R. M., E. D. Whitman, W. A. Orenstein, S. R. Preblud, L. E. Markowitz, D. L. Herrmann, A. R. Hinman. 1985. "Measles Outbreak in a Pediatric and A. R. Hinnan. 1987. "A Persistent Outbreak of Measles Despite Practice: Airbome Transmission in an Office Setting." Pediatrics 75:676-83. Appropriate Control Measures." AnsericanJournal of Epidemiology 126:438- Boerma, J. T., A. E. Sommerfelt, S. 0. Rutsein, and G. Rojas. 1990. "Im- 49 munizarion: Levels, Trends and Differentials." Institute for Resource Devel- Dietz, K. 1976. "The Incidence of Infectious Diseases under the Influence of opment, Columbia, Md. Seasonal Fluctuations." Lecture Notes in Biornathemaucs I 1:1-1 5. Borgono, J. NI. 1983. "Current Impact on Measles in Latin Anmerica." Review Dossetor, J., H. C. Whittle, and B. M. Greenwood. 1977. "Persistent IMeasles of Inlfectious Diseases 5:417-21. Infection in Malnourished Children." Bntish Medical Journal 1: 1633-35. Boyd, D. 1991. "Computerized EPI Information Systems (CEns)." Resources f or Drinkwater, H. 1885. Rernarks upon the Epidemic ofMeasles Prevalent in Sunder- Child Health, Arlington, Va. land. Edinburgh: James Thin. Brenzel, Logan. 1990. "The Cost of EPI: A Review of Cost and Cost-Effective- Duflo, B., H. Balique, P. Ranque, A. N. Diallo, G. Brucker, H. Alavi, and N. ness Studies (1979-1987)." Resources for Child Health, Arlington, Va. Prescott. 1986. "Estimation de limpact des principales maladies en zone 1991. "Cost and Financing of EPI." Resources for Child Health, rurale malienne." Revue d'Epideniologie et de Sant Publque 34:405-18. Arlington, Va. Edmonson, M. B., D. G. Addiss, J. T. McPherson, J. L. Berg, S. R. Circo, and Brenzel, L., and P. Claquin. 1991. "Immunization Prograins and Their Costs." J. P. Davis. 1990. "Mild Measles and Secondary Vaccine Failure during a World Bank Health Sector Priorities Review HSPR-04. Washington, D.C. Sustained Outbreak in a Highly Vaccinated Population."JAMA 263:2467- Bryce, J. W., F. T. Cutts, and S. Saba. 1990. "Mass Immunization Campaigns 71. and Quality of Immunization Services." Lancet 1:739-40. Ekunwe, E. 0. 1984. "Expanding Immunization Coverage through Improved Cardenas-Ayala, V. M., C. Sanchez-Vargas. 1989. "Estimation de la raz6n Clinic Procedures." World Healh Forum 5:361-63. heneficio/costo de [a vacunaci6n contra el sarampi6n." Salud Publica Mexico Ellison, J. B. 1932. "Intensive Vitamin Therapy in Measles." Britsh Medical 31:735-44. Journal 2:708-11. CCCD (Centers for Disease Control). 1990. African Child Survival Intiative- Enders, J. F., and T. C. Peebles. 1954. "Propagation in Tissue Cultures of Combating Childhood Communicable Diseases, 1989-1990. Annual Report. Cytopathogenic Agents from Patients with Measles." Proceedings of the Atlanta, Ga. Society for Expenmental Biology 86:277-86. Chand, Phool, R. N. Rai, Umwa Chawla, K. C. Tripathi, K. K. Datta. 1989. Fagbule, D., and F. Orifunmishe. 1988. "Measles and Childhood Mortality in "Epidemiology of Measles-A Thirteen Year Prospective Study in a Vil- Semi-urban Nigeria." AfricanJournal ofMedscalScience 17:181-85. lage." Journal ofCommunicable Disease 21:190-99. Feachem, R. G., and M. A. Koblinsky. 1983. "Interventions for the Control Chen, R. 1990. "Measles in Muyinga Health Sector, Burundi, 1989-1990." of Diarrhoeal Diseases among Young Children: Measles Immunization." Field report. Centers for Disease Control, Atlanta, Ga. Bulletin of the World lIealth Organization 61:641-52. Claquin. P. 1991. "Urban Ei'." Resources for Child Health. Arlington, Va. Fine, P. E. M., and]. A. Clarkson. 1982a. "Measles in England and Wales-1: Clemens, J. D., Bonita F. Stanton, J. Chakraborty, Shahriar Chowdhury, M. An Analysis of Factors Underlying Seasonal Pattems." InternationalJournal R. Rao, Mohammed Ali, Susan Zimicki, Bogdan Wojtnyiak. 1988. "Measles of Epidemiology 11:5-14. Vaccination and Childhood Mortality in Rural Bangladesh." Amenicais . 1982b. "Measles in England and Wales-lI: The Impact of Vaccina- Journal of Epidemiology 128:1330-39. tion Programmes on the Distribution of Immunity in the Population." Clements, C. J., .J B. Milstein, Mark Grabowsky, and J. Gibson. 1988. IsitemationalJournalofEpidemiology 11:15-25. "Research into Altemative Measles Vaccines in the 1990's." Working paper Foege, W. H. 1984. "Banishing Measles from the World." World Heakh Forum EPI/GEN/88.1 1. EPI Global Advisory Group. WHO/EPI, Geneva. 5:63-65. Coetzee, N., D. S. Berry, and M. E. Jacobs. 1991. "Measles Control in an Foege, W. H., and S. O. Foster. 1974. "Multiple Antigen Vaccine Strategies Urbanizing Environment." South Afncan MedicalJournal 79:440-44. us DevelopingCountries."AmericanJournalofTropicalMedmcineandHygiene Cotton,M.F.,F. E. Berkowitz,Z. Berkowitz, P.J. Becker, and C. Heney. 1989. 23:685-89. "Nosocomial Infections in Black South African Children." Pediamc Ilnfec- Forgy, L., K. Mc[nnes, S. Heinig, and B. Michaels. 1990. Projected Costs of the tious Disease Joutnial 8:676-83. Talloires Targets. Abt and Associates. Washington, D.C. Measles 185 Foster, A., and G. J. Johnson. 1988. "Measles, Comeal Ulceration, and among Unvaccinated Preschool-Aged Children: Opportunities Missed by Childhood Blindness: Prevention and Treatment." Tropical Doctor 18:74- HealthCare Providers to AdministerMeaslesVaccine."Pediatme83:369-74. 78. Imperato, P. J., and D. Traore. 1969. "Traditional Beliefs about Measles and Foster, S. 0. 1989. UNICEFEPI consultation, Sultanate of Oman. Its Treatment among the Bambara of Mali." Tropical and Geographic Medi- Foster, S. O., J. Shepperd, J. H. Davis, and A. N. Agle. 1990. "Working with cine 21:62467. African Nations to Improve the Health of Their Children." JAMA 263: Job,J. S., N. A. Halsey, Reginald Boulos, Elizabeth Holt, Dorothy Farrell, Paul 3303-5. Albrecht, J. R. Brutus, Mario Adrien, Jean Andr6, Edward Chan, Patricia Foulon, G., M. L. Klein-Zabban, L. Gnansov-Nezzi, and G. Martin-Bouyer. Kissinger, Carlo Boulos. and the Cite Soleil/JHU Project Team. 1991. 1983. "Preventing the Spread of Measles in Children's Clinics." Lancet "Successful Immunization of Infants at 6 Months of Age with High Dose 2:1498-99. Edmonston-Zagreb Measles Vaccine." Pediatric Infectous Disease Journal Gao, J. P., and M. D. Malison. 1988. "The Epidemiology ofa Measles Outbreak 10:303-11. on a Remote Offshore Island neat Taiwan." lnternational]ournalofEpidemi- John, A. MN. 1990a. "Endemic Disease in Host Populations with Fully Specified ology 17:894-98. Demography." Population Biology 37:455-71. Garenne, Michel, and Peter Aaby. 1990. "Pattem of Exposure and Measles - 1990b."Transmissionand Control of Childhood Infectious Diseases: Mortality in Senegal."Journal of Infectious Diseases 161:1088-94. Does Demography Matter?" Population Studies 44:195-215. Garenne, Michel, and P. Cantrell. "Rougeole et mortalite au Senegal: 6tude John, A. M., and S. D. Tuljapurkar. 1990. "Childhood Infectious Diseases in de l'impact de la vaccination effectu6e a Khombole 1965-1968 sur la survie LDCs: Immunization Program Design and Evaluation Using Demographic- des entants. Estimation de la mortalit6 du jeune enfant pout guider les Epidemiologic Models." Working Paper 22, Research Division, Population actions de sant6 des pays en developpement." Institut Nanonal de Sciences Council, New York. et de Recherche M6dicale, Paris. 145:512-32. John, T. Jacob, Abraham Joseph, T. 1. George, Janaki Radhakrishnan, Garenne, Michel, Odile Leroy, Jean-Pierre Beau, and Ibrahima Sene. 1991. Rajdayal Singh, Kuryan George. 1980. "Epidemiology and Prevention of "Child Mortality after High-Titre Measles Vaccines: Prospective Study in Measles in Rural South India." International Joumnal of Medical Research Senegal." Lancet 2:903-7. 72:153-58. Ghana Health Assessment Project Team. 1981. "A Quantitative Method of John, T. Jacob, M. Ray, and M. C. Steinhoff. 1984. "Control of Measles by Assessing the Health Impact of Different Diseases in Less Developed Annual Pulse Immunization." Amercan Journal of Diseases of Children: 138: Countries." Internationaljournal of Epidemiology 10:73-80. 299-300. Ginsberg,G.M., and T. H. Tulchinsky. 1990."CostsandBenefitsofaSecond Kalter, H. D., R. H. Gray,R. E. Black, and S. A. Gultiano. 1990. "Validation Measles Inoculation of Children in Israel, the West Bank, and Gaza." of Postmortem Interviews to Ascertain Selected Causes of Death in Chil- Journal of Epidemiology and Community Health 44:2 74-80. dren." International Journal of Epidemiology 19:380-86. Gordon, J. E., A. A. Jansen, and W. Ascoli. 1965. "Measles in Rural Guate- Kaschula, R. O., J. Druker, and A. Kipps. 1983. "Late Morphologic Conse- mala. Pediatrics 66:779-86. quences of Measles-A Lethal and Debilitating Lung Disease among the Grabowsky, M. 1991. "Missed Opportunities." Resources for Child Health, Poor." Review of Infectious Diseases 5:395-404. Arlington, Va. Kasongo Project Team. 1981. "Influence of Measles Vaccination on Survival Greenwood, B. M., A. M. Greenwood, A. K. Bradley, S. Tulloch, R. Hayes, Pattern of 7-35 Month Old Children in Kasongo, Zaire." Lancet 1:764467. and F. S. Oldfield. 1987. "Deaths in Infancy and Early Childhood in a Well Kenyan Ministry of Health and WHO (World Health Organization). 1977. Vaccinated, Rural, West African Population." Annals of Tropical Pediatrics "Measles Immunity in the First Year after Birth and the Optimum Age for 2:91-99. Vaccination in Kenyan Children." Bulletn of the World Health Organization Guyer, B., and A. M. McBean. 1981. "The Epidemiology and Control of 55:21-31. MeaslesinYaound6,Cameroon, 1968-1975." InternationalJournalofEpide- Kermack, W. O., and A. G. McKendrick. 1927. "A Contribution to the miology 10:263469. Mathematical Theory of Epidemics." Proceedings of the Royal Society of Halsey, N. A., Reginald Boulos, Franz Mode, Jean Andre, Linda Bowman, R. London, series A, 11 5:700-2 1. G. Yaeger, Serge Toureau, Joh Rohde, and Carlo Boulos. 1985. "Response Killewo, Japhet, Cyprian Makwaya, Emmanuel and Rose Munubhi, and to Measles Vaccine in Haitian Infants 6 to 12 Months Old." New England Mpembeni. 1991. "The Protective Effect ofMeasles Vaccine underRoutine Journal of Medicine 313:544-49. Vaccination Conditions in Dar Es Salaam, Tanzania: A Case-Control Heiby,J. R. 1990. Supertrision and the Quality of Care in the Expanded Programme Study." InternationalJournal of Epidemiology 20:508- 14. on Immunization. WHCI/EPJGAG/WP. Geneva. Kipps, A., and R. 0. C. Kaschula. 1976. "Virus Pneumonia following Measles." Hethcote, H. W. 1976. "Qualitative Analysis of Communicable Disease South African MedicalJournal 50:1083-88. Models." Mathematcal Biosciences 28:335-56. Klein-Zabban, M. L., G. Foulon, C. Gaudebout, J. Badoual, and J. Assi Adou. Hirschom, N. 1990. "Missed Opportunities for Immunization." Resources for 1987. "Fr6quence des rougeoles nosocomiales dans un centre de protection Child Health, Arlington, Va. matemelle et infantile d'Abidjan." Bulletin of the World Health Organization Holt, E. A., Reginald Boulos, N. A. Halsey, L. M. Boulos, C. Boulos. 1990. 65:197-201. "Child Survival in Haiti: Protective Effect of Measles Vaccination." Koenig, M. A., M. A. Khan, B. Wojtyniak, J. D. Clements, J. Chakraborty, V. Pidiatmie 85:188-94. Fauveau, J. F. Phillips, J. Akbar, and U. S. Barnua. 1990. "Impact of Measles Hopkins, D. R., J. F. Koplan, A. R. Hinman, and J. M. Lane. 1982. "The Case Vaccination on Childhood Mortality in Rural Bangladesh." Bulletn of the for Global Measles Eradication." Lancet 1:1396-98. World Health Organization 68:441-47. Hussey, G. D., and M. Klein. 1990. "A Randomized Controlled Trial of Koster,F.T.,G.C.Curlin,K.M.A.Azia,andAzizulHaque. 1981."Synergistic Vitamin A in Children with Severe Measles." New England Journal of Impact of Measles and Diarrhoea on Nutrition and Mortality in Bangla- Medicine 323:160Q64. desh. " Bulletin of the World Health Organization 59:901-8. Hussey, G. D., and J. Simpson. 1991. "Nosocomial Bacterias in Measles." Krober, M. S., C. E. Stracener, and J. W. Bass. 1991. "Decreased Measles Pediatric Infectious DiseaseJournal 9:715-17. Response after Measles-Mumps-Rubella Vaccine in Infants with Colds." Hutchins, S. S., J. Escolan, L. E. Markowitz, C. Hawkins, A. Kimbler, R. A. JAMA 265:2095-96. Morgan, S. R. Preblud, and W. A. Orenstein. 1989. "Measles Outbreaks Lesotho. 1990. Measles control in Lesotho. 186 Stanley 0. Foster, Deborah A. McFarland, and A. Meredith John Loras-Duclaux, I., L. David, D. Peyramond D. Floret, A. Lachaux, and M. Pison, G., and N. Bonneuil. 1988. "Increased Risk of Measles Mortality for Hlermier. 1988. "Etude 6pidemiologique et evaluation du coOt de la rougeole Children with Siblings among the Fula Bande, Senegal." Review of Infectious dans les h6pitaux lyonnais durant cinq annees." Pediatrie 43:451-54. Diseases 10:468-70. McBean, A. M., S. 0. Foster, K. L. Herrmann, and C. Gateff. 1976. "Evalua- Ponninghaus, J. M. 1980. "The Cost/Benefit of Measles Immunization: A tion of a Mass Measles Immunization Campaign in Yaounde, Cameroon." Study from Southem Zambia." Journal of Tropical Medicine and Hygiene Transacions of the Royal Society of Tropical Medicine and Hygiene 70:206-12. 83:141-49. McFarland, D. A., and D. K. Kraushaar. 1990. "Cost of the EPI and CDD Preblud, S. R., and S. L. Katz. 1988. "Measles Vaccine." In S. A. Plotkin and Programmes in Swaziland." Centers for Disease Control, Atlanta, Ga. E. A. Mortimer, eds., Vaccines. Philadelphia: W. B. Saunders. Maina-Ahlberg, B. 1979. "Beliefs and Practices Conceming Treatment of Prescott, N., A. Prost, and R. Le Berre. 1984. "The Economics of Blindness Measles and Acute Diarrhea among the Akamba." Tropical and Geographic Prevention in Upper Volta under the Onchocerciasis Control Program." Medicine 31:139-48. Social Science and Medicine 19:1051-55. Markowitz, L. 1990. "Measles Control in the 1990s. Immunization before 9 de Quadros, C. A., J. K. Andrus, J. M. Olive, C. M. da Silveria, R. E. Eikhof, Months of Age." World Health Organization, Geneva. Peter Carrasco, J. W. Fitzsimmon, and Francisco P. Pinheiro. 1991. "Erad- Markowitz, L. E., and P. Nieburg. 1991. "The Burden of Acute Respiratory ication of Pohomyelitis: Progress in the Americas." Pediatric Infectious Infection Due to Measles in Developing Countries and the Potential Impact Disease Journal 10:222-29. of Measles Vaccine." Review of Infectious Diseases 13:555-61. Rahmathullah, Laxmi, B. A. Underwood, R. D. Thulesiraj, R. C. Milton, Markowitz, L. E., N. Nzilambi, W. J. Driskell, M. G. Sension, E. Z. Rovira, P. Kala Ramaswamy, Raheem Rahmathullah, Geneesh Babu, and B. Com. Nieburg, and R. W. Rider. 1989. "Vitamin A Levels and Mortality among 1990. "Reduced Mortality among Children in Southern India Receiving Hospitalized Measles Patients, Kinshasa, Zaire."Journalof Tropical Pediatrics a Small Weekly Dose of Vitamin A." New England Journal of Medicine 35:109-12. 323:929-35. Markowitz, L. E., and W. A. Orenstein. 1990. "Measles Vaccines." Pediatric Robertson, R. L. 1985. Cost of the CCCD Project in Swaziland, 1984-1985. Clinics of North America 37:603-25. University Research Corporation. Markowitz, L. E., S. R. Preblud, P. E. M. Fine, and W. A. Orenstein. 1990. Robertson, R. L., S. 0. Foster, H. F. Hull, and P. J. Williams. 1987. "Cost' "Duration of Live Measles Vaccine-Induced Immunity." Pediatric Infec- Effectiveness of Immunization inThe Gambia-"Journalof Tropical Medicine tious Disease Journal 8:101-10. and Hygiene 88:343-51. Markowitz, L. E., J. Sepulveda, J. L. Diaz-Ortega, J. L. Valdespino, P. Rosenthal, G. 1990. "Sustainability of EPI." Resources for Child Health, Albrecht, E. R. Zell, J. Stewart, M. L. Zarate, and R. H. Bernier. 1990. Arlington, Va. "Immunization of Six-Month-Old Infants with Different Doses of Roungou, J. B. 1991. Coverage Survey, Central African Republic. Edmonston-Zagreb and Schwarz Measles Vaccine." New EnglandJournal Sabin, A. B., A. Flores Arechiga, J. Femandez de Castro, J. L. Sever, D. L. of Medicine 322:580-87. Madden, I. Shekarchi, P. Albrecht. 1983. "Successful Immunization of Mast, Eric, J. L. Berg, L. P. Hanrahan, J. T. Wassell, and J. P. Davis. 1990. Children with and without Matemal Antibody by Aerosolized Measles "Risk Factors for Measles in a Previously Vaccinated Population and Cost- Vaccine." JAMA 249:2651-52. Effectiveness of Revaccination Strategies." JAMA 264:2529-33. Sahuguede, P., A. Roisin, 1. Sanou, B. Nacro, F. Talts. 1989. "Epid6mie de Mathias, R. G., W. G. Meekison, T. A. Arcand, and M. T. Schechter. 1989. rougeole au Burkina Faso: 714 cas hospitalises a lhospital de Bobo, "The Role of Secondary Vaccine Failures in Measles Outbreaks." American Dioulasso." Annals of Pediatrics 36:244-51. Journal of Public Health 79:475-78. Schenzle, D. 1989. "An Age-Structured Model of Pre- and Post-vaccination Morley, D. 1973. "'Severe' Measles." In D. Morley, ed., Pediatric Prionties in the Measles Transmission." iMA (International Mathematics AssocLaton) Journal Developing World. London: Butterworth. of Mathematics Appled to Medicine and Biology 1:16-91. Mosely, W. H., and L. C. Chen. 1984. "An Analytic Framework for the Study Shears, P., A. M. Berry, R. Murphy, and M. A. Nabil. 1987. "Epidemiologic of Child Survival in Developing Countries." In Child Survival Strategies for Assessment of the Health and Nutrition of Ethiopian Refugees in Emer- Research. Supp. Population and Development Review 10S:25-45. gency Camps in Sudan." Bntnsh MedicalJournal 295:314-18. Narain, J. P., S. Khare, S. R. S. Rana, and K. B. Banerjee. 1989. "Epidemic Shepard, D. S., L. Sanoh, and E. Coffi. 1986. "Cost-Effectiveness of the Measles in an Isolated, Unvaccinated Population, India." International Expanded Programme on Immunization in the Ivory Coast: A Preliminary Journal of Epidemiology 18:952-58. Assessment." Social Science and Medicine 22:369-77. Nokes, D. J., A. R. McLean, R. M. Anderson, and M. Grabowsky. 1990. Smith, E. A., and S. O. Foster. 1970. "The Effect of the Smallpox Eradica- "Measles Immunization Strategies for Countries with High Transmission tion Measles Control Programme on Measles Admissions to the Lagos Rates: Interim Guidelines Predicted Using a Mathematical Model." Inter- Infectious Disease Hospital, Yaba, Nigeria." West African MedicalJournal national Journal of Epidemiology 19:703-10. 19:51-56. Orenstein, W. A., R. H. Bemier, T. J. Dondero, A. R. Hinman, J. S. Marks, Sommer, A. 1990. "Vitamin A Status, Resistance to Infection,andChildhood K. J. Bart, and B. Sirotkin. 1985. "Field Evaluation of Vaccine Efficacy." Mortality." Annals of the New York Academy of Science 587:17-23. Bulletin of the WorldHealthOrganization63:1055-68. Taylor, W. R., Ruti-Kalisa, Mambu Ma-Disu, and J. Xi. Weinman. 1988. Orenstein, W. A., L. Markowitz, S. R. Preblud, A. R. Hinman, A. Tomasi, and "Measles Control Efforts in Urban Africa Complicated by High Incidence K. J. Bart. 1986. "Appropriate Age for Measles Vaccination in the United of Measles in the First Year of Life." American Journal of Epidemiology States." Developments in Biological Standards 65:13-21. 127:788-94. PAHO (Pan-American Health Organization). 1990. "Plan to Eliminate Indig- Toole, M. J., R. W. Steketee, R. J. Waldman, and P. Nieburg. 1989. "Measles enous Transmission of Measles in the English Speaking Caribbean." Bulletin Prevention and Control in Emergency Settings." Bulletin of the World Health of the Pan-American Health Organization 24:240-46. Organization 67:381-88. Panum, P. L. 1939. "Observations Made during the Epidemic of Measles on Toole, M. J., and R. J. Waldman. 1988. "An Analysis of Mortality among the Faroe Islands in the Year 1846." Medical Classics 3:839-86. Refugee Populations in Thailand, Somalia, and Sudan." Bulletin ofthe World Phillips, M. A., R. G. Feachem, and A. Mills. 1987. Options for Diarrhoea HealthOrganization 66:237-47. Control. EPC Publication 13. London: School of Hygiene and Tropical Tuljapurkar, S. D., and A. M. John. 1991. "Disease in Changing Populations: Medicine. Growth and Disequilibrium." Population Biology 40:322-53. Measles 187 lJNICEF. 1990. Universal Child Immunization Reaching the Urban Poor. Urban . 1989a. "High Dose EZ Measles Vaccine-" EPI/GAG/89/WPI3. Geneva. Examples 16. New York. - . 1989b. "Vitamin A Update." EP1/GAG/89/WP13. Geneva. United States, National Vaccine Advisory Committee. 1991. The Measles . 1991. "PlanofAcion forGlobal MeaslesControl." EPI/GAG/91/WXPl2. Epidemic: The Problems, Barriers, Recommendations. Geneva. Verduzco, E., C. Calderon, and E. Velazquez-Franco. 1974. "Repercusiones de . 1992. "Using Immunization Contacts to Combat Vitamin A Defi- [a vacunacion contra el sarampion en M6xico." Salud PW'ilca Mexico 16: ciency." EPI/GAG/92. 707-20. WHO/EPI (World Health Organization/Expanded Programme on Immuniza- Voorhoeve, A. MN., A. S. Muller, T. W. Schulpen, W. Gemert, H. A. Valken- tion). 1984. "Expanded Programme on Immunization, Indications and burg, and H. E. Ensering. 1977. "Agents Affecting Health of Mother and Contraindications for Vaccines Used in the EPI." WHO/WER (Weekly Epi- Child in a Rural Area of Kenya. III: The Epidemiology of Measles." Tropical demological Record) 59:13-16. and Geographic Medicine 29:428-40. . 1985. "Expanded Programme on Immunization, Public Health Im- Walsh, J. A. 1983. "Selective Primary Health Care: Strategies for Control of portance of Measles." WHO/WER 60:103-5. Burma. Disease in the Developing World. IV: Measles." Reviewof Infectious Diseases . 1979. "Expanded Programme on Immunization, Measles Immuniza- 5:330-40. tion." WHO/WER 54:337-39. Kenya. Weierbach, R. 1989. "Rapport de mission au Rwanda du Juillet au 4 Aout." - . 1986. "Expanded Programme on Immunization, Measles Vaccine Centers for Disease Control, Atlanta, Ga. Efficacy." WHO/WER 61:356-57. Poland. White, C. C., J. P. Koplan, and W. A. Orenstein. 1985. "Benefits, Risks, and _ . 1985. "Public Health Importance of Measles." WHO/WER 60:95-97. Costs of Immunization for Measles, Mumps, and Rubella." AmericanJournal Sri Lanka. Of Public Health 75:739-44. - 1989. "Expanded Programme on Immunization, Missed Opportunities Whittle, H. C., A. Bradley-Moore, A. Fleming, and B. M. Greenwood. 1973. for Immunization." WHO/WER 64:32-34. Zimbabwe. "Effects of Measles and the Immune Response of Nigerian Children." WHO/UNICEF (World Health Organization/United Nations Children's Fund). Archives of Diseases of Children 48:753-56. 1985. Planning Principles for Accelerated Immunization Strategies. A Joint Whittle, H. C., M. G. M. Rowland, G. F. Mann, W. H. Lamb, and R. A. wHo/UNtCEFStatement. Geneva. Lewis. 1984. "Immunization of 4-6 Month Old Gambian Children with Yach, D., C. Metcalf, P. ILachman, G. Hussey, E. Subotsky, R. Blignaut, A. J. Edmonston-Zagreb Measles Vaccine." Lancet 2:834-37. Flisher, H. S. Schaaf, and N. Cameron. 1991. "Missed Opportunities for WHO (World Health Organization). 1988. Expanded Programme on Immuniza- Measles Immunization in Selected Western Cape Hospitals." South Afncan tion Mid-Level Managers Module: Measuring Vaccination Coverage. Geneva. MedicalJournal 79:437-39. 9 Tetanus Robert Steinglass, Logan Brenzel, and Allison Percy Tetanus is a completely preventable disease caused by contam- many instances, significant numbers of terminal patients leave ination of wounds with an anaerobic bacillus, Closn-idium the hospital against medical advice and are recorded as surviv- tetani. The organism is ubiquitous in soil and dust and has the ing patients. In one series of hospital patients, if all those who ability to form highly resistant spores. It exists harmlessly in left against medical advice are assumed to have subsequently the gut of many animals, including man. If the pathogen is died, which is most probable, the cFR increases from 50.6 introduced into necrosed tissues, it multiplies and produces a percent to 92.6 percent (Al-Mukhtar 1987). If only known powerful neurotoxin. Tetanus is an endemic environmental outcomes are included in the denominator, the cFR in the hazard, rather than a communicable disease, and consequently above study is 87 percent. does not spread in explosive epidemics (Cvjetanovic, Grab, Sharma and Sharma (1982) estimate from several hospital and Uemura 1978a). studies that the duration of clinical sickness is nineteen days Tetanus in newborns is caused by infection resulting from for newboms who recover and three days for those who die. unsterile methods of cutting the umbilical cord or dressing the Recovery in younger persons is usually complete. stump. The first sign of neonatal tetanus (NNT) is the baby's On the basis of available evidence, the CFR for non-neonatal inability to suck and swallow when a few days old. This tetanus (non-NNT) is estimated to be 40 to 50 percent, and the inability is due to rigidity of the lips and mouth (lockjaw), duration of clinical illness in adults is estimated to be 14 days which causes a characteristic ironic smile (risus sardonicus). for those who recover and 4.5 days for those who die (Sharma Rigidity quickly develops throughout the body, often accom- and Sharma 1982). Rey, Diop-Mar, and Robert (1981) esti- panied by generalized convulsions. Death, usually caused by mate the duration of hospital stay to be between 20 and 40 respiratory failure, occurs between six and ten days of life, two days. As with NNT, the CFR is largely due to treatment regimens, to three days after the onset of symptoms. availability of skilled care, and age distribution of infected In children and adults, tetanus infection follows puncture patients. The prognosis is poorer in the elderly, who represent wounds, cuts, and burns. Cases have been documented that a large proportion ofall tetanus patients in industrial countries. resulted from ear and skin infections, nonsterile injections and Some authorities report that survivors do not suffer from surgical procedures, ear-piercing, scarification rituals and tat- incapacitating sequelae (Rey, Diop-Mar, and Robert 1981), toos, circumcision, and animal bites or scratches. A relatively whereas others report persisting vertebral changes, ophthal- common cause of tetanus in adult women is postabortal or mological changes, limb deformity, and the need for convales- postpartum contamination of the uterus, which is associated cent physiotherapy (Senecal 1970; Veronesi and Focaccia 1981). with a high risk of fatality. Frequently, the portal of entry in Older patients are more likely to suffer sequelae because of non-neonatal tetanus cannot be determined by either the exacerbation of preexisting conditions. As tetanus does not con- patient or physician. fer immunity, reinfection is possible. Given the very high inci- The World Health Organization (WHO) estimates the case- dence and cFR of neonatal and non-neonatal tetanus in fatality rate (cm) of untreated neonatal tetanus to be 85 developing countries, the most effective strategy is to prevent the percent (Stanfield and Galazka 1984; Galazka and Stroh disease from occurring in the first place. 1986). Marked declines in CFRs in some hospitals are due to control of respiration by medicated relaxation and mechanical Public Health Significance of Tetanus ventilation (Simonsen, Bloch, and Heron 1987). Wide varia- tion in cFRs among treated NNT patients has been reported A sizeable body of literature exists on the public health signif- (Bytchenko 1966) and reflects treatment regimens, the vigi- icance of tetanus in both industrial and developing countries. lance and skill of nursing care, and methods of calculation. In Despite this, tetanus remains a neglected disease. 189 190 Robert Steinglass, Logan Brenzel, and Allison Percy Past and Future Trends in Incidence countries during the past thirty years have demonstrated dra- matically that, despite a contaminated environment, NNT can Well into the twentieth century, NNT continued to cause high be rapidly controlled and even eliminated by wide use of TT mortality in today's industrial countries. A review of the expe- immunization (Schofield, Tucker, and Westbrook 1961; New- rience in these countries is instructive. ell and others 1966; Berggren and Berggren 1971; Black, The decline of NNT in industrial countries to the point of Huber, and Curlin 1980). virtual elimination began before the widespread introduction Unless significant resources are allocated to its rapid reduc- of tetanus toxoid (TT) immunization for children or adults in tion, tetanus incidence will decline slowly in developing coun- the 1950s (Heath, Zusman, and Sherman 1964; Bytchenko tries. The pace ofdecline will be a function of epidemiological 1972; Christensen 1972a; Simonsen and others 1987). The risk factors, demographic trends, use of health care, socioeco- decline was due to improved socioeconomic conditions (for nomic change, political will, resource allocations, and techno- example, wearing shoes), sanitation and personal hygiene (for logical advances. The most important factors for the reduction example, cleaner maternal deliveries and immediate treatment of tetanus cases in developing countries will be increased of wounds), and advances in wound management (for exam- immunization coverage among children and adults with prep- ple, passive immunization with antitetanus serum [ATS] derived arations containing tetanus toxoid, urbanization and urban from horses). Increasing urbanization, decreasing proportion growth, and declining total fertility rates. The possible devel- of the population engaged in agriculture, mechanization of opment of a high-potency single-dose rr will also hasten the agriculture, increasing use of chemical rather than animal prevention of tetanus. fertilizers, and falling fertility were important factors contrib- To accelerate this decline in tetanus, the World Health uting to the decline. Hygiene associated with childbirth im- Assembly of WHO in 1989 set a goal for the Expanded Pro- proved (for example, hand washing and cord care) and the grammeon Immunization (EPI) ofglobal elimination ofNNT by proportion of deliveries conducted by trained health workers 1995. The goal itself has generated increased recognition of either at home (as in Holland) or in health facilities increased. the public health importance of NNT. The degree of political In the United States, NNT incidence declined from 64 per will engendered and allocation of required resources will de- 100,000 live births in 1900 to about 1 per 100,000 live births in the early 1960s (LaForce, Young, and Bennett 1969). In Japan, NNT mortality reached a similar level of 1 per 100,000 Figure 9-1. Measles, Neonatal Tetanus, and live births by 1968, despite a rate of nearly 40 per 100,000 live Pertussis Deaths Prevented and Occurring births only twenty years before. This rapid decline occurred in the absence of a TT immunization program, which was not Thousands of deaths introduced until the 1970s (Ebisawa 1967, 1972). The control 3,000 of NNT during the 1970s and 1980s in industrial countries occurred largely because of aseptic obstetric practices. The elimination of NNT owed much to the widespread immuniza- 2,500 - tion of children during the 1950s, which led to a cohort of women of childbearing age who could pass maternal antitoxin to their offspring (Christensen 1972a; Simonsen 1989). The 2,000 - fall in non-neonatal tetanus experienced in industrial coun- tries, particularly in children and young adults, was hastened by mass immunization of males and females with TT (Stanfield 1,500 - and Galazka 1984; Simonsen, Bloch, and Heron 1987). Although instructive, the experience of declining tetanus incidence and improved control in industrial countries is not 1,000 - germane to the situation prevailing in developing countries. With a few notable exceptions, documented levels of NNT today in developing areas are much higher than have ever been 500 - reported for today's industrial countries, and few cases are brought to health facilities. In many countries, prospects for rapid improvement in socioeconomic development or in the 0 proportion of births delivered hygienically are not encourag- Measles NNT Pertussis ing. Even where birth rates are falling, more babies are being born because of the increased numbers of women of childbear- ing age. Yet in the second half of this century, TrT-an effective m Prevented Occurring and affordable control measure-has become available, along with a structure of national immunization programs through Note: NNT deaths prevented = 325,000; NNT deaths occurring = 754,000. which it can be delivered. Pioneering field trials in developing Source: WHO/EPI 1989b. Tetanus 191 termine whether as many as 8 million babies and 2 million incidence per 1,000 live births dropped (Berggren 1974a). In children and adults (at current incidence rates) die from teta- an area of Indonesia with a history of regular reporting through nus during the 1990s. designated sentinel surveillance posts, NNT incidence was five times higher than indicated by the routine surveillance net- Tetanus: A Neglected Disease work operating in the same area (WHO 1986). Neonatal tetanus also is neglected for sociocultural reasons. With the exception of measles, NNT kills more children than The family of the baby with NNT is typically, but not exclu- any other vaccine-preventable disease. Called a disease of sively, poor and illiterate and does not view the disease as a "peculiar quietness" (Tateno, Suzuki, and Kitamoto 1961), biomedical entity amenable to modern medical treatment. tetanus may be the most underreported lethal infection in the From widely scattered parts of the world, it has been reported world. If tetanus had the potential to spread in sweeping that the supernatural nature of the signs of NNT suggests some epidemics or if the disease left lasting disability like polio, or if sort of spirit possession (Schofield, Tucker, and Westbrook it occurred primarily in adults like tuberculosis, NNT would 1961; Chen 1976; Bastien 1988; Blanchet 1989; Pillsbury probably have attracted the attention of public health author- 1989). In some areas of presumably lower incidence, NNT is not ities long ago. clearly distinguished from a larger syndrome affecting new- Instead, NNT kills its victims, who are generally born at borns (Bastien 1988; Nichter 1990). Cases of NNT are likely to home, before they are old enough to be registered or missed by be concealed in some cultures, where it is perceived as a punish- the health system. Routine disease surveillance systems in most ment from God (Solter, Hasibuan, and Yusuf 1986), possibly developing countries detect only a small fraction of cases, less because of parental wrongdoing (Bastien 1988). Traditional than 5 percent according to WHO (WHO/EmPI 1982). Lacking its prohibitions often preclude travel by the mother and newbom own three-digit code in the International Classification of Dis- in the intimate and secluded period immediately after delivery. eases (WHO 1977), NNT is frequently not reported separately from cases of non-NNT, despite epidemiologically significant Current Magnitude of Tetanus differences in risk factors and options for prevention. As a result, NNT incidence is often hidden within aggregate figures The magnitude and preventability of tetanus has been high- for "infections specific to the perinatal period" or cannot be lighted in several important papers in which the researchers disaggregated from tetanus in broader age groupings (for exam- attempted to define the public health agenda for the 1980s. In ple, from birth to age four). arguing for selective primary health care, Walsh and Warren Neonatal tetanus is there to be found, if one looks for it. A (1979) consider NNT to be in the highest priority group for single hospital sometimes admits more patients with NNT than disease control because of its high incidence, high mortality, are reported for the entire country (Stanfield and Galazka and cost-effective and feasible means of control. In struggling 1984; Betts 1989). In early global reviews of hospital data, NNT with problems ofscarcity and choice, Evans, Hall, and Warford was frequently found to be the leading cause of death in (1981) conclude that the investment policies of donor agen- pediatric wards (Bytchenko 1966; Miller 1972). just as re- cies should redirect resources to the areas of greatest need with ported smallpox incidence in Ethiopia, India, and elsewhere a package of matemal and child health services, including Tr rose precipitously in the face of stepped-up control and im- to pregnant women. Foege and Henderson (1986) argue that proved surveillance (Fenner and others 1988), so too the on the basis of cost, feasibility, safety, and effectiveness, the number of cases of NNT reported by a hospital in a rural area of highest priority should be given to immunization against teta- Haiti with a successful control program increased even as nus, measles, pertussis and diphtheria, and polio. Table 9-1. Estimated Worldwide Morbidity and Mortality from Non-neonatal Tetanus, 1980-1984 Population' Average morbidity Number of cases Mortalityb Number of deaths Countries (millons) (per 100,000) (thousands) (per 100,000) (thousands) Developing countries Asia' 1,510 15-30 226-528 6-14 90-211 Africa 490 15-35 73-172 6-14 29-69 America 370 3-8 11-30 1-3 3-11 Total 2,290 310-700 122-291 Industrial countries 1,160 0.15 2 0.6 1 Total' 3,550 312-702 123-292 a. Figures for 1982. This column does not total correctly in original citation. b. According to a mean cFR of 40 percent (25 percent to 40 percent in some urban hospitals,40 percent to 60 percent elsewhere in developing countries). In industrial countries, the CFR is nearly the same, despite intensive care, because most patients are elderly. c. Excluding China. Source: Rey and Tikhomirov 1989. 192 Robert Stemglass, Logan Brenzet, and Allison Percy Table 9-2. Age Distribution of Patients with Non-neonatal Tetanus and Case-Fatality Rate in Bombay and Dakar Bomba-a Dakarb 1954-68 1977-79 1960-67 1985-86 Age (years) Percent CFR Percent CFR Percent CFR Percent CFR I month-9 years 47.6 (32.8) 36.1 (6.1) 43.3 (24.9) 39.1 (17.6) 10-19 17.1 (33.7) 14.1 (15.6) 25.0 (24.9) 25.4 (22.5) 20-29 15.1 (61.5) 21.3 (23.5) 12.8 (37.7) 12.6 (45.5) 30-39 11.1 (45.9) 11.3 (38.9) 8.1 (39.2) 6.5 (41.2) 40-49 4.8 (46.8) 8.2 (23.1) 4.9 (36.4) 7.0 (37.8) 50-59 2.9 (49.3) 5.6 (66.7) n.a. n.a. 5.3 (46.4) 60+ 1.3 (53.3) 3.4 (63.6) 5.8 (63.0) 4.0 (71.4) Total 99.9 (39.1) 100.0 (21.6) 100.0 (30.0) 99.9 (29.1) n.a. Data not available. a. 1954-68 data from Patel and Mehta 1975; 1977-79 data from Vakil and Dalal 1975. b. 1960-67 data from Rev and others 1968; 1985-86 data from Sow 1989. Source. Rey and Tikhomirov 1989. The World Health Organization estimates that tetanus kills expected to occur in developing countries, as it did in indus- 754,000 newborns each year and that another 325,000 deaths trial countries (Rey, Guillaumont, and Majnoni d'Intignano are being prevented (WHo/EPI 1989b; see figure 9-1.) Neonatal 1979; Stanfield and Galazka 1984; Cottin 1987; Simonsen, tetanus typically accounts for one-fourth of infant mortality Bloch, and Heron 1987). As recently as the 1950s, NNT in the and half of neonatal mortality in unimmunized populations in United States accounted for 25 percent of all tetanus deaths developing countries (Galazka, Gasse, and Henderson 1989). (Heath, Zusman, and Sherman 1964). In Denmark, NNT ac- From community surveys, some 270,000 NNT deaths are counted for more than 50 percent of all tetanus deaths, but estimated annually in the Southeast Asia region of WHO and since 1970 no NNT deaths have been reported (Simonsen, another 200,000 in the African region (WHO/EP 1987a). Some Bloch, and Heron 1987). In Sri Lanka the proportion of all NNT 130,000 NNT deaths were estimated to occur in seven countries cases was halved after the first three years (1978-81) of the in the Eastern Mediterranean region of wHo in 1981, with establishment of the Expanded Programme on Immunization 111,000 in Pakistan alone (WHO 1982). Ninety thousand NNT (de Silva 1982). Increased incidence of adult tetanus may deaths were estimated for the American region in 1984 (Stan- occur in some developing countries that are experiencing a field and Galazka 1984). An additional 60,000 cases, including demographic transition, where high birth rates in the past 40,000 in China alone, are estimated to occur annually in the outpaced immunization coverage. Western Pacific region (F. Gasse, personal communication, Although the proportionof cases which occur at econom- July 3, 1990). These calculations are based on surveys, which ically productive ages in both developing and industrial will be discussed later. countries is likely to increase, the total number of cases at economically productive ages will decline in industrial coun- NON-NEONATAL TETANUS. The magnitude of non-NNT is poorly tries despite aging of the population (figure 9-2). Tetanus will defined and often overlooked in discussions of tetanus pre- increasingly become a disease of the elderly in industrial coun- vention. Stanfield and Galazka (1984) assume that 50 percent tries as a function of poor vaccination coverage and vanishing of all tetanus cases are non-neonatal. Rey and Tikhomirov immunity (WHO/EP 1981 a; WHO/EPI 1983b; Rosmini and others (1989) estimate that 300,000 to 700,000 cases of non-NNT, 1987; Simonsen, 1989; Sutter and others 1990). with 120,000 to 300,000 deaths, occurred yearly in the 1980s, excluding those in China (table 9-1). Industrial countries NMATERNAL MORTALITY DUE TO TETANUS. An important though account for less than 2 percent of the total number of non-NNT little recognized benefit from immunization of females with TT cases and deaths. is the prevention of tetanus mortality in adult women. Mortal- Using age distribution data from four studies conducted ity is prevented during the period of maternal risk (defined as from the 1950s to the 1980s in Bombay and Dakar, Rey and pregnancy or within six weeks of being pregnant) both from Tikhomirov (1989) show that 50 to 60 percent of non-NNT postpartum and postabortal tetanus, as well as from wounds cases and a higher proportion of deaths occur among econom- sustained at other times. Tetanus is caused by "inexpert at- ically productive individuals from ten to fifty-nine years old tempts to remove a retained placenta" at delivery and by (table 9-2). incomplete abortion (Schofield 1986). Following widespread immunization of infants and older Among 49 women fifteen to fifty years of age admitted with children with diphtheria-pertussis-tetanus ( DT) and diphthe- tetanus to one urban hospital in South Africa during a 7.5-year ria-tetanus (DT) vaccines, an epidemiological shift in the inci- period, tetanus was associated with pregnancy in 20 (40 per- dence of tetanus toward children and young adults can be cent) of these women (Bennett 1976). Seventeen (35 percent) Tetanus 193 of the total cases were postabortal, two were postpartum, and ducted widely since the late 1970s to determine mortality from one occurred during pregnancy. In another review, the genital NNT and to elucidate its epidemiological features (Galazka and tract was the portal of entry in 19 percent of more than 500 Stroh 1986). The surveys generally use a short recall period of women with tetanus (Adeuja and Osuntokun 1971). In the four to thirteen months and rely upon a "verbal autopsy" United States, LaForce, Young, and Bennett (1969) reported method based on the classic symptoms of NNT (ability to suck that 6 of 507 tetanus cases were associated with abortion (4) during first two days of life, followed by cessation of sucking, or parturition (2). In the state of Rio de Janeiro from 1966 to stiffness, spasms, and death within the first month of life). As 1968, postabortal tetanus constituted 7 (4.8 percent) of the of 1989, forty-twocountries had conducted more than seventy- 146 cases with known portal of entry (Ecuador: Ministerio de five of these surveys (WHO 1982; Gasse 1990; see table 9-3). Salud P(iblica and UNICEF 1987). Mortality from NNT ranges from as low as 0 to 2 per 1,000 Citing seven studies conducted in India and one each in live births (in Tanzania, Congo, Lesotho, Jordon, Tunisia, and Japan, Singapore, and Viet Nam, all published from 1960 to Sri Lanka) to 30 to 67 per 1,000 live births (in Pakistan, 1962, Bytchenko (1966) notes that postpartum and postabor- Bangladesh, and India). Among surveys which detected NNT tal tetanus (3 to 113 cases, median 17) accounted for 3 to 22 and reported the proportion of neonatal deaths due to tetanus, percent (median 8.2 percent) of all cases of tetanus, with cFRs 6 percent (in urban areas in West Bengal, India) to 72 percent ranging from 64 to 72 percent. A review of 981 tetanus patients (in rural areas in Uttar Pradesh, India) of neonatal mortality admitted to one hospital in New Delhi from 1963 to 1965 was attributable to tetanus (WHO 1982). found that postabortal and postpartum tetanus caused 47 per- The surveys have shown that whenever neonatal mortality cent (71) of the 150 cases occurring among women of fifteen exceeds 30 per 1,000 live births, tetanus is invariably a substan- to fifty years of age, 25 percent of the 280 female non-neonatal tial contributor (REACH 1989; see figure 9-3). In some devel- cases, and 7 percent of all tetanus cases (Suri 1967). oping countries where measles immunization coverage has In two prospective investigations of matemal mortality in a rapidly increased, NNT could soon overtake measles as the rural Bangladesh populationhaving a vital registration system, leading cause of mortality among the vaccine-preventable Chen and others (1974) found that maternal deaths (defined diseases (WHO/EPI 1987a). These surveys have succeeded in as occurring during pregnancy or within ninety days of its alerting many national decisionmakers about the magnitude termination) accounted for 27 to 30 percent of all adult female of NNT as a public health problem. On the basis of the surveys, deaths. Where cause of death was reported, 3 (7.3 percent) of WHO estimates that only 2 percent and 5 percent of NNT cases the 41 maternal deaths were due to postpartum tetanus. Thus, have been reported in the Eastem Mediterranean and South- 2 percent of all adult female deaths were due solely to post- east Asian regions of WHO, respectively (WHo/EPI 1982). partum tetanus. Postabortal tetanus deaths and tetanus not related to pregnancy would further increase the percentage of Figure 9-2. Reported Tetanus Cases in Poland, adult female deaths attributable to tetanus. The tetanus-attrib- by Age, 1965, 1975, 1985 utable maternal mortality rate was 56 per 100,000 live births. Rosenfield and Maine (1985) cite a 1979 WHo estimate that Cases 500,000 females in developing countries die annually from 2 complications of pregnancy, abortion attempts, and child- 50 birth. Maine and others (1987) construct a model for a hypo- thetical population of 1 million with a crude birth rate of 46 200 - per 1,000 and a maternal mortality rate of 800 per 100,000 births, but they assume a low proportion (2 percent) of mater- nal mortality to be due to tetanus. The tetanus-attributable 150 - maternal mortality rate in this case would be 13 per 100,000 150 live births. Tetanus represents an important cause of preventable ma- 100 - ternal mortality, although it is sometimes only mentioned in passing in references on safe motherhood (Herz and Measham 1987; Royston and Armstrong 1989). A global review of 50 - existing data to determine the magnitude of matemal mortal- ity due to tetanus has recently appeared (Fauveau, Mamdani Steinglass and Koblinsky 1993). From 15,000 to 30,000 deaths 0 due to postpartum and postabortal tetanus are estimated to 1965 1975 1985 occur annually. NEONATAL TETANUS. Much more is known about NNT than _ 0-4 _ 5-14 15-59 _ > 60 about non-NNT or maternal tetanus. Retrospective community surveys using a cluster sampling methodology have been con- Source: Galazka and Kardymowicz 1989b. 194 Robert Steinglass, Logan Brenzel, and Allison Percy Table 9-3. Estimated Neonatal Mortality and Neonatal Tetanus Mortality Rates Based on Special Community Surveys, 1978-1989 (per 1,000 live births) MortaLty rates Number of live Percent of neonatal WHO region Country Year births surveyed Neonatal NNT deaths due to tetanus Africa The Gambia 1980 4,976 - 11 Cameroon 1982 2,102 - 7 C6te d'Ivoire 1982 2,324 34 18 51 Malawi 1982 2,081 29 12 41 Ethiopia 1983 2,010 8 5 53 Zimbabwe 1983 4,103 10 4 39 Zaire 1983 4,106 - 9 Senegal 1983-86 4,164 51 16 31 Cameroon 1984 2,118 - 8 Uganda 1984 525 38 15 40 Togo 1984 4,966 11 6 52 Burundi 1984 3,099 - 8 Kenya 1984-85 6,566 16 11 67 Lesotho 1986 - - 4 Zambia 1986 3,741 14 4 30 Tanzania (Zanzibar) 1988 2,269 9 2 25 Congo 1988 3,524 15 2 15 Ghana 1989 2,694 26 7 29 Kenya 1989 2,556 21 3 15 Lesotho 1989 2,467 4 0 0 Madagascar Urban 1989 3,133 7 0.1 2 Rural 1989 2,772 2 0.8 38 Niger 1989 2,550 26 9 33 Tanzania Kagera 1989 2,118 - 3 Morogoro 1989 2,129 - 3 Eastern Dem. Yemen 1981 6,224 19 4 20 Mediterranean Egypt (urban) 1981 - - 3 Pakistan 1981 13,858 52 31 60 Somalia 1981 5,781 91 21 23 Sudan 1981 9,632 29 9 32 Syrian Arab Republic 1981 6,762 - 5 Yemen Arab Republic 1981 5,191 31 3 8 Jordan 1983 2,850 7 2 13 Pakistan 1984 9,925 - 28 Iran, Islamic Rep. of 1985 144,000 21 5 24 Egypt 1986 8,286 12 7 58 Pakistan 1987 5,859 14 4 29 Tunisia 1988 9,478 15 2 9 Southeast Asia Bangladesh 1978 2,432 48 27 56 Indonesia 1979 1,570 49 23 46 Indonesia 1980 3,933 - 12 Nepal 1980 3,346 37 15 39 Thailand 1980 13,659 21 5 23 India Rural 1980-81 23,482 19-93 5-67 16-72 Urban 1980-81 25,843 5-26 0-15 0-59 Bhutan 1982 952 19 13 67 Indonesia Rural 1982 4,971 21 11 51 Urban 1982 2,310 17 7 40 Nepal 1982 1,997 44 24 55 Indonesia 1983 4,779 - 17 Tetanus 195 Mortality rates Number of live Percent of neonatal WHO region Counr Year births surveyed Neonatal NNT deaths due to tetanus Indonesia 1984 4,836 21 35 Southeast Asia Indonesia 1984 4,769 - 9 (continued) Sri Lanka 1984 2,841 15 1 7 Burma 1985 6,000 18 6 33 Bangladesh 1986 2,077 82 41 50 India 1986 2,386 37 5 14 Indonesia 1986 4,707 - 3 - Nepal 1988 728 19 4 23 Westem Pacific Philippines 1982 8,754 13 6 48 Viet Nam 1985 8,270 12 2 16 Lao PDR 1985 4,996 16 4 25 Viet Nam 1989 9,199 8 3 40 Source: Gasse 1990. Although the NNT mortality figures are so high, some registration of all women, establishment and updating of a epidemiologists consider them to be lower than the true pregnancy registry during regular home visits by survey staff, rates, as early neonatal deaths are often missed on retrospec- and recording of births and deaths. Identifying pregnan- tive surveys (Foster 1984). Mortality rates for NNT are un- cies and tracking outcomes eliminates many of the poten- derestimated for other reasons as well. In C6te d'lvoire, the tial biases of the retrospective survey design concerning longer the recall period (for example, more than seven recall, concealment, and uncertain recording of either months), the more likely mothers forget or are unwilling to live births or deaths. report NNT in relation to other causes of neonatal mortality Longitudinal surveillance by the Intemational Center for (Sokal and others 1988). Diarrheal Disease Research (ICDDR) in Matlab, Bangladesh, Cultural factors also influence underreporting. For exam- from 1975 to 1977 among a rural population of 260,000 estab- ple, female neonatal deaths may be undercounted in some lished a NNT mortality rate of 37.4 per 1,000 live births with cultures (Galazka and Cook 1985). In Indonesia, Amold, tetanus responsible for 26 percent of all infant deaths (Chen, Soewarso, and Karyadi (1986) found a reluctance to discuss Rahman, and Sarder 1980). A prospective survey conducted infant deaths. In Senegal, self-reports by families resulted in twenty of twenty-six NNT deaths being ascribed to fevers (one), prematurity or low birth weight (three), other causes Figure 9-3. Total Neonatal Mortality and Neonatal (five), and unknown causes (eleven) because of variability Tetanus Mortality Rates in diagnosis through verbal autopsy (Garenne and Fontaine 1986). Neonatal tetanus mortality rate Biases also may lead to overestimation of NNT mortality. (per 1,000 live births) Gray, Smith, and Barss (1990) question whether the differ- ential diagnoses used in community surveys may lack spec- 50 - ificity, so that nontetanus deaths are included in estimations. / Furthermore, surveys are conducted in expected high-inci- dence areas, because the intent is to publicize the magnitude _ - of the NNT problem; generalization of survey results is then likely to overestimate NNT mortality for the country as a 20 - whole. No global review of results of prospective as opposed to 10 _ retrospective community NNT studies in the same geograph- ical area and time period has been published. In general, o V l l i i i i i very few prospective studies have been conducted, but these 10 20 30 40 50 60 70 80 90 100 too demonstrate the enormity of NNT mortality and approx- Neonatal tetanus mortality rate imate the rates found on retrospective surveys. In one of the (per 1,000 live births) few community surveys for NNT ever conducted in Latin America, Newell and others (1966) demonstrated that tet- anus killed 110 per 1,000 newboms. Like other prospective Note: If the neonatal mortality rate exceeds 30 per 1,000, neonatal tetanus is invariably a substantial contrNbutor. NN'T studies, this double-blind controlled trial relied on Source: FREACH 1989; data from WHO/EPI 1987a. 196 Robert Steinglass, Logan Brenzel, and Allison Percy by ICDDR in Teknaf, Bangladesh, determined a NNT mortality compared with that in delivery in a health facility was still rate of 27.4 per 1,000 live births with tetanus responsible for three to one in Sri Lanka (Foster 1984). 30.8 percent of neonatal and 21.3 percent of infant deaths In one Indonesian survey, a NNT mortality rate of 23 per (Islam and others 1982). These two prospective survey results 1,000 live births was recorded for babies whose mothers had are within the range reported in the retrospective studies in no prenatal contact with health facilities, as opposed to only Bangladesh (table 9-3). 4 per 1,000 for those whose mothers had at least two contacts In the Indonesian province of West Java, a prospective (WHO/EPI 1983c). In another study in Indonesia, however, NNT study (Budiarso 1984) gave a NNT mortality rate of 14.7 per mortality rates were similar regardless of the number of prena- 1,000 live births, which also falls within the range of retro- tal contacts, because TT was not being systematically given spective rates already reported. Finally, in a prospective (Solter, Hasibuan, and Yusuf 1986). In a prospective study in case-control study in a rural area of thirty villages in Sene- Senegal, it was found that washing hands with soap on the part gal, with a total population of 24,000, all births and deaths of the birth attendant had a significant effect on the occur- were systematically recorded for 43 months; the researchers rence of NNT (odds ratio 5.19, p = .0001) when other factors found a NNT mortality rate of 15.9 per 1,000 live births with were held constant (Leroy and Garenne 1991). tetanus accounting for 31 percent of all neonatal deaths (Leroy and Garenne 1991). Prevention of Tetanus This collective body of research has provided important insights into the epidemiological characteristics of NNT. Data Tetanus can be prevented by a variety of approaches at a cost are available on seasonality, age at onset of symptoms and which is affordable by most countries. death, age and parity of mothers, and circumstances of delivery and cord cutting. The preponderance of male over female Prevention Strategies deaths has been observed in many, but not all, studies and has been ascribed to real differences in risk of dying; greater like- Tetanus can be prevented by immunization, clean cutting and lihood of males being brought to the hospital, where their dressing of the umbilical cord, and hygienic wound manage- deaths would be recorded; differences in cord cutting and ment. In this section we will concentrate on prevention of NNT handling of males and females; and selectivity in recall (Stan- through immunization, because data from developing coun- field and Galazka 1984). As analyzed by week-specific mortal- tries on the effect of interventions designed to ensure aseptic ity ratios by sex, the practice of male circumcision surprisingly cord care are limited (Ross 1986a, 1986b) and information on does not appear to explain the larger number of male than associated costs is practically nonexistent. Likewise, data on female NNT deaths. the effect and cost of wound management are rare in develop- In cattle- and horse-raising areas of Punjab, Pakistan, NNT ing countries. mortality was significantly higher than in nearby farming and Immunization against tetanus is achieved by vaccinating urban areas, presumably because of a greater risk of exposure different target groups with vaccines such as DPT, DT, TT and Td (Suleiman 1982). In Uttar Pradesh, India, exposure-related (tetanus-diphtheria with a reduced component of diphtheria variables (for example, a previous NNT death in the family, antigen)-all of which contain tetanus toxoid. Tetanus toxoid presence of large animals in the home, assistance by an un- and Td are suitable for adults, whereas DPT vaccine is given to trained birth attendant) were better predictors of NNT than socioeconomic variables, such as education, income, land Table 9-4. TT Immunization Schedule for Women ownership, and caste (Smucker and others 1980). The signif- Percent Duraton of icance of epidemiological factors for NNT, as opposed to the role Dose Time to immunize protected protection of socioeconomic factors, was noted also in Senegal (Leroy and Garenne 1991). rr-1 At first contact or as early Nil None Garenne 1991). ~~~~~~~~~~~~~~as possible in pregnancy Surveys in urban areas have shown consistently lower NNT mortality rates than in neighboring rural areas in Egypt, Iran, rr-2 At least four weeks after 80 Three years and India (Galazka and Stroh 1986; WHO/EPI 1987b, 1987c). rT+I Nevertheless, urban rates are high in some cities, such as TT-3 At least six months after 95 Five years Jakarta, Indonesia, which has a NNT mortality rate of 6.9 per TT+2 or during subsequent 1,000 live births (Arnold, Soewarso, and Karyadi 1986). pregnancy In most studies, NNT wasmore likely to occur if the child was TT-4 At least one year after TT+3 99 Ten years delivered at home or by an untrained attendant. Hospital or during subsequent delivery, however, does not guarantee protection from NNT, pregnancy because infection may still occur as a result of unclean delivery TT-5 At least one year after TT+4 99 Throughout or of unhygienic cord dressing after discharge. In Sri Lanka, or during subsequent childbearing from 1975 to 1980, nearly half ofNNT deaths (206 of 423) and pregnancy years 75 percent of all births occurred in health facilities (de Silva a. Original document states "for life." 1982). Nevertheless, the relative risk of NNT in home delivery Source: WHO/EPI Programme on Immunization, 1988b. Tetanus 197 children less than five years old and preferably during infancy. In Haiti, mass immunization of the entire population of a The DT vaccine is used for young children unable to receive rural area eliminated NNT and reduced non-NNT to a negligible DPT and is mainly administered in schools. The schedule rec- level (Berggren 1974a). A two-round mass TT campaign in ommended by the World Health Organization requires five 1985 in Pidie District, Indonesia, achieved 84 percent im- doses of rT for protection throughout the childbearing years munization coverage of all women of childbearing age with two (w[io/Epi 1988b; see table 9-4.) doses of TT. Pre- and postcampaign NNT mortality surveys Protective levels of antibody in the woman ensure protec- indicated an 85 percent reduction in NNT mortality. This tion for the newborn (as well as for the mother herself), since reduction was likely due to the mass Tr campaign, because antibody crosses the placenta from mother to baby. Studies on neonatal mortality attributable to causes other than tetanus the immunological response to TT have been reviewed by Rey remained unchanged (WHO/EPI 1988a). Despite some disrup- (1982) and Galazka (1982, 1983). The standard series of three tion to routine programs, the Pidie campaign covered a high DrT injections given at monthly intervals during infancy proportion of the unimmunized backlog, resulting in a dra- counts as the first two of the five TT injections required for matic reduction in NNT nmortality. protection throughout the childbearing years. Immunization was shown to be dramatically effective in Tetanus toxoid costs about $0.02 perdose in multidose vials, reducing NNT in Colombia (Newell and others 1966), New can withstand temperatures of 37C for at least six weeks Guinea (Schofield, Tucker, and Westbrook 1961), Bangladesh (WHO/EPI 1990), has more than 95 percent efficacy when used (Black, Huber, and Curlin 1980; Rahman, Chen Chakraborty, according to the correct schedule, and is extremely safe. Nev- Yunus, Chowdhury, Sarder, Bhatia, and Curlin 1982), and ertheless, less than half the babies in developing countries have many other countries. In Bangladesh, varying NNT mortality immunity at birth against NNT. between the fourth and fourteenth days explained most of the Reactions to TT are minor and local, usually lasting less than difference in neonatal and infant mortality rates between the one day. Severe systemic reactions are rare, occurring in 1 per Maternal and Child Health (MCH) and Family Planning inter- 50,000 to 250,000 injections (Christensen 1972b; White and vention area and the control area in a population of 260,000 others 1973). In the German Democratic Republic, severe (Bhatia 1989). residual damage was virtually unknown (WHO/EPI 1983a). Tet- Mosley (1989) has argued that, unique among vaccines, rT anus toxoid can be given at any stage of pregnancy without is highly effective because NNT incidence is high and the increased risk of abortion or congenital abnormality (Heino- disease has high fatality among otherwise healthy persons. For nen, Shapiro, and Slone 1977). Contraindications to TT im- that reason, a disease-specific intervention against tetanus is munization are virtually nonexistent (Rey and Tikhomirov able to have a large demographic effect. Unlike measles or 1989). pertussis-related deaths, which come at the end of a cycle of Researchers have documented the effect of rT immunization synergistic insults (infection, growth retardation, and reduced on reducing NNT mortality and lowering overall neonatal mor- resistance), NNT has a discrete cause and a specific interven- tality in developing countries using a variety of control strate- tion-TT. Consequently, the so-called replacement mortality gies. Immunization of pregnant women in Sri Lanka and phenomenon probably does not occur in the case of NNT. Burma resulted in rapid declines in NNT (Stroh and others Henry, Briend, and Fauveau (1990) state that given the con- 1987; Galazka, Gasse, and Henderson 1989). In Sri Lanka after the introduction of EPI in 1978, NNT dropped from 2.16 to 0.06 per 1,000 live births by 1986. In Burma, a community survey Table 9-5. Effect of Three Different Interventions found that the NNT mortality rate (3 per 1,000 live births) in on Mortality from Neonatal Tetanus, Bunna, 1985 EPI operational areas was only one-third the rate in other areas Intervention (Stroh and others 1987). The effect of three interventions in Immunization Burma was calculated to determine their proportional contri- with two doses of Hospital Trained birth bution to the reduction in neonatal tetanus mortality (Stroh Characteristics tetanus toxoid delivery attendant and others 1987; see table 9-5). Hospital delivery, although Efficacy' 0.91 0.85 0.33 highly efficacious (85 percent),contributedtoonly 17 percent Coverage 0.44 0.14 0.51 of the reduction in NNT mortality because coverage with this Efficacy x coverage 0.40 0.12 0.17 intervention was only 14 percent. The programmatic efficacy, Contribution to which is a product of the efficacy and coverage of the inter- reduced neonatal vention, was slightly higher for deliveries by traditional birth mortality (percent)b 58 17 25 attendants (TBAs) and much higher when two doses of TT a. Defined as: immunization were given to the pregnant woman. attack rate without intervention - attack rate with intervention ,,, ,, . ~~~~~~~~attack rate without intervention Mass immunization of 95 percent of women of childbearing h. Defined as: age with one dose ofTT from 1978 to 1979 in Maputo, Mozam- (efficacy i x coverage l x 10a bique, followed by routine immunization of pregnant women, where i i nterention. sum of (efficacy i x coverage i) resulted in an eightfold drop in reported NNT cases in one Source: Galazka, Gasse, and Henderson 1989; based on data in Stroh and hospital (Cliff 1985; WHO/EPI 1988b; Cutts and others 1990). others 1987. 198 Robert Steinglass, Logan Brenzel, and Allison Percy straints in Bangladesh on implementing the full EPI, TT (along early school grades before attrition. School enrollment levels with measles vaccination) is the most cost-effective immuniza- of boys and girls are greater than 95 percent in some of the tion strategy for child survival. same districts where Tr coverage among pregnant women is Active and permanent immunization of the entire popula- lowest. Approximately 90 percent of women in Kenya deliver tion and of successive cohorts will be required first to control their babies at home, despite high levels of prenatal care, so a tetanus and then, because of continuous environmental risks strategy aimed at schoolchildren will help solve the problem of contamination, to sustain its elimination as a public health of NNT in the medium term while having a marked, immediate problem. Each individual will require five doses of the current effect on non-NNT. tetanus toxoid preparation at appropriate intervals for full Through high coverage with DPT in infancy and DT in school, lifetime protection. Rey and Tikhomirov (1989) propose a some of the Gulf States in the Middle East have virtually three-stage approach for tetanus control: universal immuniza- eliminated NNT-despite a TT coverage rate in pregnant tion of infants, children, and women of childbearing age, women of less than 20 percent. Most girls attend school and including pregnant women; extension of immunization to women deliver their babies in hygienic conditions. Conse- schoolchildren and high-risk, easy-to-reach adults (for exam- quently, no systematic attempt to immunize all women to ple, military recruits), as well as more systematic prophylaxis eliminate NNT is indicated. In Denmark, a vaccination program of wounds; and extension of immunization to all other adults, consisting of three high-potency DT injections in infancy and including such neglected groups as the elderly and immigrants. a single revaccination five years later resulted in continuous The relative emphasis and timing of each of these control protection to about the age of twenty-five (Simonsen and stages is a decision best made at national and subnational levels others 1987). based on disease epidemiology, organization of health service A strategy which relies exclusively on the identification and delivery, operational and behavioral considerations, availabil- immunization of women during pregnancy is unlikely to suc- ity of resources, and the tradeoffs between costs and benefits ceed in many areas for operational and cultural reasons. Use of of early and late control. The appropriate choice and degree of prenatal services at fixed facilities is low or frequently occurs implementation of strategies will depend on whether the goal very late in pregnancy, leaving insufficient time to administer is total elimination of all tetanus, the total elimination of NNT, two doses to the previously unimmunized woman. With peri- or the elimination (that is, control) of both as public health odic outreach or mobile strategies, trying to identify only problems. pregnant women is like trying to hit a moving target when the The advantages and disadvantages of using various im- marksman, or health worker, is also on the move. A campaign munization strategies and target groups can be delineated in Bangladesh in which health workers went door-to-door to (WHO/EPI 1986) on the basis of discussions by Rey (1982), identify pregnant women was not very successful, given shy- Cvjetanovic and others (1972), and Schofield (1986). (See ness to declare pregnancy, outright resistance to vaccination table 9-6.) There is a long window of opportunity during which during pregnancy, and the health workers' failure to refer the it is possible to immunize women to prevent NNT. Ideally, women for vaccination sufficiently early in pregnancy (Rah- women entering their childbearing years already should have man, Chen, Chakraborty, Yunus, Faruque, and Chowdhury received five doses of tetanus toxoid, which can be in the form 1982). of properly spaced doses of DPT or DT (in childhood) and TT. Historically, the exclusive focus on pregnant women as a The prevailing belief in many countries that two doses of -T target group for TT has been well intended but operationally are sufficient must be changed (Steinglass 1989). The earlier impractical in many developing countries. Administration of the protection, the greater the reduction of non-NNT, as well. rT to individuals in this high-risk group is known to have an This is important, given that the highest age-specific incidence immediate effect on protecting the newbom. But unless the of tetanus after the neonatal period in developing countries is health services are well developed and appropriately used, a among children. target group focusing on pregnant women exclusively will have There is no global blueprint for NNT control. Strategies need low programmatic efficacy and will not achieve a rapid reduc- to be determined locally and may differ from one area to tion of NNT or non-NNT. Thus a population-wide strategy is another within the same country. In Kilifi District, Kenya, a required because susceptibility to tetanus is general. NNT mortality survey conducted in 1989 by the Resources for For this reason, WHO recommends continuous immuni- Child Health (REACH) project found low levels ofNNT mortality zation of women of childbearing age, including pregnant (3 per 1,000 live births) as a result of high TT coverage through women. Every contact with the health services is an oppor- well-attended prenatal care services (Bjerregaard, Steinglass, tunity to screen a woman's TT status and provide immuniza- Mutie, Kimani, Mjomba, Orinda 1993). Targeting Tr to preg- tion. This strategy is less immediate in protecting individual nant women was an appropriate strategy in this district because births than one that focuses on women already pregnant, prenatal care coverage is high, although mothers with one but its effect on the population will be more rapid. For this child were significantly less well protected with Tr than were strategy to work, a change in attitude of health workers, multiparous women. particularly curative staff, may be needed. Because this Elsewhere in Kenya, however, where prenatal care services target group is less specifically at risk than women already are unavailable ot not used, girls need immunization during pregnant, more doses of TT will be needed per NNT case Tetanus 199 Table 9-6. Advantages and Disadvantages of Different Immunization Strategies in the Prevention of Neonatal Tetanus Approach Advantages Disadvantages When to use Immunization of pregnant Few additional resources needed Hesitation about injections When over 80 percent of women attending antenatal Potentially rapid impact on during pregnancy pregnant women attend at least services disease incidence Women at highest risk rarely twice in antenatal period come for antenatal care As part of overall effort to Only very short periods immunize women to immunize women and to maintain immune status Immunization of women of child- Any contact of women with Cooperation of health staff Preferred when coverage for bearing age through regular health worker can be used needed antenaral care is less than health services Better chance of reaching high- More complex logistics complete and there is reason- risk women (who may not Accessibility of health services able degree of access to general come for preventive care, but may be limited health services would come for curative care May need to be supplemented for their child) with (limited) mass campaigns Immunization of women coming Few additional resources needed Women are not reached for first Should be part of any approach with children to immunization Women with children are likely pregnancy If used as only approach, needs session to become pregnant again Coverage cannot exceed periodic supplementation with maximum coverage of children mass campaigns Immunization of women coming Few additional resources needed Accessibility may be limited Should be part of any approach with or without children to Women reached for first and may eliminate need for immunization session pregnancy mass campaign if coverage is high Special outreach clinics Increases accessibility Needs organization and some In places with regular, well- (markets, meetings) considerably extra resources attended markets or other spe- cial events and limited access to regular health services Immunization of school- Few additional resources needed Impact on disease incidence Wherever a school health children Can be incorporated into delayed (ten to twenty years) program can be activated ongoing school health programs High-risk groups have low school without distracting resources School immunization programs attendance from MCH care may provide good stimulus for No school health program in improving health education on most rural areas immunization Mass campaign Rapid impact Resource intensive Wherever incidence is 10 per High visibility has good Might distract resources from 1,000 live births or more promotional value development of regular MCH When special high-risk areas or Men as well as women can be care groups are not reached included May need repetition otherwise As part of any accelerated immunization activity Source: WHO/EPI 1986. averted because many doses will be given to older, less fertile Economic and Financial Considerations of Prevention age groups. The World Health Organization recommends the use of long-lasting cards for appropriate screening, im- Tetanus has important economic consequences for the devel- munization, and documentation of protection. oping world. Mortality rates for neonatal tetanus are generally A promising strategy being introduced by the Pan-American highest in the poorest countries, which have inadequate pre- Health Organization in Latin America makes use of existing ventive services. incidence data from routine reporting systems to identify areas For infected babies in the poorer countries, tetanus is almost at higher-than-average risk for NNT. This strategy permits always fatal. Children and adults not adequately protected by health staff to target immunization efforts (Cvjetanovic 1972; TT immunization experience a continuous risk of tetanus from de Quadros 1990). Another novel strategy is being tried in wounds throughout their productive years. Deaths from teta- Indonesia, where prospective brides are required to show proof nus, in 40 percent of adult cases and 85 percent of neonatal of rr immunization for marriage registration (Lanasari and cases, affect the productive capacity of the population and Rosenberg 1989). represent an economic loss to society. Further, hospital treat- 200 Robert Steinglass, Logan Brenzel, and Allison Percy ment of tetanus is expensive for families and society. In coun- vaccinators' salaries but did not include costs at the national tries with limited public health resources, provision of ade- level or a portion of the routine salaries of health officials quate treatment for tetanus constitutes a high opportunity cost (UNICEF/Jakarta 1985). Kielmann and Vohra (1977) predicted compared with health services for other diseases. Given the the cost in Punjab, India, of importing, storing, and transport- high cost and poor outcome of treatment, and the fact that ing vaccine, as well as managing and administering a hypothet- most tetanus patients are never brought to health facilities in ical program. the first place, the most cost-effective strategy is to prevent this Among the studies in which economic costs were calcu- common disease. lated, the identification and calculation of program costs The cost to society of tetanus includes economic loss due to varied. In several studies the researchers adapted methodol- death and disability, the cost of treating individuals, and the ogies developed for costing EPI (Brenzel 1987; Brenzel and cost of preventing disease. Although there is a growing body others 1987; de Champeaux and Martin 1989; Narcisse of empirical and theoretical evidence on the cost of preventing 1989; Phonboon and others 1989; Brenzel and others 1990; tetanus, little work has been done on the loss of productivity Berman and others 1991). The authors of one study mea- because of the disease. In this section we review and discuss sured the incremental cost of adding different vaccines, like cost-benefit and cost-effectiveness analyses of the prevention DPT, to EPI in Indonesia on the basis of additional inputs and of NNT and non-NNT through immunization. Other methods of resources required for each (Barnum, Tarantola, and Setiady preventingtetanus-suchastrainingTBAstoprovide improved 1980). Berman and colleagues (1991) were the only re- obstetrical care-have not been included in the following cost searchers who distinctly evaluated the economic cost of a analyses because there is a near total absence of data on costs tetanus immunization campaign. The diversity of underly- and the effect of alternative interventions. ing assumptions, not always explicit in each study, makes Fifteen studies and four simulation models are reviewed comparing results risky. (see appendix tables 9A-1-9A-4). The studies, which in- Cost-effectiveness studies differed in how shared health clude six from Africa, six from Asia, and three from the resources were allocated to tetanus prevention. For instance, Latin American and Caribbean region, were conducted by the cost of the tetanus component of EPI was estimated to be a variety of researchers during a twenty-year period, from 37.5 percent of the total cost in the Gambia (Robertson and 1970 to 1990. Examination of TT costs was not the purpose others 1985). This proportion was based on the number of of eight of the studies. In these instances, available cost and contacts for tetanus toxoid (three) compared with total con- coverage data have been used to estimate the cost-effective- tacts for all doses for full immunization (eight). In many ness of immunization strategies. studies, the cost of immunizing women with TT was based on the proportion of annual doses administered to women com- COST-EFFECTIVENESS STUDIES. The resultsfromcost-effectiveness pared with total annual doses given by EPI (Brenzel 1987; studies of immunization against tetanus are difficult to com- Brenzel and others 1987; Shepard and others 1987; de Cham- pare because of the variability in methods used to determine peaux and Martin 1989; Narcisse 1989; Phonboon and others program costs and outcomes (see appendlix tables 9A-1 and 1989; Brenzel and others 1990). 9A-2). Some researchers use actual cost and incidence data, The time frame used in the cost studies affects the results. whereas others rely on estimates based on various assumptions. Although annual costs of a tetanus control program were The methods used to estimate program costs vary in four calculated in most of the cost-effectiveness studies included in principal ways: (a) type of costs measured, (b) method of this review, there was no uniform time period across all studies. allocating shared resources, (c) time frame of analysis, and (d) The researchers in several studies evaluated the cost of an the strategy and scale of tetanus control. immunization program over a period of several years but did Researchers measured either the expenditure for the not discount costs to a present value estimate (Berggren 1974a; immunization program (Berggren 1974a; Kielmann and Vohra Barnum, Tarantola, and Setiady 1980). Others measured costs 1977; Rey, Guillaumont, and Majnoni d'Intignano 1970; for less than one year, for mass campaigns, for example, which UNICEF/Jakarta 1985) or the economic costs (Bamum 1980; last for a period of months (UNICEF/Jakarta 1985; Narcisse 1989; WHO/EPI 1981b; Robertson and others 1985; Brenzel 1987; Berman and others 1991). Further, the point at which a Brenzel and others 1987; Shepard and others 1987; de Cham- cost-effectiveness study is conducted in the life of an im- peaux and Martin, 1989; Narcisse 1989; Phonboon and others munization program affects the generalizability of the results. 1989; Brenzel and others 1990; Berman and others 1991).1 Estimates are likely to be located on different average cost Rey, Guillaumont, and Majnoni d'lntignano (1979) include curves at different times during the immunization program, only vaccine costs in the cost calculations for Dakar, Senegal. making comparisons of efficiency suspect. In Deschapelles, Haiti, Berggren (1974a) estimated the value The studies focused on routine immunization of pregnant of the time spent by health workers to administer tetanus women, or mass immunization of pregnant women or women toxoid vaccine during visits to marketplaces, though he used of childbearing age. The target population, delivery strategy the total expenditure for the five-year program for the cost- used, and the scale of the intervention differed. Most studies benefit calculation of the program. The costs of the Indonesian focused on immunization of women with rr. The required campaign in Central Lombok reflected UNICEF expenditure and number of doses, and therefore contacts, per woman for full Tetanus 201 protection was two in most studies, although it ranged from results were used even though they represented a wider or one (Kielmann and Vohra 1977) to three doses (Berggren different geographic area within the country. 1974a; Robertson and others 1985). Pregnant women were the The assumed duration of protection from TT also differed target group in seven of the studies, whereas women of repro- among studies. Kielmann and Vohra (1977) assumed that a ductive age were the focus in another seven studies. The cost single, high-potency dose of TT gave lifelong immunity, but of vaccinating the entire population with TT was calculated in NNT deaths averted were calculated only for the year of the one study (Rey and others 1979). In another, the researchers campaign. For the mass campaign in Pidie, Indonesia, Berman examined the cost of the tetanus component of DPT for infants and others (1991) considered the duration of protection from in addition to the cost of TT for women (Barnum, Tarantola, two doses of -T to be four years for women of childbearing age and Setiady 1980). but only one year for women immunized through routine Mass campaign and routine strategies could be compared in prenatal care. Berggren (1974a) assumed that three doses of TT only two studies. In Haiti, the cost-effectiveness of mobile provided protection for a minimum of five years. teams, rally posts, and national campaign and fixed centers None of the studies considered the cumulative benefits from were evaluated (Narcisse 1989). In Indonesia, a provincial- priming the immune system with previous tetanus doses. As- level campaign was compared with routine services (Berman sumptions that benefits from TT immunization begin at the and others 1991). Berggren (1974a) evaluated the cost of the start of a campaign or the beginning of a year for a routine marketplace strategy in Haiti, as well as that of a comprehens- program are made in order to reduce the complexity of assess- ive intervention that included TBA training. Some researchers ing the cost and effectiveness of immunization. included the cost of providing routine immunization services We conclude, then, that the total costs of NNT reduction to women in a sample of health facilities (WHO/EPI 1981b; are likely to be underestimated in this sample of cost-effective- Robertson and others 1985; Shepard and others 1987; Berman ness studies, whereas the benefits are overestimated or under- and others 1991; Brenzel and others 1990). Others examined estimated among studies and even within studies because of the total cost of a tetanus control strategy at regional, provin- varying assumptions. cial, or national levels. Because tetanus incidence is influenced by general social COMPARISON OF STUDIES BY PROGRAM STRATEGY. Most of the and economic factors, attributing the reduction in incidence studies in appendix tables 9A-1 and 9A-2 provide analyses of exclusively to a specific immunization intervention oversim- the two broad categories of strategies for TT immunization: plifies the situation. In Haiti, the incidence of tetanus declined routine programs and mass campaigns. Routine programs may prior to the onset of the immunization program (Berggren include fixed-facility services, mobile teams, outreach, and 1974a). The training program for TBAS may have contributed otherstrategies; but separate data are rarely available regarding to the eventual decline in cases, although the benefit of the cost and cost-effectiveness of each of these types of strate- training TBAS in Haiti is difficult to measure because training gies. As a result, we considered all routine programs as a group overlapped with provision of ATS at delivery and in treatment and compared them with campaigns. For both campaigns and of tetanus. In few of the studies was immunization other than routine programs, target groups may include all women of TT to women considered. Yet DPT and DT administered in childbearing age, pregnant women, schoolgirls, infants, and childhood primes the immune system so that even a single males as well as females. In table 9-7 we compare the unit costs future dose of TT before a woman delivers could be protective. and cost-effectiveness of the routine programs and campaigns None of the researchers, with the exception of Rey, Guillau- in the review. Included in the table are the median values and mont, and Majnoni d'lntignano (1979), addressed the cost of ranges summarized by strategy. reducing both non-NNT mortality and NNT mortality, and none We were able to compare the cost of routine TT immuniza- of them examined the costs or benefits of reaching the target tion programs in eight studies (see table 9-7). Several of the population of school-age children. studies included estimates for different districts, regions, or The comparability of studies is also limited by variability types of facilities, yielding a total of seventeen observations for in methods used to assess outcome. Berman and others the cost per TT dose, per TT second dose (TV2), per case (1991) were the only researchers who used NNT mortality prevented, or perdeath averted; the study done in the Gambia data derived from community surveys before and after the was the only one to include estimates under all four categories. intervention. Community survey data on NNT mortality The outcome measure most frequently available was cost per were available before the intervention in Deschapelles, TT2. Haiti, although comparison of pre- and postintervention As shown in table 9-7, the cost per TT dose found in the eight rates had to be based on hospital admissions data (Berggren studies ranged from $0.40 to $1.76, with a median of$1.14 (all 1974a). The report on the study done in Central Lombok, costs are given in 1989 U.S. dollars). The cost per TT2 ranged Indonesia, had data only from a preintervention community from $0.66 to $11.87, with a median of $3.38. The wide range survey (uNicEF/Jakarta 1985). Robertson and others (1985) in cost per TT2 may reflect different methods for estimating and Kielmann and Vohra (1977) relied on estimates of NNT costs as well as varying levels of efficiency and effectiveness of mortality from community surveys conducted in notably the programs studied. In only two studies, in the Gambia and earlier time periods. In some of the above cases, survey Indonesia (Aceh), was the cost per case prevented and the cost 202 Robert Steinglass, Logan Brenzel, and Allison Percv Table 9-7. Unit Costs and Cost-Effectiveness of Tetanus Immunization Programs (1989 U.S. dollars) Cost per Cost per Cost per case Cost per death Source Locaticn Subdivision TT dose TT2 prevented averted Routine Programs de Champeaux and Martin 1989 Burkino Faso Yako 1.01 4.15 - - Gourcy 0.64 2.13 - - Shepard and others 1987 Ecuador n.a. 0.40 - - - Robertson and others 1985 The Gambia n.a. 1.58 5.23 205 228 Narcisse 1989 Haiti Fixed centers 0.91 - - - Horse teams 1.14 - WHO/EPI 1981b Indonesia n.a. 1.18 3.92 - - (Central Java) Berman and others 1991 Indonesia (Aceh)Y Tanah Pasir - 2.20 89 105 Samudera - 2.76 113 133 Matangkuli - 1.47 60 70 Jeumpa - 0.66 27 32 Brenzel and others 1990 Sudan Darfur 1.76 4.80 - - Kordofan 0.86 2.16 - - Capital 1.33 2.84 - - Nationwide 1.73 4.20 - - Phonboon and others 1989 Thailand Hospitals - 10.25 - - Health centers - 11.87 - - Minimum n.a. n.a. 0.40 0.66 27 32 Maximum n.a. n.a. 1.76 11.87 205 228 Median n.a. n.a. 1.14 3.38 89 105 Campaigns Brenzel 1987 Cameroon n.a. 1.34 4.16bc c Berggren 1974'b Haiti n.a. 0.34 1.05 98 115 (Deschapelles) Narcisse, 1989 Haiti Mass campaign 1.71 - - - Rallyposts 0.21 - Kielmann and Vohra 1977 India Initial 0.29 - - 97 (Narangwa)d Maintenance n.a. - - 0.34 UNICEF 1985 Indonesia n.a. 0.23 0.46 44e 52 (Central Lombok) WHO/EPI 1988a Indonesia n.a. 0.82 1.84 122 144 Berman and others 1991 (Pidie) M 25 Rey, Guillaumont and Senegal Mass campaign - - 825 2,750 Majnoni d'lntignano 1979 (Dakar) Brenzel and others 1987 Senegal n.a. 0.76 - - Minimum n.a. n.a. 0.21 0.46 44 52' Maxinium n.a. n.a. 1.71 4.16 825 2,750 Median n.a. n.a. 0.55 1.45 110 115 Total (routine programs and campaigns) Minimum n.a. n.a. 0.21 0.46 27 32 Maximum n.a. n.a. 1.76 11.87 825 2,750 Median n.a. n.a. 0.91 2.80 98 110 - Data not available. n.a. Not applicable. a. Protective effect of TT2 on all deliveries within three years not considered; hypothetical incidence data. h. Derived from information reported in cited document. Schedule included three doses of TT. c. Excludes an additional 630 non-NNT cases prevented. d. Study was for one high-potency dose of n with assumed 80 percent efficacy and lifelong immunity; cost data are hypothetical. e. Assumed NNT mortality rate equals 28.0 (level for seven clusters in Central Lombok district in a thirty-cluster survey). f. Includes cost per NNT and non-NNT case and death prevented. g. Excludes India/Narangwal maintenance level estimate. Source: See first column of this table and tables 9A-2 and 9A-4. Tetanus 203 per death averted estimated.- The median cost per case pre- than one year, resulting in a large difference between the cost vented was $89, with a range from $27 to $205. The cost per per case prevented and the cost per death averted. death averted ranged from $32 to $228, whereas the median Berggren (1974a, 1974b) estimated that in Haiti an addi- was $105. The cost per death averted does not differ greatly tional 630 cases of non-NNT were prevented. This estimate has from the cost per case prevented because of the high case- not been incorporated into the cost per case or death averted fatality rate of neonatal tetanus. in appendix table 9A-2 or in table 9-7. Eight studies listed in table 9-7 also included unit costs and In general, this review shows a relatively small range in unit cost-effectiveness for TT programs employing campaign strate- costs and cost-effectiveness. This is surprising given the differ- gies. All but one included estimates of the cost per rr dose ent methods used in the 15 studies for analyzing the cost and delivered. As shown in table 9-7, these estimates ranged from effect of diverse interventions. Little disparity was found be- $0.21 to $1.71, with a median of $0.55. One of the lowest tween the median values for routine strategies as opposed to estimates ($0.29) was from Kielmann and Vohra's study in campaign strategies, which implies that each strategy may be Narangwal, India, where a single dose of high-titer (30 Lf per cost-effective in different circumstances. Whereas median milliliter) calcium phosphate-adsorbed vaccine was used. costs per rr and T12 doses are lower in campaigns, routine Kielmann and Vohra (1977) found that this single dose was strategies were more cost-effective in cases and deaths averted. more effective than three doses of aluminum phosphate- The choice of strategy must take into account its likely pro- adsorbed TT (10 LF per milliliter) given at one-month inter- grammatic efficacy-that is, the efficacy of the technology and vals. If this assertion is true, it may be more appropriate to irs expected coverage within the population. In other words, compare this cost ($0.29) with the cost per TT2 found in it makes little sense to base a delivery strategy solely on Cameroon ($4.16), Deschapelles in Haiti ($1.05), and Central pregnant women attending fixed facilities for prenatal check- Lombok ($0.46) and Pidie ($1.84) in Indonesia, because these ups when, in many countries, only a small minority of women costs represent the nearest estimates of the cost to protect a seek prenatal care. woman fully against tetanus for several years. The median cost per Tr2 was $1.45. SIMULATION MODELS. We review four hypothetical models Estimates of the cost per NNT case prevented are available which were used to simulate the costs and benefits of alterna- for four studies, and range from $44 to $825, with a median of tive tetanus immunization strategies (Cvjetanovic and others $110. The wide range in these estimates may be due in large 1972; Kessel 1984; Sharma and Sharma, 1984a, 1984b; and part to different methods of estimating costs and morbidity Smucker, Swint, and Simmons 1984; see appendix tables 9A-3 reductions. and 9A-4). Given the complicated immunologic aspects of The median cost per NNT death averted in the studies which tetanus immunization, modeling exercises provide valuable examined campaign strategies was $115, reflecting a range insights from an epidemiologic perspective into the effective- from $52 to $2,750. The study in Narangwal, India, includes ness of altemative strategies. The greatest weakness of each two estimates for the cost per death averted: one for the initial model is the lack of empirical basis for cost estimates. Kessel phase, during which 87 percent of all women from fifteen to (1984) uses an arbitrary system of financial units (from I to forty-four years ofage were immunized; the other, a much lower 100), which ranks the difficulty of implementing alternative estimate ($0.34), for maintenance of the program, in which immunization strategies with regard to their effect on unit cost only the girls fifteen years of age who enter the eligible cohort per dose (unit costs are expected to rise as the most remote each year were immunized. This lower estimate has been left population groups are reached, usually through outreach and out of the calculation of the median cost per death averted mobile services). The results of Kessel's simulations, however, ($115) because of its extremely low and hypothetical nature. are heavily influenced by the order of magnitude and ranking Because Kielmann and Vohra (1977) assume lifelong immun- of these delivery strategies. Cvjetanovic and others (1972) do ity from one dose of high-titer vaccine, they claim that only not consider economic costs of implementing an immuniza- those women who are entering childbearing age will need to tion program but include only vaccine and treatment costs in be vaccinated during the maintenance phase of the program. their model. Sharina and Sharma (1984a, 1984b) base their Nevertheless, despite the assertion of lifelong immunity, they model on vaccine costs as well, including (unlike the other include in their cost-effectiveness calculations only those models) costs of three DPT and three DT shots to young children deaths prevented in a single year. This results in an underesti- and schoolchildren, respectively. Only Smucker, Swint, and mate of the cost-effectiveness, if indeed one agrees with the Simmons (1984) estimate the economic cost of an immuniza- assumption of lifelong protection. tion strategy aimed at TT vaccination of women. Yet they do The extremely high estimate of cost per death averted (NNT not consider the integration of TT vaccination (through MCH plus non-NNT) found in Dakar, Senegal, may have resulted services) with EPI in India. Joint implementation of immuni- from the high-intensity program used in that instance: seven zation of women and children would substantially alter their semestral mass vaccination campaigns were undertaken assumptions about frequency of contact by women with the against tetanus (Rey, Guillaumont, and Majnoni d'lntignano health system and productivity of health workers (for exam- 1979).Moreover,thecase-fatalityratewasonlyapproximately ple, the number of women who could be immunized per 30 percent because of the high number of cases in those older day). 204 Robert Steinglass, Logan Brenzel, and Allison Percy Cvjetanovic and colleagues (1972) assume high immuniza- tions are not directly comparable to those of the studies, they tion coverage (90 percent) of pregnant women through rou- provide useful insights into the effect of programmatic changes tine services and medium coverage (50 percent) of the total on both the costs and outcomes of tetanus control programs. population through mass strategies. Empirically, however, Cvjetanovic and others (1972) estimate a range of costs per these assumptions are reversed. Mass campaigns in Central case prevented from $245 to $595 (all costs are in 1989 U.S. Lombok and Pidie District, lndonesia; Narangwal, India; and dollars), depending on delivery strategy and target population Deschapelles, Haiti, resulted in greater than 80 percent cover- assumed. Smucker, Swint, and Simmons (1984) vary their age, whereas coverage from routine fixed-tacility services was coverage assumptions and cost estimates to compare a routine less than 50 percent of the target population of pregnant program with a campaign using teams of vaccinators. For the women. The cost-effectiveness of -r immunization in Kessel's campaign approach, costs per TT dose range from $1.34 to (1984) model is not dependent on coverage levels, which $2.33, whereas the cost perTr2 varies from $2.74 to $5.25. The appears counterintuitive from economic theory: average costs cost per death averted for this strategy ranges from a low of change as coverage increases. There is no simulation in any $6.75 to a high of $11.86. The routine program shows some- model of the costs and benefits of continuous immunization of what better results: the cost per TT dose ranges from $0.50 to women of childbearing age, the strategy recommended by WHO. $0.98; costs per TT2, from $1.03 to $2.20; and the cost per death As in the empirical studies, varying assumptions were averted, from $2.59 to $4.87. It should be noted that these costs made in the hypothetical models regarding the efficacy of per death averted are far less than the estimates found in all tetanus immunization. The duration of immunity differed the empirical studies, which may indicate that the cost esti- among these models from five years (Sharma and Sharma mates were too low or that effectiveness assumptions were 1984a, 1984h; Smucker, Swint, and SimmonF 1984) to ten overly optimistic. years (Cvjetanovic and others 1972). The primary series of TT Kessel (1984), using theoretical financial units as a means for full protection ranged from two doses (Smucker, Swint, and of comparing the resource requirements of different strategies, Simmons 1984 and Sharma an:d Sharma. 1984a, 1984b) to concludes that school-based programs are the most economical three doses (('vjetanovic and others 1972). Sharma and in the long run because all the residual fertility of the girls is Sharma (1984a, 1984b) include the costs and benefits of three protected. Sharma and Sharma (1984a, 1984b) favor contin- doses of DPT and two doses of PT, whereas Kessel (1984) uous immunization of pregnant women, because mass im- includes four doses of DPT for preschool children. Cvjetanovic munization programs are projected to lead to only short-term and others (1972) and Sharma and Sharma (1984a, 1984b) declines in tetanus incidence. include the effect on NNT and non-NNT. Only one model discounts the future costs of the immtunization program-at a REVIEW OF STUDIES AND SIMULATION MODELS: CONCLUSIONS. rate of 12 percent per year (Smucker, Swint, and Simmons This review of empirical work and simulation models has 1984). The benefits of the program (for example, deaths shown that immunization of women with TT isacost-effective averted) are not discounted. Kessel (1984) estimates the effec- means of controlling NNT. The median cost per NNT case tiveness of tetanus immunization programs by calculating the averted is $98 and the median cost per NNT death averted is residual protected fertility achieved by directing efforts at $110. These estimates are similar for both routine delivery of preschool girls, school-age girls, and adult women seeking TT and campaigns. Although cost-effectiveness is but one prenatal care. criterion for choosing among alternative health interventions, As with cost-effectiveness studies, the modeling exercises these figures compare favorably with those of other interven- probably underestimate the costs of delivering immunization tions presented in the rest of this collection. services to women and children, whereas benefits are some- For three reasons, however, the results of this review can be times underestimated and sometimes overestimated. Costs of generalized only in a limited fashion. First and most important, complementary programs, such as immunization of school- the cost of controlling tetanus depends on country-specific children with DT and infants with DpT and training of tradi- characteristics, such as the health infrastructure, delivery strat- tional birth attendants, are not factored into all the egies used, incidence rates of the disease and coverage rates simulations. Similarly, the benefits from NNT reduction are achieved, and the degree of integration of tetanus control assumed to accrue solely from TT immunization, without regard within routine EPI, MCH, and school health services. A country for active-passive immunization for wound management (in- with a well-developed health system and universal access to cluding ATS for newborns) and safe deliveries. Thus, the cost services is more likely to have lower average costs for a TT per dose and cost per death averted are also underestimated, immunization program than a country which is still extending although, compared with other interventions described in this basic health services to its population. Integration of tetanus collection, efforts to control NNT by immunization are likely to control with basic health services will also tend to reduce be highly cost-effective. average costs. The empirical studies included in this review reflect the COMPARISON OF RESULTS OF SIMULATION MODELS. Only two of cost of providing services at particular coverage levels through the four models produced usable cost-effectiveness estimates specific strategies. The cost of providing routine services at for our present purposes. Although the results of these simula- low coverage levels will not be representative of average Tetanus 205 costs at higher levels, because economic theory predicts that of DPT has been an underused strategy. Besides a simulation average costs change as volume increases. All the studies were exercise (Kessel 1984) which evaluates the costs and benefits conducted in countries with high NNT incidence, so results of school immunization against tetanus, little empirical data cannot be generalized to countries where tetanus incidence are available about the potential costs (in relation to benefits) is low, and marginal costs of averting cases will consequently of a nationwide school program in a developing country. be much higher. In no study or simulation did researchers The Kessel (1984) model predicts that school immuniza- examine the costs and benefits associated with the contin- tion is the most cost-effective strategy for reducing NNT uous immunization of women of childbearing age, the cur- deaths, compared with a preschool program, prenatal clinic, rent strategy recommended by WHO for the worldwide and outreach immunization of pregnant women. This model, elimination of tetanus. however, assumes universal access to education and a well- The second reason generalization from country-specific re- developed institutional base from which to deliver immuni- sults is difficult is because the opportunity cost of health zations. The actual situation in most developing countries resources is not consistent among countries. A cost per death differs from these assumptions: access is often not universal and averted of $144 in Indonesia (Berman and others 1991) may attrition rates are high, particularly among young girls-the not represent a different set of tradeoffs for resource allocation target population for immunization. Similar to the health from a cost of $228 in the Gambia (Robertson and others sector, the education sector is weak institutionally and finan- 1985), and the situation in Indonesia may differ greatly from cially in most developing countries. Therefore, the applicabil- that in Haiti ($115 perdeathaverted; Berggren 1974a, 1974b). ity of these simulation results for decisionmaking in many The decision to invest public resources in tetanus control developing countries is limited. Still, in countries with high through immunization must be made within the context of the enrollment of female primary school children, DT immuniza- marginal product resulting from competing country-specific tion may prove to be highly cost-effective for tetanus preven- use of those same resources. tion, especially when it is an incremental measure in an Finally, variations in the methods used in the empirical integrated primary health care strategy (which may at the same studies and simulations hamper casual generalizations of these time include vitamin A distribution, administration of anthel- results. Still, the convergence of many study results around a mintics, and screening for trachoma). figure of $110 per death averted suggests that this median Single-dose tetanus toxoid vaccine which would confer estimate can be used as a starting point for determining re- lifelong immunity is likely to become available for human use source allocations for tetanus control (see table 9-7). Tetanus within five to ten years. Testing in animals has already begun, control through immunization is a highly cost-effective and with testing in humans expected soon. Funds may be needed inexpensive means of reducing infant mortality in the devel- in the future to support trials in humans. The vaccine would oping world. contain biodegradable microcapsules of different thicknesses which would slowly release tetanus toxoid over a prolonged ADDITIONAL ECONOMIC ISSUES. In addition to immunization period. Administered early in life, the vaccine would provide of women and children, other strategies exist which can po- protection for the individual and for future pregnancies. Al- tentially affect the incidence of NNT. Among these are training though unit costs per dose of this vaccine cannot be predicted, of TBAs in more hygienic delivery practices. Several studies it is unlikely to be expensive. The single-dose vaccine would allude to the effectiveness of TBA training in reducing NNT, also significantly reduce the number of required immunization though controlled community studies with pre- and post- contacts, resulting in significant cost savings. This same bio- intervention evaluation have been rare (Berggren 1974a; All- technology relying on impregnated microcarriers may be useful man 1986). Allman (1986) estimated TBA training to cost $10 for delivering other inactivated vaccines. per worker between 1977 and 1982. James Heiby reports a Pregnant women and those of childbearing age are fre- figure of $92.50 per worker in Nicaragua during the late 1970s quently not screened and immunized with tetanus toxoid (cited in Allman 1986). when they come into contact with the health system for By contrast, many observers conclude that TBA training is whatever reason. The World Health Organization recom- not as effective in preventing NNT as vaccination of women mends immunization of eligible women and infants at every (Ross, 1986a, 1986b; Solter, Hasibuan, and Yusuf 1986; Bhatia encounter with the health system in order to reduce missed 1989; Jordan 1989). The expense of training, supervising, and opportunities. Immunization of pregnant women during supporting workers to cover the population would also be routine prenatal care visits as early as possible during preg- prohibitive for most developing countries, particularly if case nancy and immunization of women of childbearing age at loads per birth attendant are low. Although some evidence childhood immunization sessions should lead to reduced suggests that TBA training programs are relatively effective in costs of the immunization program by substantially increas- reducing nontetanus neonatal mortality (Rahman 1982), ing efficiency (WHO/EPI 1988b). large-scale efforts are unlikely to be as cost-effective as targeted Routine immunization of women either is conducted immunization strategies for NNT reduction. through antenatal clinics and MCH services or is the respon- Immunization of schoolchildren either with a primary series sibility of EPI. Where possible, routine TT immunization (and of DT or with additional reinforcing doses ofDT after three doses mass immunization strategies) should be designed as inte- 206 Robert Steinglass, Logan Brenzel, and Allison Percy grated efforts with MCH services. If TT immunization is given rorespiratory resuscitation and use of specialized intensive care as part of an integrated package with childhood immuniza- units have led to improved outcomes. Nevertheless, even tion and prenatal care services, the cost is likely to be treated tetanus remains extremely serious with the course and reduced. prognosis dependent on age, preexisting conditions, superim- The World Health Organization has recommended expand- posed infections and complications of treatment, and avail- ing the target population for tetanus toxoid immunization from ability of medical facilities with advanced equipment and pregnant women to all women of childbearing age (WHO/EPI expert staff(Veronesi and Focaccia 1981). 1986). Initially, there was concern that enlarging the popula- Expert use and timing of sedatives, muscle relaxants, and tion would not be an affordable strategy for developing coun- respiratory assistance (including tracheotomy and artificial tries. Gerard Foulon investigated the incremental cost of ventilation, if indicated) are typically required, but these pro- implementing a five-dose schedule for all women of childbear- cedures carry their own iatrogenic risk. Inaccessibility of spe- ing age (personal communication, July 12, 1990). He found cialized treatment facilities in much of the world results in that total costs would increase within the first five years of delayed admission, a factor associated with increased adopting this strategy but would then revert to preexpansion mortality. costs. If the incremental costs of additional vaccine, storage, Treatment of NNT and non-NNT patients consists of excising, training, and monitoring could be financed through donor cleansing, and disinfecting the wound; antibiotic therapy; use resources, this policy would not represent an economic burden of benzodiazepines for their sedating, anticonvulsant, and mus- to developing countries in the short run. cle-relaxing properties; maintenance of effective ventilation, To the extent that parents will quickly replace their lost particularly by tracheotomy; parenteral administration of human baby, a death from NNT will continue to exact incalculable tetanus immune globulin or antitetanus serum of equine origin, costs within the family. In the absence of lactational amenor- which is less desirable because of frequent adverse side effects; rhea following a baby's death from NNT, the mother may well maintenance of water, electrolyte, and nutritional balance; become pregnant sooner. Shorter pregnancy intervals are as- and intensive nursing care. (Neonates in particular require sociated with higher infant and child mortality. Even if the assisted respiration in specialized neonatal intensive care older child dies in infancy, too short a birth interval still places units.) The effectiveness of this symptomatic treatment de- the younger child at very high risk. Another pregnancy so soon pends on financial and human resources rarely available where also jeopardizes the health of the mother. tetanus most frequently occurs. The REACH project determined that half of immunization Following the treatment outline above, hospitals in France costs (capital and recurrent) in developing countries are cur- and Japan have recorded impressive reductions in fatality of rently financed by external resources from donor organizations non-NNT patients to a rate of approximately 10 percent despite (Brenzel 1989). Tools for making resource allocation deci- the increasing age of patients in recent years (Rey, Diop-Mar, sions, such as cost-effectiveness analyses, have not been used and Robert 1981; Ebisawa and Homma 1986). The principal often in cases where resources are abundant. Sustaining im- determinant of survival in Japan was admission to an intensive munization coverage gains, however, is becoming a greater care unit where staff, including anesthesiologists, were trained priority for program managers, and this translates into selecting in treatment of tetanus. After 1974, gastrointestinal and car- the most affordable and effective strategies for tetanus preven- diac complications overtook respiratory insufficiency with or tion. The estimates calculated in this chapter underscore the without pulmonary infection as the leading cause of death for need for continued donor assistance in the financing of rT Japanese tetanus patients. Rey, Diop-Mar, and Robert (1981) immunization programs, given the economic declines and report that intensive care of newborns has in a few instances growing populations faced by most developing countries. even reduced neonatal fatality to 10-20 percent from 90 percent, although the risk of incapacitating sequelae has in- Case Management of Tetanus creased owing to survival after intensive care. Rey, Diop-Mar, and Robert (1981) suggest that in develop- This section discusses strategies for the treatment of tetanus ingcountriesacompromisebetween modern medicaladvances and their costs and benefits. and available resources is needed. They advocate establish- mentofspecial care units in facilities admittingnmore than 100 Case Management Strategies cases per year. Such a unit would admit other patients also requiring continuous monitoring. Much of the following discussion on case management of NNT Practically, the financial and human resources which they and non-NNT is based on an authoritative essay entitled "Treat- recommend for tetanus patients may still be out of reach for ment of Tetanus" (Rey, Diop-Mar, and Robert 1981), to which most developing countries: one doctor continuously available; the reader is referred. Death from tetanus commonly occurs in two daytime nurses and one nighttime nurse, which would association with spasms, which lead to acute asphyxia. Treat- require a team of six to eight trained nurses; several orderlies; ment is largely symptomatic and attempts to prevent and oxygen; apparatus for aspiration, intravenous infusion, cathe- counteract the effects of spasticity and spasms on respiration. terization of veins or bladder, and tracheotomy; nasogastric Improvements over the past few decades in methods of neu- tubes; and appropriate sedatives, antibiotics, and serum. Lab- Tetanus 207 oratory tests, bacteriological examinations, and air condition- Table 9-8. Costs Per Capita of Tetanus Treatment ing are desirable but may have to be omitted. and Immunization in Selected Areas Even with relatively simple and inexpensive treatment, it is (1989 U.S. dollars) in the newbom that the most noticeable improvement in Cost of Cost of survival rate is found (Rey, Diop-Mar, and Robert 1981). Location Type of case treatment vaccination Schofield (1986) observes, however, that with routine treat- ment an overall fatality reduction of only 10 percent can be 1 expected. Even in the United States as recently as 1953-61, India Average 319 - the case-fatality rate for all tetanus remained as high as 63 (Delhi, Safdar- percent (Bytchenko 1966). In the absence of sophisticated jang Hospital) equipment and the most advanced drugs, the quality and Iran, Islamic Govemment hospital 319-807 1.00 continuity of nursing care, which allows early recognition and Rep. of Private hospital 957-2,042 6.40-10.50 treatment of potentially life-threatening complications, is Senegal Fatal 255-511 probably the most important factor when case fatality varies (Dakar) Surviving 460-2,374 0.90-1.80 greatly from place to place (Barten 1969). Average 638-766 Yugoslavia Simple 287 Cost and Benefits of Treatment Artificial respiration 0.30-1.60 required 2,872 1 There have been some attempts to examine the costs and Developing Average 319-957 1.00-1.60 benefits of treatment of tetanus compared with prevention countries though immunization. Treatment costs are more relevant for non-NNT, since NNT cases are rarely brought to the hos- Note: All costs have been converted into 1989 U.S. dollars, assuming costs pital and the case-fatality rate of NNT approaches 100 per- in original studies were in 1970 dollars. cent. Cvjetanovic and others (1972) state that treatment Source: Cvjetanovic, Grab, and Uemura 1978. costs may vary from $50 to $900 ($148 to $2,665), with an average of $200 per case ($592) in developing countries.3 tetanus immunization program, which allowed a redistribution Rey, Guillaumont, and Majnoni d'Intignano (1979) esti- of $600,000 worth of care from tetanus to other priorities. He mate treatment costs from $15,000 to $20,000 ($28,500 to estimated a benefit-to-cost ratio of 9 to 1. $38,000) per case in the United States and $10,000 to Rey, Guillaumont, and Majnoni d'Intignano (1979) esti- $16,000 ($19,000 to $30,500) in France. Berggren (1974a) mate the costs of altemative immunization strategies and estimates treatment costs of $12 ($30) per day at Albert benefits in treatment costs saved over a thirty-year period (see Schweitzer Hospital in Haiti, with an average of seventeen table 9-9) based on the model developed by Cvjetanovic and days of treatment. Rey and Tikhomirov (1989) report a others(1972).Assumingthatthecostpervaccinationis$1.18 mean hospital stay for non-NNT patients of sixteen days, (all costs have been converted to 1989 U.S. dollars) and the though a significant proportion of cases die within the first treatment cost for one case is $592, they calculate that contin- two or three days. Griffith and Sachs (1974) report a mean uousvaccinationof pregnantwomenwould result in the lowest hospital stay in Ludhiana, India, of eighteen and one-half cost per case averted ($245), compared with one mass vacci- days and a direct daily cost of 991 rupees ($226). When only nation campaign ($463 per case averted), repeated mass cam- surviving patients are considered, a total of 1,857 rupees paigns ($595), and a combination of repeated mass campaigns ($423) were spent to treat each infant. and continuous vaccination ($468). This model does not, Table 9-8 provides treatment costs per case of tetanus (NNT however, examine costs of continuous (routine) immunization and non-NNT) in various locales. It must be emphasized that of all women of childbearing age. treatment protocols (use of drugs, ATS, tetanus immune globu- This information on the costs of treatment of non-neonatal lin, ventilation, and so on) are not uniform and that cost tetanus can be used to estimate the cost-effectiveness of case methodologies vary; therefore, comparison of these estimates management.4 The range of treatment costs for developing is not straightforward. Nevertheless, with a median cost per countries cited in table 9-8 is from $319 to $957. Footnote b NNT death averted of$1 10by means of Tr immunization (table of table 9-1 indicates that treatment in hospitals of non- 9-7), prevention of tetanus is by far more cost-effective than neonatal tetanus cases may reduce the case-fatality rate by 15 treatment. Cvjetanovic and others (1972) estimated that cu- to 20 percentage points. Dividing the cost of treatment by this mulative savings in treatment costs over the course of thirty change in cFR gives a range in estimated cost per death averted years would exceed by a factor of more than 2.5 to I the cost of $1,595 to $6,380. As mentioned above, Cvjetanovic and of continuous immunization of pregnant women. Costs associ- others (1972) estimate an average cost of $592 per case treated ated with passive immunization as part of wound management in developing countries, resulting in a cost per death averted would also decrease as the population becomes protected by of $2,960 to $3,947, depending on the assumed reduction of TT. Berggren (1974a) calculated that more than 50,000 hospi- 15 to 20 percentage points in the cFR. Thus, treatment of tal days were saved over a four-year period by Haiti's vigorous non-neonatal tetanus appears to be substantially less cost- 208 Robert Steinglass, Logan Brenzel, and Allison Plercy Table 9-9. Cost-Benefit and Cost,Effectiveness protect this population for life against tetanus and their off- of Various Immunization Programs in Developing spring against NNT. This figure is less than the total potential Countries over a Thirty-year Period savings resulting from avoided economic loss, avoided treat- (1989 u.s dollars) ment, and prevention of disability. The cost to protect all Number Treatment Cost per unimmunized women of childbearing age would be less than Cumulative of cases cost case half this amount and would eliminate NNT and reduce adult Immunization program cost averted saved averted tetanus by half. None' n.a. n.a. n.a. n.a. The estimates of tetanus deaths can be used to project crude cost-effectiveness figures for prevention of adult tet- One mass campaignh 592,288 1,278 756,944 463 anus cases.5 Murray, Yang, and Qiao (1992) estimate that Repeated mass campaigns 10.6 million deaths occur annually in developing countries at ten-year intervals 2,194,391 3,686 2,183,173 595 in the adult age group (fifteen to fifty-nine years of age). Continuous vaccination From tables 9-1 and 9-2 earlier in this chapter, we can of pregnant women' 1,513,761 6,184 3,662,708 245 estimate that 132,500 (1.25 percent) of these deaths are due Combination of repeated to tetanus. An individual in a developing country has ap- mass campaigns and proximately a 24 percent chance of dying of any cause continuous vaccination between the ages of fifteen and fifty-nine (Murray, Yang, of pregnant women 3,708,152 7,922 4,692,105 468 and Qiao 1992). By derivation, an adult in a developing n.a. Not applicable. country has approximately a 0.3 percent chance of dying of Note: Assumes population of one million at beginning of the perioi; tetanus between those ages. thereafter, annual growth rate is 2 percent. Cost of one vaccination is $1.18; treatment cost for one case is $592. Full immunization by the age of fifteen (through a combi- a. Without immunization program, incidence would be 400 per 100,000 nation of DPT, DT, Td, and TT during infancy, at school, and newborns and 18 per 100,000 population, during other contacts with the health system) would require b. Assumes coverage of 50 percent, vaccine effectiveness of 95 percent. c. Assumes coverage of 90 percent, vaccine effectiveness of 95 percent. five properly spaced shots and would most likely fully protect Source: Rey, Guillaumont, and Majnoni d'lntignano 1979, based on an individual from the risk of tetanus for life. Cvjetanovic and others (1972). Unfortunately, few of the studies reviewed earlier included examinations of the cost of providing tetanus as a part of infant effective than prevention. Prevention should therefore remain and school immunization programs. For the most part, they the intervention of top priority. focused on reaching adult women in order to prevent neonatal Non-NNT is important from an economic perspective in that tetanus deaths. The marginal cost of providing the "T" in DPT treatment costs are high, death is common, long-term disabil- is likely to be very small, and the costs of providing DT or Td ity may ensue, and prolonged illness may result in lost produc- to schoolchildren might be substantially different from the tivity for the family and society (Rey, Guillaumont, and costs of providing Tr to pregnant women. Nevertheless, the Majnoni d'lntignano 1979). Age-distribution data from stud- median cost per rr dose of $0.91 (table 9-7) is the most ies conducted in Bombay (1954-79) and Dakar (1960-86) reasonable estimate we have available of the unit cost of indicate that a high proportion of deaths and between 50 and delivering tetanus immunization. If five doses were provided 60 percent of non-NNT cases occur among the economically at a median cost of $0.91 per dose, full protection could be productive groups age ten to fifty-nine years (Rey, Guillau- bought for $4.55. mont, and Majnoni d'lntignano 1979). If this cost is divided by the individual risk of an adult dying With increasing immunization of infants and older children of tetanus between the ages of fifteen and fifty-nine (0.3 with DPT and DT, an epidemiological shift in incidence to older percent), we can derive a cost per death averted of$ 1,517. This age groups is expected to continue. These epidemiological cost of preventing adult tetanus is nearly fourteen times higher shifts will have an effect on the economic productivity of than the median cost per NNT death averted ($110) found in developing societies. table 9-7, but it compares well with other adult interventions The annual cost fully to protect the entire adult popula- discussed elsewhere in this collection. tion of the developing world is difficult to estimate because vaccinations administered in past years may still be protec- Research Agenda for the 1990s tive. Unlike infants (a cohort which renews itself annually and is consequently easily calculated), women receive mul- A review of the recent literature has identified a research tiple doses of TT at varying intervals over a thirty-year agendaforthe 1990s. Such alengthy list isnot meant to suggest reproductive span; and women enter and leave the eligible that current control efforts should await research findings. age range continuously (Steinglass 1990). Still, assuming a Enough is already known about the benefits of NNT control cost of $0.91 per dose of TT (table 9-7), a five-dose schedule efforts to justify vigorous implementation at country level. The for lifelong protection, and a total adult population of 2.4 research agenda includes topics in vaccinology, epidemiology, billion, we arrive at a cost of approximately $10.9 billion to programmatic concems, and behavioral science. Tetanus 209 Vaccinology * Include in the standard thirty cluster community surveys of immunization coverage questions on matemal TT status, * Determine the nature, action, and duration of immunity age and parity of mother and protection status of delivery, from primary and reinforcing doses of TT with varying po- number of prenatal care visits, place of delivery and atten- tencies and intervals in different settings (Jones 1983). dant, and circumstances of delivery. * Definetheeffectofthevaryinglevelsofmaternaltetanus * Include TT in surveys on missed opportunities for antitoxin on the level and duration of infant tetanus anti- immunization to determine magnitude and correction of toxin following varying doses of DPT (wHo/EPI 1989b). problem. * Develop and test a TT vaccine which is more immunoge- * Conduct serological (gold-standard) assessments of im- nic with fewer doses, such as a single-dose high-potency TT munologic and immunization status in the community as with pulsed or continuous release of toxoid, with or without part of NNT mortality and thirty cluster coverage surveys adjuvants, that uses alternative polymers as the vehicle (Schofield 1986). (Galazka 1983; WHO/EPI 1989a). SurveiUance Operational Strategies * Develop various methods of identifying high-risk popu- * Identify and implement altemative cost-effective NNT lation subgroups and areas for focused interventions. prevention strategies in a variety of settings (including hard- * Determine the magnitude of matemal mortality due to to-reach areas), identify costs and operational constraints, tetanus. and document the effect on immunological status and NNT incidence. Such strategies include expanding the TT schedule lae Determine the feasibility of community-based surveil- to five doses and the target groups to all women of child- lance and reporting of NNT cases and the utility of case bearing age, immunizing at every contact with the health ivestigations. services, offering immunization at markets, immunizing * DeterminemethodsofdocumentingtheabsenceofNNT schoolchildren in the early grades, launching mass campaigns as part of the elimination effort. every five or ten years, incorporating TT in national vaccina- * Determine in selected areas the relationship of altitude tion days, enforcing compulsory TT before marriage certifi- to NNT incidence to concentrate control activities in high- cates are issued, scheduling a routine dose of TT at the start risk areas. of every decade of life-at ages ten, twenty, thirty, and forty. * Determine practical methods of screening and immuniz- Impact Evaluation ing all women of childbearing age on every contact with the health services. * Assess sustained effect of TBA training on changing de- livery and cord care practices and on NNT and neonatal * Determine methods of identifying and routinely im- mortality in "before" and "after" control and experimental munizing women entering the childbearing age. areas (Ross 1986a). * Explore potential use of TT outside the cold chain, includ- * Conduct retrospective case-control studies of the effect ing the possibility of administration by prefilled single-use of TBA training (and TT immunization) on neonatal mortal- injection devices which can be used by peripheral workers- ity and NNT in areas where some TBAs have been trained and for example, TBAs or village health workers (WHO/EPI 1989a; others have not (Ross 1986b). WHO/EPI 1989b). * Conduct studies on the incremental effect of TBA train- * Study use, distribution, and effect of disposable delivery ing, above that achieved by TT immunization alone, on NNT kits in a variety of settings. and neonatal mortality (Ross 1986b). Monitoring Social Factors * Develop and apply valid criteria, guidelines, operational * Review experiences of the use of techniques of social indicators, and methodologies to monitor levels of TT cov- marketing and social mobilization directed at the problem erage and protection, clean delivery, and cord care OfNNT. (Steinglass 1988; WHO/EPI 1989b). * Conduct market research and practical behavioral re- * Review experience using lifetime home-based records search on immunization acceptability, intrapartum and post- and develop several record-keeping options forTT protection partum care, and cultural perceptions of NNT to identify and in areas employing different NNT prevention strategies. overcome resistance on the part of the public and providers Study factors for promoting retention of records by the and to promote TT immunization and clean delivery prac- public. tices (Bastien 1988; Pillsbury 1989). 210 Robert Steinglass, Logan Brenzel, and Allison Percy Study methods of community involvement for routine * Cost-effectiveness and opportunity cost of other health identification and referral of females for TT, including use of interventions and strategies TBAS, women's groups, political structure, and religious leaders. * Incremental cost of different TT strategies Cost-Effectiveness * Operational and behavioral considerations As was demonstrated by the studies in Deschapelles, Haiti * Study the cost and cost-effectiveness of alternative NNT (Berggren, 1974a), and Pidie District, Indonesia (Berman and control strategies, especially TT immunization of all women others 1991; WHO/EPI 1988a), mass campaigns can be effective of childbearing age and TBA training. Study the logistical in rapidly reducing the backlog of unimmunized individuals in implications and resource requirements for widened im- the target population. Continuous immunization through rou- munization target groups, so as to influence local decisions tine services is a more common approach and will be necessary on resource allocation. in most cases to ensure continued protection of the population * Refine simple costing guidelines on alternative NNT pre- over time. In the cost studies reviewed, the median cost per vention strategies for use by program managers for decisions NNT death averted was $110. on resource allocation. As Berman and others (1991) noted: "The appropriate agenda for planning is not an absolute choice amongst differ- Conclusions ent strategies, but a flexible schedule for how different ap- proaches can be combined over time to maximize results at an Tetanus kills 750,000 babies annually, and non-NNT kills an affordable cost. This approach was suggested by Cvjetanovic additional 120,000 to 300,000 persons. Neonatal tetanus is et al. (1972) and still remains valid." completely preventable by means of matemal immunization The World Health Organization now recommends efforts with tetanus toxoid or aseptic care of the umbilical cord. to eliminate NNT worldwide by 1995. Achievement of this Prevention OfNNT will reduce neonatal mortality by up to half global target will require a global commitment of re- and infant mortality by up to one-quarter in unimmunized sources and mobilization of political will at all levels. populations. Increasingly, the level OfNNT is being recognized Unlike other eradication and elimination efforts (for ex- as a barometer of the health status and well-being of mothers ample, smallpox and polio), there can be no cessation of and newborns, with each case attesting to multiple failures of vaccination and revaccination efforts once NNT elimina- the health system (Galazka and Cook 1985). tion is achieved, because the infective agent exists in the Prevention of NNT should be a priority for resource alloca- environment and cannot be eradicated. Elimination itself tion in manydevelopingcountries, given the magnitude of the will need to be sustained forever by means of active disease (high incidence rates in poorer countries), the severity immunization. Neonatal tetanus is easily preventable and of the disease (high case-fatality rates even with treatment), can be eliminaced as a public health problem in most the high cost of treatment, and the availability ofa safe, highly countries at a reasonable cost. This cost would be afford- efficacious, and cost-effective vaccine. able for most countries, although many of the poorer The strategies chosen for immunization, as well as the target countries (which also tend to have the largest tetanus groups, should be defined locally and will depend on a number problem) will require donor assistance for years into the of different factors, including: future. * Level of incidence * Level of resources available (nationally and from donors) Appendix 9A. Cost-Effectiveness of Tetanus * Organization and utilization of health services (particu- Immunization Programs larly preventive and MiCH services) Two of the following four tables examine the cost-effectiveness * Existence of other channels for contact (schools, bride of tetanus immunization programs from the standpoint of registration, TBAs, and so on) epidemiologic data, cost, and effectiveness data. The other two * Immediacy of desired effect (use of campaigns or routine tables use the same data to examine simulation models of immunization) cost-effectiveness. Table 9A- 1. Studies of Cost-Effectiveness of Tetanus Immunization Programs: Epidemiologic Data Number of TT immunizations given Intervention Population Population Three or NNT incidence NNT mortalitY Case- Location (source) year Strategy Target group total Target One Two more Total ratea ratea fatality rate Burkina Faso (de 1987 Fixed centers Women 15-44 - - 6,043 1,957 0 8,000 Champeaux and 13,832 5,928 0 19,760 Martin 1989)' Cameroon (Brenzel 1987) 1987 Mass campaign Women 15-49 - - - 151,415 Ecuador (Shepard and others 1987) 1974-85 Fixed centers Pregnant women - - - 179,765 -- - The Gambia 1980-81 Fixed centers Pregnant women - - 84 - - - 40 in 90 (Robertson and and outreach nationwide percent unimmunized percent others 1985) (1965) (1982) Haiti/ 1967-71 Mass Females ten and 94,000 247,677 213,002 178,327 639,006 64(1967) 60(1966) 85 d Deschappelles older percent (Berggren Marketplace All people ten - -- - - - 9(1972) - 1974a,b)' and older Rally posts Children Haiti (Narcisse 1988 Fixed centers Women 15-44 - - - - 165,713 - 1989) Masscampaign Women 15-44 - - - - - 542,461 - - - Rally posts Women 15-44 - - _- - 165,713 - - Horse teams Women 15-44 - -- - - 127,496 - - - India/Narangwal 1972-73 Campaign Women 15-44 13,000,000 1,820 1,583 0 0 1,583 - 25 - (Kielmann and (87 Vohra 1977)c percent) Indonesia/Central 1979-80 Fixed centers Pregnant women 1,400,000 - - Java (WHO/MPI and outreach two rural 1981 b) sites Indonesia/Central 1979-84 Fixed centers Children Java (Barnum, and outreach Pregnant women Tarantola, and Setiady 1980)f Indonesia/Pidie 1985 Masscampaign Women 10-45 380,000 95,300 83,642 67,962 151,604 Provincialj 85 (WHO/PI 1988; (88 (71 20.9 (1984) percent 1988a; percent) (percent) Pidie (5 clus- Berman and ters): 32.1 others 1991) 15 (1984) Pidie (30 clusters): 4.9 2.6 (1987) (Table continues on the following page.) Table 9A- I (continued) Number of TT immunizations given Intervention Population Population Three or NNT incidence NNT mortality Case- Location (source) year Strategy Target group total Target One Two more Total ratea rate3 fatality rate Indonesia/Aceh 1985 Fixed centers Pregnant women - 358 - 43 - - - 85 (Berman and in four sub- - 462 - 97 - - - - pecentd others 1991)h districts - 732 - 205 - - - 2,435 - 1,948 - - Indonesia/Central 1985 Campaign All women of 577,000 140,000 129,728 125,982 - 255,710 - Seven clusters: 85 Lombok (UNICEF reproductive age (93 (90 28 9.7 percentd 1985)' percent) percent) (1983) Provincial level: 16.7 Senegal/Dakar 1970 Mass campaign Total population 650,000 650,000 - - - 300/650,000 100/650,000 30-40 (Rey and (two times) percent others 1979)' (hypothetical) 1970-73 Masscampaign Total population 650,000 650,000 - - - - - - -- (seven times) (real data) Senegal (Brenzel 1987 Mass campaign Pregnant women - - - - - 71,546 - - - and others 1987) Sudan (Brenzel 1988 Fixed centers Pregnant women - 190,120 40,686 23,575 - 64,261 - - - and others and mobile Pregnant women - 180,177 69,548 45,405 - 114,953 - - - (1987) teams Pregnant women - 119,556 68,864 60,854 - 129,718 - - - Pregnant women - 920,030 332,131 231,848 - 563,979 - - - Thailand 1987 Fixed centers Pregnant women 3,025,000 - - 42-70 - - - -- - (Phonboon and and outreach percent others 1989)k - Not available. a. Per 1,000 live births unless otherwise noted. b. In Yako, EPI used oral polio vaccine and DPT. In Gourcy, EPI used DPTE (DPT with injectable polio). c. Three doses of TT one month apart; minimum five years protection. Numbers of TT immunizati(ons derived from information in cited docLiment. Incidence based on hotspiral admissions. Mortality escimate rom retrospective community survey. Calculations use actual program intervention data. d. WHO estimate of case-fatality rate. e. Assumes lifelong immunity from one dose of high-potency (30Lf/ml) calcium phosphate-adsorbed TT with 80 percent efficacy. f. Uses hypothetical model of incidence. g. Two doses of TT. Coverage figures differ slightly by source. h. Hypothetical incidence data. Targets derived from information reported in document. i. Two doses of TT. Previous vaccination history ignored. j. Hypothetical study of ten-year program of nT2. Low CER due to high proportion of cases in people over one year of age. k. Two doses of -r. Source: See first column of this table. Table 9A-2. Studies of Cost-Effectiveness of Tetanus Immunization Programs: Cost and Cost-Effectiveness Data Total Cost Cost per 7r dose Cost per Tl2 Cost per case preventeda Cost per death averteda Current 1989 Current 1989 Current 1989 Current 1989 Current 1989 Location (source) Subdivision Year dollars dolars dollars dolars dollUars dollars dollars dollars dollars dollars BurkinaFaso (de Gourcy 1987 11,568 12,607 0.59 0.64 1.95 2.13 Champeauxand Yako 7,447 8,116 0.93 1.01 3.81 4.15 Martin 1989)h Cameroon (Brenzel n.a. 1987 186,218 202,955 1.23 1.34 3.82 4.16 1987)' Ecuador (Shepard and n.a. 1985 62,918 72,482 0.35 0.40 - - - - - - others 1987)b The Gambia (Robertson n.a. 1980 125,315 188,463 1.05 1.58 3.48 5.23 136.36 205.07 151.53 227.89 and others 1985)' Haiti/Deschapelles n.a. 1969 67,000 226,346 0.10 0.34 0.31 1.05 28.88 97.57 33.98 114.79 (Berggren 1974a,b)d Haiti (Narcisse 1989)b n.a. 1988 143,926 150,867 0.87 0.91 - - - - - 888,990 931,862 1.63 1.71 - - - - - - 32,846 34,430 0.20 0.21 - - - - 138,890 145,588 1.09 1.14 - - - - - - IndialNarangwal Initial period 1972 253,846 751,750 0.10 0.29 - - - - 32.85 97.27 (Kielmann and Vohra Maintenance n.a. n.a. n.a. n.a. n.a. n.a. n.a. n.a. 0.12 0.34 1977)Y phase Indonesia/CentralJava n.a. 1980 - - 0.62 1.18 2.06 3.92 - - - - (WHO/EPI 1981b)' Indonesia/Central Java n.a. 1978 32,800,000 62,352,475 - - - - 8.70 16.54 135.00 256.63 (Bamum and others 1980)' Indonesia/Pidie n.a. 1985 108,355 124,825 0.71 0.82 1.59 1.84 106.25 122.40 125.00 144.00 (wHo/Epi 1988a; (92-127) (106-146) (108-147) (124-179) Berman and others 1991)g Indonesia/Aceh Tanah Pasir 1985 82 94 - - 1.91 2.20 77.38 89.15 91.04 104.88 (Bermanandothers Samudera 233 268 - - 2.40 2.76 97.90 112.78 115.18 132.69 1991)" Matangkuli 261 301 - - 1.27 1.47 51.98 59.88 61.15 70.45 Jeumpa 1,114 1,284 - - 0.57 0.66 23.34 26.89 27.46 31.64 Indonesia/Central NNT mortality, 28.0 1985 50,000 57,600 0.20 0.23 0.40 0.46 38.43 44.27 45.21 52.09 Lombok (UNICEF NNT mortality, 18.3 - - - - - - - 58.80 67.74 69.18 79.69 1985) NNT mortality, 37.7 - - - - - - - 28.54 32.88 33.58 38.68 NNT mortality, 16.7 - - - - - - - 64.44 74.23 75.81 87.33 (Table continues on the follounng page.) Table 9A-2 (continued) Total Cost Cost per TT dose Cost per r72 Cost per case preventeda Cost per death averteda Current 1989 Current 1989 Current 1989 Current 1989 Current 1989 Location (source) Subdivision Year dollars dollars dollars dollars doUars dollars dollars dollars dollars dolars Senegal/Dakar (Rey and Hypothetical 1970 - - - - - - 175.00 558.45 528.00 1,684.92 others 1979)Y Real data 1971 70,000 214,468 - - - - 269.23 824.88 897.44 2,749.59 Senegal (Brenzel and n.a. 1987 49,786 54,261 0.70 0.76 - - - - - - others 1987)b Sudan(Brenzeland Darfur 1988 108,065 113,277 1.68 1.76 4.58 4.80 - - - - others 1990)b Kordofan 93,776 98,298 0.82 0.86 2.07 2.16 - - - - Capital 164,819 172,768 1.27 1.33 2.71 2.84 - - - - National level 929,577 974,406 1.65 1.73 4.01 4.20 - - - Thailand (Phonboon Hospitals 1985- - - - - 8.90 10.25 - - - and others 1989)S Health centers 86 - - - - 10.30 11.87 - - - - - Not available. n.a. Not applicable. a. NNT only, unless otherwise specified. h. Calculated from cost-effectiveness study of entire EPI. c. Cost of expatriate personnel excluded. Costs based on three contacts required to receive T73 (three of eight contacts 37.5 percent of total costs). d. Total cost is expenditures only. Cost per case and death averted excludes another 630 non-NNT cases prevented. If these were included, the cost per case prevented would fall to $23 ($77 in 1989 dollars), and the cost per death averted would drop to $29 ($99 in 1989 dollars). e. Hypothetical cost data for 2.6 million women age fifteen to forty-four. f. Range in cost per nT = $0.56-$0.76; range in cost per rT2 = $1.43-2.61. Cost data include five years' cost of DPT and TT and cases and deaths from diphtheria, pertussis, and tetanus. g. Cost perTT dose and case prevented derived from infonnation reported in cited documents. Applying upper and lower confidence intervals (17-47) around 32.1/1,000 in five clusters of l984survey in Pidie and 2.3-7.5/1,000 in thirty-cluster survey in 1987, cost per death averted ranges from $45-$211 ($52-$243) in 1989 U.S. dollars. h. Cost per case and death prevented does not take into account the protective effect of TT2 on all deliveries within three years. Cost per case prevented data derived from information reported in cited document. i. Cost excludes administrative salaries at national level. Cases and deaths averted estimated from data reported in cited document, assuming vaccine efficacy of 95 percent; general fertility rate of 110/1 ,000; average duiration of protection of three years. j. Total costs include only vaccine costs. Costs for second study, using real data, were likely incurred over entire four-year period. Authors assume cost year is 1971. Source: See first column of this table. Table 9A-3. Simulation Models of Cost-Effectiveness of Tetanus Immunization Programs: Epidemiologic Data Population Tr coverage Source Location Strategy Target group Total Target TTI TT2 NNT incidence5 NNT mortality" Case-fatality rate Cvjetanovic and - Mass campaign Total population 1,000,000 1,000,000 - - 4 Neonatal - With treatment: others 1972" Repeated mass Total population 1,000,000 1,000,000 - - 18/100,000 - 80 percent general campaign (3x) (adult) 30 percent newbom Continuous Pregnant women 1,000,000 1,000,000 - - - Without treatment: pregnant 90 percent general Continuous preg- Pregnant women 1,000,000 1,000,000 - - - 40 percent newborn nant and repeated and total mass (3x) population Kessel 1984' - Preschool Children - - - - - 10-30 Primary school Children - - - - - Antenatal TF Pregnant women - - - - - - Antenatal Pregnant women - - - - outreach Sharma and India/rural Continuous Pregnant women - - - - Adult: - Adult: 50 percent Sharma 1984a,bd Uttar Pradesh (50-80 percent) 185/100,000 Children 5-10 - - - - Neonatal: 66.7 Rural: 97 percent Children 10-15 68.8 Children < 5 Mass-repeated Women 15-44 every five years All adults (50 percent coverage) Smucker and others India/Uttar Campaign (teams) Women 10-44 3,529,048 882,262 75-95 60-90 - 53 1984' Pradesh every 5 years percent percent (2 districts) Outreach Women 10-44 3,529,048 882,262 75-95 60-90 - 53 (continuous) percent percent - Not available. a. Per 1,000 live births unless otherwise indicated. b. Includes NNT and not NNT. Assumes 50 percent coverage for campaign and 90 percent coverage for continuous strategy. Booster given one year afrer TT2. Thirty-year time horizon. c. Excludes cases and deaths from non-NNT. d. Vaccine efficacy is 90 percent, twenty-five year time horizon, two TT per woman. Boosters given every five years. NNT mortality from sample of 3,267 births. Excludes cases and deaths from non-NNT. Intervention year 1978. Table 9A-4. Simulation Models of Cost-Effectiveness of Tetanus Immunization Prograrm: Cost and Cost-Effectiveness Data Total cost Cost per TT dose Cost per rm2 dose Cost per case prevented' Cost per death averted' Current 1989 Current 1989 Current 1989 Current 1989 Current 1989 Source Location Cost year dollars dollars dollars dollars dollars dollars dollars dollars dollars dollars Cvjetanovic and - 1972 - - - - - - 156.50 463.47 others 1972h 201.00 595.25 -- - 82.70 244.91 - 158.10 468.20 -- - Kessel, 1984' - - - - Uses hypothetical financial units, not actual dollars. Concludes that school-based immunization is most cconomical for long term complenientary strategy. Sharma and India/rural Uttar - - Favors continuous immunization of pregnant women. Mass immunization leads to - - Sharma 1984a,b Pradeshc short-lived declines in cases and deaths. Smucker and India/Uttar 1978 () 1,147,795 2,181,947 0.70 1.34 1.44 2.74 - - 3.55 6.75 others 1984d Pradesh 1,461,477 2,778,443 1.23 2.33 2.76 5.25 - - 6.24 11.86 (2 districts) 431,222 819,749 0.26 0.50 0.54 1.03 - - 1.36 2.59 613,517 1,166,290 0.52 0.98 1.16 2.20 - - 2.56 4.87 a. Cost per case and death avented is for NNT only, unless otherwise specified. h. Nodiscountingofcosts and benefits. c. Gives rankings of cost per prevented death at different NNT mortality rates. Makes assumptions regarding ranking of financial units and strategies. d. First line for each strategy: most cost-effective scenario. Second line for each strategy: least cost-effective scenario costs for twenty-five year period discounted at 12 percent per year; deaths not discounted. Tetanus 217 Notes Bhatia, Shushum. 1989. "Pattems and Causes of Neonatal and Postneonatal Mortality in Rural Bangladesh." Studies in Family Planning 20:136-46. The authors wish to acknowledge their gratitude to Mary Carnell, Pierre Bjerregaard, P., R. Steinglass, D. M. Mutie, G. Kimani, M. Mjomba, and V. Claquin, Nancy Cylke, Michael Favin, Rebecca Fields, Stanley Foster, Artur Orinda. 1993. "Neonatal Tetanus Mortality in Coastal Kenya: A Commu- Galazka, Fran,ois Gasse, Mary Harvey, Diane Hedgecock, Norbert Hirschh- nity Survey." Internationaljournal of Epidemiology 22(1). orn, and Dean Jamison for their assistance and comments on earlier drafts of Black, R. E., D. H. Huber, and G. T. Curlin. 1980. "Reduction of Neonatal this chapter and to the Resources for Child Health Project (John Snow, Inc.) Tetanus by Mass Immunization of Non-pregnant Women: Duration of under a contract (DPE-5982-Z-00-9034-00) with the U.S. Agency for Inter- Protection Provided by One or Two Doses of Aluminum-Adsorbed Tetanus national Development for their support and encouragement. Toxoid." Bulletin of the W'orld Health Organization 58:927-30. 1. The term "economic costs" refers to the value of all resources used, Blanchet, Th6rese. 1989. Perceptions of Childhood Diseases and Attitudes towards including those which were donated and those for which there was no Immunization amongSlum Dwellers, Dhaka, Bangladesh. Arlington, Va.: John additional expenditure (for example, personnel time and amortization of Snow, REACH. vehicles and equipment). Brenzel, L. E. 1987. Cost-Efecbveness of Immumzanon Strategies m the Repubic 2. In the case of Indonesia (Aceh), as well as Indonesia (Pidie), only the of Carneroo. Arlington, Va.E John SInow i REACH. cost per death averted was given in the source cited. The World Bank authors o 1989. The Cost of EPI: A RevieSn of Cost and Cost-Effec.veness Studies have calculated the cost per case averted from this information using the WHO (1979-1987). Arlingtonf Va.: Aohn Snow REACH. estimate of 85 percent case fatality for NNT. 3. All dollar amounts in parentheses in this section are 1989 U.S. dollars. Brenzel, L. E., Pierre Claquin, Ian McLellan, and Sally Stansfield. 1987. Rapid 4. The authors would like to thank Dean Jamison for his assistance with Assessment of Senegal's Acceleration Phase. Evaluation Publication 5. New the calculations in this paragraph. York: UNICEF. 5. The authors would like to thank Dean Jamison for his assistance with Brenzel, L. E., Michael Enright, Rachel Feilden, K. S. Lwin, Siraj Mustafa, A. the calculations in this paragraph. H. Omer, Atahir Osman, Ahmed Babiker, and Allison Percy. 1990. Report on the Cost-Effectveness of the Expanded Program of Immunization in the Republic of Sudan, vols. 1-5. Arlington, Va.: John Snow, REACH. References Budiarso, L. R. 1984. Prospectve Study on Infant and Childhood Mortality in Sukabumi, 1982-83. Jakarta: Govemment of Indonesia, Department of Health. Adeuja, A. 0. G., and B. O. Osuntokun. 1971. "Tetanus in the Adult Nigerians. A Review of 503 Patients." Eas tAfnican Medical]Joumal 48:683-91. Byrchenko, Boris. 1966. "Geogra phic a,lBDistibtin ofthezau a heorld,Hat raia A Revew of503 Ptiens." Est AMan Melcaiourmi48:68-91.1951-60: A Reviewkof the Problem." Bulletin of the W'orldHealth Organization Allman, Suzanne. 1986. "Childbearing and the Training of Traditional Birth 34:71-104. Attendants in Rural Haiti." Medical Anthropology Quarterly 17:4L-43. .1972. "Recent Trends ofTetanus Mortality in the World." In Geoffrey Al-Mukhtar, M. Y. 1987. "Birth Care Practice and Tetanus Neonatorum: A Edsall, ed.. Proceedings of the Third International Conference on Tetanus. Hospital-Based Study in Mosul." Public Health 101:453-56. Pan-American Health Organization Scientific Publication 253. Washing- Amold, R. B.. T. 1. Soewarso, and Albertus Karyadi. 1986. "Mortahlty from ton, D.C. Neonatal Tetanus in Indonesia: Results of Two Surveys." Bulletin of the Chen, L. C M. C. Gesche, Shamsa Abmed, A. 1. Chowdhury, and W. H. World Health Organization 64:25962. Moslev. 1974. "Matemal Mortalitv in Rural Bangladesh." Studies in Family Barnum, H. N., Daniel Tarantola, and 1. F. Setiady. 1980. "Cost-Effectiveness Planning 5:3 34-41. of an Immunization Programme in Indonesia." Bullenn of the World Health Chen, L. C., Mizanur Rahman, and A. M. Sarder. 1980. "Epidemiology and Organization 58:499-503. Causes of Death among Children in a Rural Area of Bangladesh." Interna- Barten, J. J. C. 1969. "Neonatal Tetanus: A Review of 134 Cases." Tropical aid nonaljounsal of Epidemiology 9:2 5-33. Geographical Medicine 21:383-88. Chen, P. C. Y. 1976. "The Traditional Birth Attendant and Neonatal Tetanus: Bastien, J. W. 1988. Cultural Perceptzons of Neonatal Tetanus and Programming The Malavslan Experience." Journal of Tropical Pediatrics and Environmental Implications. Bolivia. Arlington, Va.: John Snow, REACH. Child Health 22:263-64. Bennett, M. J. 1976. "Postabortal and Postpartum Tetanus: A Review of 19 Christensen. P. E. 1972a. "Comments on Epidemiology and Immunology of Cases." South African MedicalJounial 50:513-16. Tetanus in Denmark." In Geoffrey Edsall, ed., Proceedings of the Third Berggren, G. G., Adeline Verly, Nicolle Gamier, Walbourg Peterson, Douglas International Conference on Teranus. Pan-American Health Organization Ewbank, and Wooly Dieudonne. 1983. "Traditional Midwives, Tetanus Im- Scientific Publication 253. Washington, D.C. munization, and Infant Mortality in Rural Haiti." Tropical Doctor 13:79-87. . 1972b. "Side Reactions to Tetanus Toxoid." In Geoffrey Edsall, Berggren, W. L. 1974a. "Administration and Evaluation of Rural Health ed., Proceedings of the Third International Conference on Tetanus. Pan- Services. 1. A Tetanus Control Program in Haiti." Aniencan Jounial of American Health Organization Scientific Publication 253. Washington, Tropical Medicine and Hygiene 23:936-49. D.C. 1974b. "Control of Neonatal Tetanus in Rural Haiti through the Cliff,]. 1985. "Neonatal Tetanus in Maputo, Mozambique. Part 11: Preventc- Utilization of Medical Auxiliaries." Bulletin of the Pan-American Health tive Measures." The Central AfricanJournal of Medicine 31:27-32. Organization 8:24-29. Cottin, J. F. 1987. "Le tetanos en France 1984-85." Bulletin Epidemiologique Berggren, W. L., and G. G. Berggren. 1971. "Changing Incidence of Fatal Hebdomadaire 10:37-39. Tetanus of the Newbom: A Retrospective Study in a Defined Rural Haitian Cutts, F. T.. A. Soares, A. V. Jecque, J. Cliff, S. Kortheek, and S. Colombo. Population."AmericanJournalofTropicalMedicineandHygiene20:491-94. 1990. "The Use of Evaluation to Improve the Expanded Programme on Berman, Peter, John Quinley, Burhannudin Yusuf, Syaifuddin Anwar, Udin Immunization in Mozambique." Bulletin of the World Health Organization Mustaini, A. Azof, and Iskander. 1991. "Maternal Tetanus Immunization 68:199-208. in Aceh Province, Sumatra: The Cost-Effectiveness of Altemative Strate- Cvjetanovic, B. 1972. "Epidemiology of Tetanus Viewed from a Practical gies." Social Science and Medicine 33:185-92. Public Health Angle." In Geoffrev Edsalt, ed., Proceedings of the Third Betts, C. D. 1989. Assessment of Neonatal Tetanus and Its Control in Bolivia. International Conference on Tetanus. Pan-AmeTican Health Organization Arlington, Va.: John Snow, REACH. Scientific Publication 253. Washington, D.C. 218 Robert Steinglass, Logan Brenzet, and Allison Percy Cvjetanovic, B., B. Grab, and K. Uemura. 1978a. "Epideemiological Models of Garenne, Michel, and Olivier Fontaine. 1986. "Assessing Probable Causes of Acute Bacterial Diseases." Bulletin of the World Health Organization 56 Death Using a Standardized Questionnaire: A Study in Rural Senegal." (Supplement 1):1 1-28. Paper presented at the Seminar on Comparative Studies of Mortality and 1978b. "Tetanus: An Endemic Disease without Interhuman Trans- Morbidity: Old and New Approaches to Measurement and Analysis, Uni- mission." Bullentin of the WorldHealthOrganizanon 56 (Supplement 1):29-45. versity of Siena, Intemational Union for the Scientific Study of Population Cvjetanovic, B., B. Grab, K. Uemura, and Boris Bytchenko. 1972. "Epidemi- and Institute of Statistics, July 7-12, Siena, Italy. ological Model of Tetanus and Its Use in the Planning of Immunization Gasse, Franqois. 1990. NeonatalTetanus ELiminainon Initiatve: Progress Reportand Programmes." International Journal of Epidemiology 1:125-37. Recommendation. EPIMCIN-IfNTAGEN/90. 1. Geneva: World Health Organization. de Champeaux. Antoine, and Bruno-Jacques Martin. 1989. Evaluation du Gray, R. H., G. S. Smith, and P. G. Barss. 1990. The Use of Verbal Autopsy Programme Elargi de Vaccination, Province de la Sissili. Centre Muraz, Burkina Methods to Determine Selected Causes of Death in Children. Baltimore: Johns Faso: Organisation de Coordination et de Coop6ration pour la Lutte contre Hopkins University, Institute for Intemational Programs. les Grandes Endemies (ocCGE), Unit6 de Vaccinologie. Griffith, J. A., and D. L. Sachs. 1974. "Tetanus Neonatorum: Social and de Quadros, C. A. 1990. "Defining High Risk Areas and Groups." In NeonaLal Economic Dimensions for Teaching and Program Planning." Indian Pediat- Tetanus Elimination: Issues, and Future Directions. Arlington, Va.: John ncs 11:409-15. Snow, REACH/MotherCare. Haaga, J. G. 1986. "Cost-Effectiveness and Cost-Benefit Analyses of Im- de Silva, A. V. K. V. 1982. "Neonatal Tetanus Problem in Sri Lanka." Pakistan munization Programmes in Developing Countries." In D. Jelliffe and E. F. PediamcJournal 6:214-27. P. Jelliffe, eds., Advances in International Maternal and Child Health, vol. 6. Ebisawa, lsao. 1967. "Mortality of TetanIs in Japan: An Uninitentional Oxford: Clarendon Press. Control Observation." In Leo Eckmann, ed., Principles on Tetanus: Proceed- Heath, C. WXI., Jack Zusman, and ]. L. Sherman. 1964. "Tetanus in the United ings of the Second International Conference on Tetanus. Bern, Switzerland: States, 1950-1960." AmericanJournal of Public Health 54:769-79. Hans Huber. Heinonen, 0. P., Samuel Shapiro, and Dennis Slone. 1977. "Immunizing - 1972. "Mortality and Sex Ratio of Tetanus in Japan." In Geoffrey Agents." In BirthDefectsandDrugs in Pregnancy. Littleton, Mass.: Publishing Edsall, ed., Principles on Tetanius: Proceedings of the Third Internatonal Con- Sciences Group. ference on Tetanus. Pan-American Health Organization Scientific Publica- Henry, Fitzroy, Andre Briend, and Vincent Fauveau. 1990. "Child Survival: tion 253. Washington, D.C. Should the Strategy be Redesigned? Experience from Bangladesh." Health Ebisawa, Isao, and Reiko Homma. 1986. "Tetanus inJapan: Trends of Mortal- Policy and Planning 5:226-34. ity, Case Fatality, and Causes of Death." Japanese Journal of Experimental Herz, B. K., and A. R. Measham. 1987. The Safe Motherhood Initiative: Proposals Medicine 56:155-61. for Action. World Bank Discussion Papers 9. Washington, D.C. Ecuador: Ministerio de Salud Publica and UNICEF. 1987. Curso Basico de Islam, IM. S., M. M. Rahaman, K. M. S. Aziz, Mizanur Rahman, M. H. Munshi, Vigilancia Epidemiologica de Enfermedades Immunoprevenibles. Modulo 4 and Yakub Patwari. 1982. "Infant Mortality in Rural Bangladesh: An p. 40: Quito. Analysis of Causes During Neonatal and Post-neonatal Periods." Joumnal of Evans, J. R., K. L. Hall, andJeremy Warford. 1981. "Shattuck Lecture-Health Tropical Pediatrics 28:294-98. Care in the Developing World: Problems of Scarcity and Choice." Newl Jones, T. S. 1983. "The Use of Tetanus Toxoid for the Prevention of Neonatal England Journal of Medicine 305:1117-27. Tetanus in Developing Countries." In N. A. Halsey and C. A. de Quadros, Fauveau, Vincent, M. Mamdani, Robert Steinglass, and Marjorie Koblinsky. es., Recent Advances in Immunization: A Bibliographic Review. PAHO (Pan- 1993. "Matemal Tetanius: Magnitide Epidemiology and Potential Control American Health Organization) Scientific Publication 451. Washington, Measures." Internationaljournal of Gynecology and Obstemcs 40(1):3-12. D.C. Fenner, Frank, D. A. Henderson, Isao Arita, Zdenek Jezek, and 1. D. Ladnyi. Jordan, Brigitte. 1989. "Cosmopolitical Obstetrics: Some Insights from the 1988. Smallpox and Its Eradication. Geneva: World Health Organization. Training of Traditional Midwives." Social Science and Medicine 28:925-44. Foege, W. H., and D. A. Henderson. 1986. "Management Priorities in Priniary Kessel, Elton. 1984. "Strategies for the Control of Neonatal Tetanus."Journal Health Care." Reviews of Infectious Diseases 8:313-21. of Tropical Pediatsics 30:145-49. Foster, S. 0.1984. "Immunizable and Respiratory Diseases and Child Mortal- Kielnsann, A. A., and S. R. Vohra. 1977. "Control of Tetanus Neonatorum in ity." In W. H. Mosley and L. C. Chen, eds., Child Survival: Strategies for Rural Communities-Immunization Effects of High-Dose Calcium Phos- Research, special supplement, PopulationandDevelopment Review 10:1 19-40. phate-Adsorbed Tetanus Toxoid." Indian Journal of Medical Research 66: Galazka, Artur. 1982. "Tetanus Toxoid-Nature and Action." Pakistan Pedi- 909-16. arricJournal 6:120-43. LaForce, F. M., L. S. Young, and J. V. Bennett. 1969. "Tetanus in the United - 1983. Immunization of Pregnant Women against Tetanus. EPtGEN/83/5. States (1965-1966): Epidemiologic and Clinical Features." New England Geneva: World Health Organization. Journal of Medicine 280:569-74. Galazka, Artur, and Robert Cook. 1985. "Neonatal Tetanus Today and Lanasari, Rosaline, and Zeil Rosenberg. 1989. "TetanusToxoid Immunization Tomorrow." In Guiseppe Nistico, P. Mastroeni, and M. Pitzurra, eds., of Prospective Brides in Central Java, Indonesia." Health Policy and Planning Proceedings of the Seventh International Conference on Tetanus. Rome: 4:235-38. Gangemi. Leroy, Odile, and Michel Garenne. 1991. "Risk Factors of Neonatal Tetanus Galazka, Artur, Francois Gasse, and R. H. Henderson. 1989. "Neonatal in Senegal." Internanonal]ournalof Epidemiology 20:521-26. Tetanus and the Global Expanded Programme on Immunization." In Elton Maine, Deborah, Allan Rosenfield, Marilyn Wallace, A. M. Kimball, B. E. Kessel and A. K. Awan, eds., MaternalandChildCare in DevelopingCountries: Kwast, Emile Papiemrik, and Sharon White. 1987. "Prevention of Matemal Assessment, Promotion, Implementation. Proceedings of the Third International Deaths in Developing Countries: Program Options and Practical Consider- Congress for Maternal and Neonatal Health. Thun, Switzerland: Ott. ations." Paper presented at International Safe Motherhood Conference, Galazka, Artur, and B. Kardymowicz. 1989. "Tetanus Incidence and Immunity 10-13 February, Nairobi, Kenya. in Poland." EuropeanJournalof Epidemiology 5:474-79. Miller, J. K. 1972. "The Prevention of Neonatal Tetanus by Matemal Im- Galazka, Artur, and George Stroh. 1986. Neonatal Tetanus: Guidelnes on the munization." EnvironmentalChild Healh 18:160467. Community-Based Survey on Neonatal Tetanus Mortality. WHO/EPMI,cEN/86/8. Mosley, W. H. 1984. "Will Primary Health Care Reduce Infant and Child Geneva: World Health Organization. Mortality? A Critique of Some Current Strategies, with Special Reference Tetanus 219 to Africa and Asia." In J. C. Caldwell and Gigi Santowi, eds., Selected Robertson, R. L., S. 0. Foster, H. F. Hull, and P. J. Williams. 1985. "Cost- Readings in the Cultural, Social, and Behavioral Determinants of Health. Health Effectiveness of Immunization in The Gambia."Journal of Tropical Medicine Transition Series no. 1. Canberra: Highland Press for Health Transition and Hygiene 88:343-51. Center of the Australian National University. Rosenfield, Allan, and Deborah Maine. 1985. "Matemal Mortality-A Ne- Murray, C. J. L., Gonghaan Yang, and Xinjian Qiao. 1992. "Adult Mortality: glected Tragedy: Where is the M in MCH?" Lancet 2:83-85. Levels, Pattems, and Causes." In R. G. A. Feachem, Tord Kjellstrom, C. J. Rosmini, F., M. Wirn, G. Gentili, C. Colotti, M. Rossini Ricci, E. Franco, l. L. Murray, Mead OCver, and M. A. Phillips, eds., The Health of Adults iii the Terzn, and P. Pasquini. 1987. "Year of Birth, Sex, and Residence as 'Deter- Developing World. Oxford: Oxford University Press. minants' of Tetanus Incidence and Immunity in Italy." European Journal of Narcisse, Maryse. 1989. Analyse du coat, coat-efficaciti des srrategses du Pro- Epidemiology 3:377-80. grainme Elargi de Vaccination eni Haiti. Port-au-Prince, Haiti: Ministere de la Ross, D. A. 1986a. "Does Training TBAs Prevent Neonatal Tetanus?" Health Sante Publique et de la Population, Bureau de Coordination du Programme Policy and Planning 1:89-98. Elargi de Vaccination. - 1986b. "The Trained Traditional Birth Attendant and Neonatal Newell, K. W., A. D. Lehmann, D. R. Leblanc, and N. G. Osorio. 1966. "The Tetanus." In Amelia Mangay-Maglacas and John Simons, eds., The Potential Use of Toxoid for the Prevention of Tetanus Neonatorum: Final Report of of the Tradinonal Birth Attendant. Geneva: World Health Organization. a Double-Blind Controlled Field Trial." Bulletin of the World Health Organi- Royston, Erica, and Sue Armstrong, eds. 1989. Preventing Maternal Deaths. zation 35:863-71. Geneva: World Health Organization. Nichter, Mark. 1990. "Vaccinations in South Asia: False Expectations and Schofield,Frank. 1986. "Selective PrimaryHealthCare: StrategiesforControl Commanding Metaphors." In Jeannine Coreil and ]. D. Mull, eds., Anthro- of Disease in the Developing World. XXII. Tetanus: A Preventable Prob- pology and Prinmary Health Care. Boulder, Colo.: Westview PTess. lem." Reviews of Infectous Diseases 8:144-56. Patel, J. C., and B. C. Mehta. 1975. "Tetanus: A Study of 8,697 Cases." In Schofield,F.D.,V.M.Tucker,and G.R.Westbrook. 1961."NeonatalTetanus Proceedings of the Fourth Internaional Conference on Tetanus. Lyons: Fonda- in New Guinea: Effect of Active Immunization in Pregnancy." Bntish tion Merieux. MedicalJournal 2:785-89. Phonboon, K., D. S. Shepard, S. Ramaboot, P. Kunasol, and S. Preuksaraj. Senecal, J. 1970. "Tetanus." In D. B. Jelliffe and H. C. Trowell, eds., Diseases 1989. "The Thai Expanded Programme on Immunization: Role of Im- of Children in the Subrropics and Tropics, 2d ed. London: Edward Amold. munization Sessions and Their Effectiveness." Bulletin of the World Health Sharnma, Monica, and P. D. Sharma. 1984. "The Dynamics of Tetanus Epide- Organization 67:181-88. miology and Control in Rural Uttar Pradesh Studied through Computer Pillsbury, Barbara. 1989. Immunizanion: The Behavioral Issues: Behavioral Issues Siniulation." Social Science and Medicine 18:303-13. in Child Survival Programs: A Synthesis of the Literature with Recommendations Sharma, P. D., and Monica Sharma. 1982. "An Approach to Health Planning for Project Design and Implementation. Monograph 3. Washington, D.C.: in IndiaviaComputerSimulationofEpidemiological Models: ACase Study Intemational Health and Development Associates. of Tetanus." Healthand Population: Perspectives and Issues 5:139-67. Rahman, Makhlisur, L. C. Chen,J.Chakraborty, M. Yunus, A. I. Chowdhury, . 1984. "Impact of Altemate Immunisation Strategies on Tetanus A. M. Sarder, Shushum Bhatia, and G. T. Curlin. 1982. "Use ofTetanus Neonatorum in India." Indian Pediamcs 21:839-52. Toxoid for the Prevention of Neonatal Tetanus. 1. Reduction of Neo- Shepard, D. S., R. L. Robertson, C. S. M. Cameron, Pedro Satumo, Marjorie natal Mortality by Immunization of Non-pregnant and Pregnant Women Pollack, and Jacques Manceau. 1987. The Cost-Effectiveness of Immunization in Rural Bangladesh." Bulletin of the World Health Organization 60: Strategies in Ecuador. Arlington, Va.: John Snow, REACH. 26 1-6 7. Simonsen, Ole. 1989. "Vaccination against Tetanus and Diphtheria: Evalua- Rahman, Makhlisur, L. C. Chen, J. Chakraborty, M. Yunus, A. S. G. tion oflmmunity in the Danish Population, Guidelines for Revaccinations, Faruque. and A. I. Chowdhury. 1982. "Use of Tetanus Toxoid for the and Methods for Control ofVaccination Programs." Danish Medical Bulletin Prevention of Neonatal Tetanus. 2. Immunization Acceptance among 36:24-47. Pregnant Women in Rural Bangladesh." Bulletin of the World Health Simonsen, Ole, M. W. Bentzon, Keld Kjeldsen, H. A. Venborg, and Iver Organization 60:269-77. Heron. 1987. "Evaluation of Vaccination Requirements to Secure Contin- Rahman, Shafiqur. 1982. "The Effect of Traditional Birth Attendants and uous Antitoxin Immunity to Tetanus." Vaccine 2:115-22. Tetanus Toxoid in Reduction of Neonatal Mortality." Journal of Tropical Simonsen, Ole, A. V. Bloch, and Iver Heron. 1987. "Epidemiology ofTetanus Pediatrics 28:163-65. in Denmark 1920-1962." Scandinavian Journal of Infectious Diseases 9: REACH (Resources for Child Health Project). 1989. EPI Essentials: A Guide for 437-44. Program Officers. Arlington, Va.: John Snow. Smucker, C. M., G. B. Simmons, Stan Bemstein, and B. D. Misra. 1980. Rey, Michel. 1982. "Prevention of Neonatal Tetanus by Immunization." "Neo-natal Mortality in South Asia: The Special Role of Tetanus." Popu- Pakistan PediatricJournal 6:103-19. laton Studies 34:321-35. Rey, Michel, 1. Diop-Mar, C. Lafaix, M. Gu6rin, and B. Schaff. 1968. "Le Smucker, C. M., J. M. Swint, and G. B. Simmons. 1984. "Prevention of t6ranos a Dakar, probleme de sant6 publique." Bulletin etrMnmoires de FaculteW Neonatal Tetanus in India: A Prospective Cost-Effectiveness Analysis." des Mddicins Pharnocologmques 16:77-87. Journal ofTropical Pediatrics 30:227-36. Rey, Michel, I. Diop-Mar, and D. Robert. 1981. "Treatment of Tetanus." In Sokal, D. C., G. Imboua-Bogui, G. Soga, C. Emmou, and T. S. Jones. 1988. Ricardo Veronesi, ed., Tetanus: Important New Concepts. Amsterdam: Ex- "Mortality from Neonatal Tetanus in Rural Cote d'lvoire." Bulletin of the cerpta Medica. World Health Organization 66:69-76. Rey, Michel., P. Guillaumont, and B. Majnoni d'Intignano. 1979. "Benefits Solter, S. L., A. A. Hasibuan, and Burhanuddin Yusuf. 1986. "An Epidemio- versus Risk Factors in Tetanus." Developments in Biological Standardization logical Approach to Health Planning and Problem-Solving in Indonesia." 43:15-23. Health Policy and Planning 1:99-108. Rey, Michel, and E. Tikhomirov. 1989. "Non-neonaIt Tetanus over the Sow, Abdourahmane. 1989. "Tetanus inDakar,Senegal 1985-1986." Personal World." In Giuseppe Nistico, Bemard Bizzini, Boris Bytchenko, and Rene communication. Cited in Rey and Tikhomirov 1989. Triau, eds., Proceedings of the Eighth Internatonal Conference on Tetanus. Stanfield, J. P., and Artur Galazka. 1984. "Neonatal Tetanus in the World Rome: Pythagora Press. Today." Bulletn of the World Health Organization 62:647-69. 220 Robert SteingLass, Logan Brenzel, and Allison Percv Steinglass, Robert. 1988. Monitoring Tetanus Toxoid Coverage through Routine WHO (World Health Organization). 1977. International Classification of Dis- Reporting: Presentand ProposedMethods. Arlington, Va.:John Snow, REACH. eases. Manual of the International Statistcal Classification of Diseases, Injunes, - 1989. "The Control of Neonatal Tetanus." Mothers and Children, and Causes of Death. Vols. I and 2. Ninth revision. Geneva. Bulletin on Infant Feeding and Maternal Nurtion 9:4-5. . 1982. Prevention of Neonatal Tetanus. Eastern Mediterranean Re- 1990. Monitonng Tetanus Toxoid Coverage and Activity Using Service gional Office Technical Publication 7. Alexandria, Egypt. Statistics. In Neonatal Tetanus Elimination: Issues and Future Directions. . 1986. The Expanded Programme on Immunization in South-East Asia. Arlington, Va.: John Snow. REACH/MotherCare. Regional Health Papers 12. New Delhi: South-East Asia Regional Office. Steinglass, Robert, D. M. Mutie, Geoffrey Kimani, Mary Mjomba, Vincent WHO/EPI (World Health Organization/Expanded Programme on Immuniza- Orinda, and Peter Bjerregaard. 1989. "Neonatal Tetanus Mortality in Kilifi tion). 1981a. "Disease Incidence and Immunization Coverage, Poland District, Kenya: Results of a Community Survey, 1989." Forthcoming in 1950-1980." Weekly Epzdemiological Record 56:281-84. Intemnationaljouirnal of Epidemiology. International Jounal of Epidemiology.-. 1981 b. "An Economnic Appraisal: Indonesia." Weekly Epidemiological Stroh, George, U Aye Kyu, U Thaung, and U Kyaw Lwin. 1987. "Measurement Record 56:99-101. of Mortality from Neonatal Tetanus in Burma." Bulletin of the World Heakh Organization 65:309-16. . 1982. "The Use of Survey Data to Supplement Disease Surveillance." Suleiman, Omar. 1982. "Mortality from Tetanus Neonatorum in Punjab Weekly Epidemiological Record 57:361-62. (Pakistan)." Pakistan PediatricJournat 6: 152-83. . 1983a. "Adverse Reactions to Immunization, German Democratic Suri, J. C. 1967. "The Problem of Tetanus in India." In Leo Eckniann, ed., epubhc." Weeds Epidemiologmcal Record 58:62-64. Principles on Tetanus: Proceedings of an Irnternational Conference on Tetanus. . 1983b. "Disease Incidence and Immunization Coverage, Hungary." Bern, Switzerland: Huber. Weekly Epidemiological Record 58:77-80. Sutter, R. W., S. L. Cochi, E. W. Brink, and B. 1. Sirotkin. 1990. "Assessment . 1983c. "Neonatal Tetanus Mortality Surveys, Indonesia." Weekl' of Vital Statistics and Surveillance Data for Monitoring Tetanus Mortality, Epidemiological Record 58:56-57. United States, 1979-1984." AmericanJournal of Epidemiology 131:132-42. _ . 1986. Prevention of Neonatal Tetanus through Immunization. WHOrrPI' Tateno, Isao, Shigeto Suzuki, and Osama Kitamoto. 1961. "Epidemiology of GEN/86.9. Geneva. Tetanus in Japan." Japanese Journal of Experimental Medicine 31:365-80. -. 1987a. Issues in Neonatal Tetanus Control. EPuIAG/87/WP. I1. Geneva. UNICEF (United Nations Children's Fund)/Jakarta. 1985. "Tetanus Toxoid . 1987b. "Neonatal Tetanus Mortality Surveys, Egypt." Weekly Epide- Campaign in West Nusa Tenggara, Indonesia." Assignment Children 69/72: miological Record 62:33 2-3 5. 369-80. 1987c. "Neonatal Tetanus Mortality Surveys, Iran." Weekly Epidemi- Vakil, B. J., and N. J. Dalal. 1975. "Ways of Inoculation, Portals of Entry of ological Record 62:38"40. Tetanus, and Risk Evaluation of a Wound." In Fourth International Confer- 1988a. "Impact of Tetanus Toxoid Immunization, Indonesia." Weeklv ence on Tetanus. Lyons: Fondation Merieux. Veronesi, Ricardo, and R. Focaccia. 1981. "The Clinical Picture." In Ricardo - 1988b. Neonatal Tetanus: Immunize All Women of Childbearing Age. Veronesi, ed., Tetanus: Important New' Concepts. Amsterdam: Excerpta Geneva. Medica. Walsh, J. A., and K. S. Warren. 1979. "Selective Primarv Health Care: An --. 1989a. Progress report: March 1989-October 1989. Research and Interim Strategy for Disease Control in Developing Countries." New Eng- Development Group Meeting. EPI/RD/89 Wi'. 4. Oct. Geneva. Iand Journal of Medicine 301 :967-74. . 1989b. A Vision for the World: Global Elimination of Neonatal Tetanus White, W. G., G. M. Bames, E. Barker, D. Gall, P. A. Knight, A. H. Griffith, by the Year 1995: Plan of Action. EPi/GAG/89/WP.9. Geneva. R. M. Morris-Owen, and J. W. G. Smith. 1973. "Reactions to Tetanus . 1990. "Stability of Vaccines." Weekly Epidemiological Record 65:233- Toxoid."JournalofHygiene 71:283-97. 35. 10 Rheumatic Heart Disease Catherine Michaud, Jorge Trejo-Gutierrez, Carlos Cruz, and Thomas A. Pearson Rheumatic heart disease (RHD) is the most common form of Interventions to prevent RHD can be directed at different heart disease among children and young adults in many devel- points in the chain of events leading from GASP to intractable oping countries. It affects more than 4 million people world- cardiac failure. Primary prevention targets cases of pharyngitis wide and causes approximately 90,000 deaths each year. This and consists of a single injection of benzathine penicillin, heavy toll could be reduced, because RHD is always triggered by which effectively treats GASP and prevents the occurrence of a controllable infectious agent: group A streptococci. RF. Secondary prevention targets RF cases and consists in The chain of events leading to RHD is complex and evolves monthly injections of benzathine penicillin for several years to over several years. It starts with acute group A streptococcal prevent recurrences ofGASP that might trigger new bouts of RF. pharyngitis (GASP), or strep throat, which is extremely com- Tertiary prophylaxis targets severe cases of RHD. It interrupts mon among school-age children. If they are not treated effec- the progression to intractable cardiac failure by means of tively, 3 percent of GAsP episodes lead to rheumatic fever (RF), surgical repair or replacement of damaged heart valves. A a disease that damages the heart, particularly the heart valves. vaccine that could prevent the occurrence of GASP is not yet Heart valve injuries occur because group A streptococci pre- available. cipitate an immunological assault against the body's own heart The cost-effectiveness of various strategies is determined by valves and some other tissues. the number of total cases that have to be treated to prevent Rheumatic fever lasts for only several weeks but often leaves one death and is limited by difficulties in finding the cases. permanent scars which cause rheumatic heart disease. Dam- Primary prevention in developing countries is impractical aged valves no longer open or close properly, thus disrupting when the diagnosis of GASP cannot be confirmed. Secondary normal blood flow. These hemodynamic changes overload the prevention therefore remains the cornerstone of RHD preven- heart and lead to progressive cardiac insufficiency and often to tion. Tertiary prophylaxis is a cost-effective alternative in spite premature death. Finally, the severity of valve injuries tends to of the high cost of surgery because it can be very specifically increase over time, because RF recurs in 75 percent of children directed to a small number of cases. who had a first attack of RF when they suffer new episodes of In industrial countries the incidence of RF and prevalence of streptococcal pharyngitis (figure 10-1). In the absence of any RHD decreased during the past decades as socioeconomic con- intervention, among 1,000 children with pharyngitis, 200 ditions improved and penicillin treatment became available. would have pharyngitis caused by group A streptococci, 6 Fewer than 1,000 cases of RHD, and almost no deaths, now would suffer an initial RF attack, 5 would later have recurrences occur each year in industrial countries. The challenge is to of RF, and 2 would die from intractable cardiac failure. achieve a similar reduction in developing countries, despite Figure 10-1. Chain of Events Leading to Rheumatic Heart Disease Showing Percentage That Will Develop the Next Stage Percentage that will develop the next stage GASP 3 -+ RF initial attack + GASP recurrences 75 + RF recurrences 70 -+' RHD Source: Authors, from epidemiologic data. 222 Catherine Michaud, Jorge Trejo-Gutierrez, Carlos Cruz, and Thomas A. Pearson scarce resources, slow socioeconomic growth, and limited ac- In the Developing World cess to health care. In this chapter we review the epidemiology, pathogenesis, In sharp contrast to the situation prevailing in industrial and clinical manifestations of RF and ROID; discuss possible countries, RF and RHD are quite common in developing coun- interventions and analyze their cost-effectiveness; and, finally, tries, where the incidence of RF ranges from 6.9 to 100 cases outline areas which require further research. per 100,000 people (table 10-1), and the prevalence of RHD ranges from 1.0 to 18.6 cases per 1,000 (table 10-2). These rates Background are similar to rates that were common in Western countries at the beginning of this century (WHO 1988). Their broad range The distribution of RF and RHD is quite different in industrial illustrates disparities existing among different geographic re- and developing countries, and has evolved over time. The gions, among various ethnic and socioeconomic groups, and distribution of the disease also varies within each country between urban and rural populations. For instance, several between rural and urban areas and among different population minority groups in the Pacific Islands still suffer from a very groups. The major determinants of the disease are socioeco- high prevalence ofRHD, rangingbetween 7.6 and 18.6 cases per nomic conditions, access to medical care, and the changing 1,000 persons, even though the prevalence of RHD is quite low virulence of group A streptococci. in the general population. In New Zealand, most cases of RHD were reported from the northern part of North Island, where In Industrial Countries Maori and Polynesian populations are concentrated (Neutze 1988a). In Australia, RF and RHD persist only among aboriginal Rheumatic fever and rheumatic heart disease have virtually communities living in the Northern Territory (MacDonald disappeared in industrial countries. The average annual RF and Walker 1989). The differences observed between various incidence is now below 0.5 per 100,000, and RHD prevalence ethnicgroupsareprobablyattributabletotheirlowersocioeco- is less than 0.05 per 1000. nomic conditions, rather than to a particular genetic suscepti- The progressive decline in RF and RHD started simultaneously bility to the disease. in several Westem countries at the beginning of the twentieth The prevalence of RHD is determined worldwide by the century, several decades before penicillin treatment became level of socioeconomic development and by access to health available. With the introduction of penicillin, at the end of care, but in addition, it differs strikingly between urban and World War 11, the decline accelerated rapidly among higher rural settings. The prevalence of RHD tends to be higher socioeconomic groups but lagged behind among the poor. among the urban poor than among rural poor. It ranges from Inadequate housing and overcrowding favor the spread of 8.5 to 11 cases per 1,000 individuals in the largest cities of streptococcal infection from person to person, undemutrition Africa, Asia, and Latin America, whereas in rural areas, RHD impairs the immune response, and streptococci remain longer prevalence does not exceed 3.5 per 1,000 on average (Strasser in the throat in the absence of penicillin treatment. The 1985). number of rheumatic fever and rheumatic heart disease cases have nevertheless slowly decreased even in communities in which the use of penicillin was not widespread, perhaps as a Table 10c1. Annual Incidence of Rheusatic Fever consequence of lower virulence of group A streptococci. These n Selected Areas observations indicate that penicillin treatment was only one Incidence Age group among several factors that prompted the decline of RF and RHD Location Year (per 100,000) (years) (Gordis 1985; Gordis, Lilienfeld, and Rodriguez 1969a and England and Wales 1963 4.7 1-14 1969b; Massell and others 1988). Baltimore (United The frequency of occurrence of pharyngitis has also de- States) 1964 15.3 5-19 creased in industrial countries, but not the proportion of pha- Denmark 1970 10.7 All ryngitis caused by group A streptococci. Streptococci cause 15 Singapore 1971 92 All ot Cyprus ~~1972 27-43 All to 30 percent of all pharyngitis, and this proportion does not HypngKong 1972 23 Al vary significantly within and between nations worldwide Czechoslovakia 1972 8.5 <15 (Markowitz 1985). The potential to develop RF and RHD there- Iran, Islamic Rep. of 1973 59-100 All fore persists in industrial countries, as was evidenced by out- Kuwait 1983 19.6 <14 breaks of RF that have occurred in several small communities Auckland (New 1984 70.7 <15 since 1986 in the United States and northern Italy. In each Zealand)Y b instance, RF followed a similar new epidemiologic pattern: RF Hawaii (United States 1976-80 14.4 4-18 Salt Lake County cases were concentrated among middle-class families with (UnitedStates) 1985 18.1 5-17 ready access to medical care. The resurgence of RF, following this new pattern, has raised great concern and requires close a. Maori population. monitoring (Hosier and others 1987; Kaplan and Markowitz 272; Icidence(per l00,0hJ) by ethnic group was samoan,F96.5; Hawaiian, 1988; Veasy and others 1987; Wald and others 1987). Source: Adapted from Majeed and others 1987. Rheumatic Heart Disease 223 Table 10-2. Prevalence of Rheumatic Heart cases include infections of the skin and throat (GASP), forms of Disease in School-age Children in Different Areas pneumonia, and a recently identified disease resembling toxic Prevalence shock. Scarlet fever is the result of infection by streptococci Locaton Year (per 1,000) that elaborate an erythrogenic toxin against which the host has no antibodies. Algeria 1970 15.0 The exact nature of the interaction between group A strep- Nigeria 1970 0-3.0 tococci and the human host leading to RF is not fully under- Egypt 1973 10.0 stood. Several hypotheses were developed to explain how Morocco 1973 9.9 streptococci might damage the heart: Soweto, South Africa 1975 6.9 C6te d'lvoire 1985 1.9 * Direct tissue invasion by group A streptococci Latin America * Toxic effects of streptococcal products, particularly Brazil 1968-70s 1.6-6.8 streptolysins S or 0, which are known to be capable of Montevideo, Uruguay 1970 1.0 inducing tissue injury 1985 10.0 * Reactions like serum sickness, mediated by antigen- La Paz, Bolivia 1973 17.0 antibody complexes Mexico City, Mexico 1977 8.5 San Juan, Puerto Rico 1980 1.6 * Autoimmune phenomena induced by similarity or iden- Caracas, Venezuela 1985 10.0 tity between certain streptococcal antigens and human tis- Porto Allegre, Brazil 1985 10.0 sue components Sao Paulo, Brazil 1985 10.0 Asia Even though direct tissue invasion and toxic effects of Tokyo, Japan 1966 0.3 streptolysins may play a role, most attention has recently been Taiwan, Republic of China 1970 1.4 directed to the immunological process involved. Almost every India 1970s 6.0-11 part in the immune system, cellular and humoral, is involved Pakistan 1970s 1.8-11 in RF. Group A streptococci selectively amplify or downregul- Thailand 1974 1.2-2.1 ate various immune pathways which are likely to be critical in Mona d.4-2. the subsequent development of RF after an episode of strepto- NewgDelhiIndia 9n8. 11.0 coccal pharyngitis (Caims 1988; Goldstein 1967). Recent research has unraveled the key role played by M PTscific proteins in the immunological process (Fischetti 1991). M Rarotonga, Cook Islands 1982 18.6 proteins cover the surface of the bacterial cell wall and appear French Polynesia 1988 11.2 as hairlike projections. They give a streptococcus the ability to Waikito, New Zealanda n.d. 7.6 (Maori) resist ingestion by white blood cells. To overcome the effect of 1.0 (non-Maori) M proteins, the human host produces antibodies directed nd. No dare available, against M proteins. These antibodies neutralize the protective n.d. No date available.caaiyothMprtianalotesrpoocutob a. Subjects in this study were age five to twenty-nine years. capacity of the M protein and allow the streptococcus to be Source: Adapted from WHO, 1988. engulfed and destroyed by white blood cells. Group A strepto- cocci have further increased their ability to evade the immune Trends observed during the past several decades in develop- system through antigenic variation. There are more than ing countries vary. The prevalence of RHD is increasing in the eighty different serotypes, or varieties of M proteins, and largest cities undergoing rapid growth, particularly in slum laboratory tests suggest that antibodies against one serotype do areas. In India the prevalence of RHD has increased with rapid not offer protection against others. The low incidence of GASP urbanization, and now reaches 6 cases per 1,000 people on observed in adulthood may be due either to an undefined, average for the whole country (WHO 1980a). The prevalence age-related factor or to a broad immunity that individuals of RHD seems to be much lower in China than it is in India. acquire through contact with streptococci as children. Richard Bumgamer estimated in 1990 that there were 410,000 Different serotypes of group A streptococci cause the two cases of RHD in China, which corresponds to an average RHD major nonsuppurative sequelae of GASP: RF and glomerulo- prevalence of 1.4 per 1,000 for the whole country (Bumgamer, nephritis. Serotypes of group A streptococci isolated from personal communication). In middle-income economies-for children with RF and their families differ from those isolated example, Thailand-the prevalence of RHD is decreasing. from children with poststreptococcal nephritis and their fam- ilies (Majeed and others 1987). This chapter deals only with Pathogenesis RF. Rheumatic fever occurs when antibodies developed against Group A streptococci are very common human pathogens. M proteins cross-react with the host's own heart tissues because Between 20 and 30 million cases of group A streptococcal some fragments (epitopes) of the M protein closely resemble infections occur each year in the United States alone. These fragments of valve glycoproteins and valve fibroblasts, leading 224 Catherine Michaud, Jorge Trejo-Guterrez, Carlos Cruz, and Thomas A. Pearson Table 10-3. Frequency of Major Manifestations rapid random, nonrhythmic movements of chorea most often in Initial Attacks of Rheumatic Fever in Prospective affect the muscles of the face and arms. Subcutaneous nodules Studies are round, hard, painless swellings, which occur in 5 to 10 (percent) percent of rheumatic patients. Erythema marginatum is the United United characteristic skin rash of RF and occurs in fewer than 5 percent Manifestation Kuwait India Kingdom States of patients. Symptoms observed during the initial attack of RF tend to be different from those observed during subsequent Polyarthritis 79 67 85 76 episodes of RF (Nelson 1983). Chorea 8 20 13 8 Nodules 0.5 3 ** I Initial Rheumatic Fever Attack Erythema marginatum 0.5 2 ** 4 During the initial attack of RF it is common to observe migrat- ** Negligible amount. ing arthritis with painful, red, hot swellings of one or several Source: Adapted from Markowitz 1988. large joints-knees, elbows, wrists, and ankles. Arthritis may last up to three months; it causes significant morbidity but to permanent valvular damage. The exact role of such cross- ultimately leaves no sequelae (WHO 1988). Arthritis tends to reactive antibodies in the genesis of RHD, however, is not yet be less frequent in subsequent episodes of RF. The frequency of understood (Fischetti 1991). carditis, on the contrary, tends to increase with recurrent Other researchers have identified genetic markers-B lym- episodes of RF. phocyte alloantigens-which might determine greater suscep- Carditis often causes few symptoms, and it is rarely life tibility to RF (Patarroyo and others 1979). For instance, RF threatening. The onset of carditis may be asymptomatic-so- happened more often in monozygotic than dizygotic twins, but called silent carditis-or insidious, with only a few vague the numbers studied were very small (Taranta and others symptoms such as lethargy, poor appetite, and chronic respira- 1959). These findings still require confirmation in various tory infections. In this instance, patients seek health care only ethnic groups. much later, when they suffer from shortness of breath There is no single determinant for RF. Several risk factors (Markowitz 1988). pertaining to the environment, the host, and the agent interact and ultimately cause RF. Only a few of the mechanisms in- Recurrent RF Episodes volved are fully understood. For instance, the observation that the sera of patients contained antibodies against heart tissues After the initial attack, RF is characterized by frequent recur- dates back half a century, but the exact amino-acid sequence rences of the disease after a varying number of intercurrent of heart cross-reactive epitopes was identified in M proteins years of freedom from symptoms. The risk for RF recurrence, only recently. following subsequent GASP, increases dramatically from 3 per- cent to 75 percent (Denny 1987). Recurrent bouts of carditis Clinical Aspects often cause long-term sequelae because when the inflamma- tory process heals in the heart, it leaves scars on the valves, Rheumatic fever occurs in 3 percent of all GASP episodes which impair their normal opening or closure. (Denny 1987). No direct relationship exists between the se- verity of GASP symptoms-fever, sore throat-and the devel- opment of RF ten to thirty-five days later. Furthermore, GASP Table 10-4. Outcomes of 10,000 Hypothetical causes no symptoms in 30 to 50 percent of all children who Cases of Rheumatic Fever without Treatment later develop RF (Maharaj and others 1987). Initial attack Inital attack The clinical findings vary greatly and are determined by the without with site of involvement, the severity of the attack, and the stage at recurrence recurrence Total which the patient is first examined. The most common symp- Outcome Number Percent Number Percent Number Percent toms of RF are redness and pain of the joints (arthritis) and a Total cases 2,500 25 7,500 75 10,000 100 typical heart murmur (carditis). The clinical diagnosis of RF may be difficult because the severity and the combination of Recover 1,475 59 1,500 20 2,975 30 RF symptoms vary among different individuals. The onset is Mild 369 15 840 11 1,209 12 usually acute when arthritis is the presenting manifestation Moderate/ and more gradual when carditis is the initial clinical feature. severe cardio- When carditis is the sole clinical manifestation, it can be megaly 431 17 1,920 26 2,351 24 difficult to determine when the attack began. Congestive Arthritis and carditis are sometimes accompanied by cho- heart failure rea, subcutaneous nodules, or erythema marginatum. Chorea leading'to_death _22 _9_3_240 _43_3_465 _35 occurs in 10 to 15 percent of patients and is self-limited. The Source: Authors, from epidemiologic data. Rheunatic Heart Disease 225 Table 10-5. Age Distribution at First Attack disease. Table 10-4 illustrates the outcome of 10,000 RF hypo- of Rheumatic Fever and Death in 10,000 thetical episodes based on different epidemiologic studies. Hypothetical Cases without Treatment These studies have shown that initial episodes of RF occurred Ages (years) Number Percent before age fifteen in 90 percent of cases (table 10-5). Thirty percent of all children having one or more RF episodes recover At first attack completely even in the absence of treatment; 12 percent < 5 o00 8 develop mild carditis; 23 percent develop cardiomegaly but no 6-8 2,000 20o 9-11 3,600 36 congestive heart failure. Among the remaining 35 percent, the 12-15 2,600 26 damage to the heart valve is so severe that it causes progressive > 15 1,000 10 heart failure and early death. Total 10,000 100 Diagnosis At death <55 87 2.5 The diagnosis of RF requires a combination of clinical and 5-14 485 14 laboratory criteria because no single symptom or laboratory 45-64 814 23.5 test is pathognomonic of the disease. Rheumatic carditis, or >65 104 3 permanent valvular lesions of RHD are suspected when a heart Total 3,465 IQQ murmur is audible at auscultation. The initial detection of heart murmurs can be successfully done by health workers Source: Authors, from epidemiologic data. (Irwig and others 1985). To confirm the diagnosis and to determine the extent of underlying heart damage, however, In developing countries, the initial episode of RF is often chest x-ray, electrocardiogram, and, if possible, echocardiogra- unnoticed, and arthritis tends to be reported less often than in phy or cardiac catheterization are required. Westem countries. Because of these differences, RF seemed to follow a different clinical course in developing countries. Care- Jones Criteria ful longitudinal studies have not substantiated such differ- ences. Carditis occurs in 34 to 55 percent of children during Dr. T. Duckett Jones systematized clinical diagnosis of RF in the initial RF attack and in 65 to 85 percent of subsequent RF 1944 and classified the observed symptoms into major and episodes in different parts of the world (Majeed and others minor manifestations. Jones criteria have been only slightly 1981; Padmavati and Gupta 1988; Potter and others 1978; modified by subsequent revisions and remain the mainstay for Sanyal and others 1974; see table 10-3). Carditis, nevertheless, the diagnosis of rheumatic fever today (tables 10-6 and 10-7; runs a more severe course in developing countries because the figure 10-2). Standardized diagnostic criteria are important to initial attack of RF occurs at a much younger age, and RF recurs ensure comparability of epidemiological data, for the evalua- more often. tion of prevention and care programs, and to guide therapeutic decisions. Underdiagnosis will prevent affected children from Rheumatic Heart Disease Table 10-6. Changes in the Original Jones Criteria Approximately 90 percent of children who have carditis dur- for the Diagnosis of Acute Rheumatic Fever ing RF episodes will develop RHD. This disease is a chronic and e progressive condition resulting from permanent scarring of M Year heart valves, following RF. It often causes little morbidity in its early stages. Children attend school, run, and play (McLaren Carditis M M M M and others 1978). A few, however, suffer from exhaustion Polyarthritis - M M M when they exercise. As RHD becomes more severe with recur- Chorea M M M M . , .. .. .., . ...................Subcutaneous nodules M MI M M rent bouts Of RF, children and young adults can no longer attend Srytheoa m nadum m - m m school or work, and they withdraw from play and other social Arthralgia Ni - m m activities. Finally, severe congestive cardiac failure occurs, and Fever m m m m medical treatment often fails. Heart surgery then remains the Erythrocvte sedimentation rate m m m m only possible intervention. History of acute rheumatic fever or rheumatic heart Disease Outcone disease M m m m Evidence of prior streprococcal infection - m m R The age at which the first RF attack occurs and the frequency and severity of RF recurrences determine the outcome of RHD. - Not available. Note: N = Major manifestation; m = minor manifestation; a = required Young children six to thirteen years old are at highest risk for manifestation. recurrences, and they suffer a more fulminant course of the Source: WHO 1988. 226 Catherine Michaud, Jorge Trejo-Gutierrez, Carlos Cruz, and Thomas A. Pearson Table 10-7. Cases of Rheumatic Heart Disease body developed against extracellular products of group A strep- in the Developing World, 1985 tococci, peak within three weeks following GASp and provide Population age DePT retrospective evidence of infection with group A streptococci. 6-16 Nears Total casesa per year A single Aso titer does not provide a reliable measure of the Locanon (thousands) (thousands) (thousands) time elapsed since infection because the rate of decrease of ASO Asia titers is highly variable (Denny 1987). Urban 189,000 1,510 Rural 485,000 971 - The Public Health Significance Total 674,000 2,481 56 Children bear the major burden of disease resulting from RF and Latn America R-ID. This burden is likely to increase with continuing popula- Urban 73,000 584 - tion growth in developing countries. RF and RHD have high Rural 33,000 66 - economic costs which could be reduced through effective Total 106,000 650 15 prophylaxis of RF and RHI). Middle East! North Africa Urban 32,000 253 - Current levels in the Developing Countries Rural 81,000 162 - Total 113,000 415 9 The first African case of lH-1D was reported by Procter and Hargreaves in eastern Africa in 1932, but the importance of RF Sub-Saharan Africa and RHD in developing countries was not recognized until the Urban 40,000 319 - 1950s (Anabwani, Amoa, and Muita 1989). Rheumatic heart Rural 100,000 200 - disease is the most common form of heart disease among Total 140,000 519 12 children and one of the most common cardiovascular diseases Total 4,065 92 - Not available. Figure 10-2. Jones Criteria (Revised) for Guidance a. These are conservative estimates, based on the following assumptions: (a) mean urban prevalence 8/1,000; (b) mean rural prevalence 2/1,000. Those in the Diagnosis of Rheumatic Fever severely affected will die within twenty years of the initial attack of RF. b. Assumes 45 percent of cases each year result in death within 20 years. Calculation for number of deaths per year is as follows: Total number of cases Major Manifestations Minor Manifestations in the region/2000n45. Source: Authors, from epidemologic data. Carditis Clinical receiving proper treatment, whereas overdiagnosis may create Polyarthritis - Previous rheumatic fever or rheumatic heart disease unnecessary emotional and psychological suffering among pa- Chorea - Arthralgia tients and parents (Jones 1944). Erythema Marginatum - Fever GASP: Laboratory Diagnosis Subcutaneous Nodules Laboratory - Acute phase reactants Erythrocyte sedimentation rate, Laboratory tests can provide direct identification of group A C-reactive protein, leukocytosis streptococci or retrospective evidence of infection with group - Prolonged P-R interval A streptococci. Direct identification of group A streptococci from the throats of children suffering from pharyngitis is re- Supporting evidence of quired to diagnose GASP. This can be done through cultures, or streptococcal infection by means of rapid antigen detection tests. Cultures on blood - Increased Titer of Anti-Streptococcal Antibodies agar plates are exceptionally effective as a diagnostic tool. ASO (anti-streptolysin 0), others However, they require laboratory facilities, and bacteriologic - Positive Throat Culture identification takes eighteen to twenty-four hours. Rapid an- tot Group A Streptococcus tigen detection tests provide an immediate diagnosis. They - Recent Scarlet Fever have a good specificity of 98 percent, but their sensitivity varies with the amount of antigen present between 44 percent and 100 percent. Negative results, therefore, must be confirmed by Note: Two major or one major and two minor manifestations plus evidence of a culture (Kaplan and Markowitz 1988). The sensitivity of preceding streptococcal infection indicates a high probability of acute rheumatic rapid antigen detection tests needs to be increased to ensure This figure shows the recommendation of the American Heart Association reliable diagnosis and improved clinical management. (Circulation 1984). It has been approved by the WHO Study Group (1988) with the provision that the following be dealt with separately and be exempted Retrospective evidence is important to confirm the diagno- from fulfilling the Jones critera: 'pure'chorea, late on-set carditis, and rheumatic recurrence. sis of RF. Titers of antistreptolysin 0 (ASO), which is an anti- Source: WHO 1988. Rheumatic Heart Disease 227 Table 10-8. Hospital Admissions for Rheumoatic and children were sometimes referred abroad at high cost. Heart Disease For instance, by mid- 1985, 238 RHD cases from French Poly- (percent) nesia had been evacuated to France to undergo heart surgery Admissions as percentage (Vigneron 1989). Country of all cardiac admissions Women of childbearing age are at the highest risk of suffer- ing complications from RHD during pregnancy. The important Bsangaeh34.0 hemodynamic changes occurring during pr-egnancy may preci- Burma 30.0 pitate cardiac failure in women with RHD. Rheumatic heart Mongolia 30.0 disease is an important obstetric complication in Africa, Pakistan 23.0 making up 75 to 90 percent of all symptomatic heart disease Thailand 34.0 cases during pregnancy. In Cape Town, South Africa, closed Africa mitral valvotomies were performed in 41 women age eighteen Ethiopia 34.8 to forty-one between 1965 and 1985 (Vasloo and Reichart Ghana 20.6 1987). Malawi 23.0 Approximately 90,000 deaths due to RHD occur each year Nigeria (Ibadan) 18.1 among all those affected. Peak mortality reported from autopsy Nigeria (Kano) 23.0 studies in Mulago Hospital, Kampala, Uganda, occurred be- South Afrnca 29.7 tween ages twenty-four and thirty-four (Shaper 1972; see figure Uganda 24.7 10-3). Adults living beyond age thirty-five tend to suffer from Zambia 18.2 less severe cardiac damage and, therefore, have lower mortality rates. The death rate from RHD is highest among young adults. Source: WHO 1980a; Hutt 1991. In Sub-Saharan Africa, up to 20 percent of all deaths con- firmed at autopsy were due to RHD (Hutt 1991). Rheumatic among young adults in tropical and subtropical countries heart disease causes approximately I out of 150 deaths occur- (Sharper 1972). In 1985 an estimated 4.2 million people- ring between the ages of sixteen and forty-nine in developing mostly children and young adults-were suffering from RHD; countries. 500,000 of these people had at least one episode of RF, and approximately 90,000 of those affected died (table 10-8). The outstanding features of RF and RHD in tropical and subtropical regions are the youngage atwhich the initial attack Table 10-9. Cost-Effectiveness of Different of RF occurs, the high recurrence rate, and the more fulminant Prophylactic Strategies course of RHD. In Thailand, peak RF incidence occurred be- Cases Cost per tween ages nine and eleven, and peak RHD prevalence between treated to death Cost per twelve and sixteen (Vongprateep, Dharmasakti, and Sindha- Unit cost prevent averted DALYs DALY vanonda 1988). A similar age distribution was observed in Treatment (dollars) one death' (dollars) gainedb (dollars) other developing countries. Prevention of Rheumatic fever and rheumatic heart disease cause severe pharyngitisc 60 682 40,920 39 1,049 disability and are a frequent cause for hospitalization. About Prevention 10 to 35 percent of all clinical cardiac patients in hospitals in of RFd 1,380 4 5,520 39 142 Sub-Saharan Africa and Asia were suffering from RHD (Hutt of RHiDe 8,500 1.5 12,750 30 425 1991; see table 10-9). In Barangwanath Hospital, Soweto, in the Republic of South Africa, RF and RHD accounted for 11 per a. Assuming 100 percent efficacy of each treatment, approximately 500 pharyngitns cases, 100 GASP, three RF, or one severe RHD case would have to be 1,000 of all pediatric patients under age ten (Edington and treated to prevent one death. Gear 1982). The average length of hospital stay of three to four b. Twenty DALYs per death averted. DALYs per disability reduction: three weeks, for both RF and RHD, meant prolonged absence from for mild carditis (10 percent for thirty years); six for moderate carditis (20 percent for thirty years); ten for severe carditis (50 percent for twenty years). school or work. The first two prevention strategies provide a gain of thirty-nine DALYs because In the 1960s, cardiac valvular surgery became available. these interventions reduce the disability from mild and moderate carditis in The high number of children and young adults referred for addition to reducing disability and death from severe carditis; the last strategy does not reduce disability from mild or moderate carditis. surgery confirmed the young age of occurrence and the c. Primary prevention entails one benzathine-penicillin injection each severity of RHD in developing countries. In Nairobi, Kenya, year for ten years. Efficacy of intervention is 70 percent. 33 percent of all valvotomies were done in children under d. Secondary prevention entails one benzarhine-penicillin injection per month for five years and assumes one hospitalization of twenty-four days. age sixteen; in Johannesburg, South Africa, 44 percent of Efficacy of intervention is 80 percent. patients that underwent valvuloplasty were fifteen or youn- e. Tertiary prophylaxis entails valvuloplasty or valve replacement and ger, and 26 percent were under twelve (Anabwani, Amoa, includes hospitalization. Efficacy of intervention is 70 percent. Unit costs range from $5,000 to $12,000; costs per death averted from $7,500 to $18,000; and Muita 1989; Antunes and others 1987). Heart surgery andcostsperDALYfrom$250to$600. was not available, however, in many developing countries, Source: Authors, from epidemiologic data. 228 Catherine Michaud, Jorge Trejo-GuneTTez, Carlos Cruz, and Thomas A. Pearson Figure 10-3. Ages of Subjects with Rheumatic tional ten DALYS, assuming that RHD causes 50 percent disability Heart Disease at Autopsy in Mulago Hospital, during the twenty years preceding death. The total indirect Kampala, Uganda, 1960-65 cost per RHD death, therefore, was estimated to be thirty DALYs. Percent Percent Elements of Preventive Strategy 30 - Male 30- Female 20 - 20 - Interventions to prevent RHDcan be targeted at different points in the chain of events leading from GASP to intractable cardiac 010 - 10 failure. Two different intervention strategies-mass penicillin o - _ o - ~ - - - - - - - prophylaxis or a vaccine-could prevent GASP. Mass chemo- V ~r P 0 b+>,V< ro '5 45 'I >>> Pr prophylaxis is not practical, however, unless the population at risk can be precisely targeted (for example, military recruits), and a vaccine is not yet available. Primary and secondary prevention therefore aim at reducing the occurrence of RF. Source: Sharper 1972. Primary prevention averts the first attack of RF with a single injection of benzathine penicillin, and secondary prevention Possible Patterns of Morbidity and Mortality by 2015 hinders recurrences of RF by providing monthly injections of benzathine penicillin. Finally, rehabilitation through cardiac Total morbidity and mortality attributable to RF and RHD is surgery corrects damage to the valves and interrupts the pro- likely to increase during the next twenty-five years because the gression to cardiac failure (figure 10-4). total population at highest risk for RF, those between the ages of five and fifteen, will increase and reach 1.2 billion by the Mass Chemoprophylaxis year 2015. Half of them will live in urban areas (United Nations 1988). Assuming no changes in the incidence or in Routineadministrationofbenzathinepenicillininjectionswas the prevention of RHD by the year 2015, approximately 6 introduced among all U.S. military recruits after World War million school-age children will suffer from RHD. The level of 11 because more than 20,000 cases of RF had occurred in navy socioeconomic development, access to health care, and progress and marine recruits between 1942 and 1945. The strategy was toward better preventive strategies (that is, a vaccine) will ulti- carried out effectively, and it prevented further epidemics of mately determine the disease burden resulting from RF and RHID. RF. Routine administration was discontinued in the early 1980s but was resumed after two outbreaks of RF occurred in 1986-87 Economic Costs ar the naval training center in San Diego, Califomia, and at the Fort Leonard Wood Army Training Base in Missouri. Both Rheumatic fever and rheumatic heart disease incur direct as military epidemics were effectively terminated by the reintro- well as indirect costs. Estimates of direct costs of RHD are not duction of mass benzathine penicillin prophylaxis to all mili- available in most developing countries. Neutze estimated the tary recruits who were not allergic to penicillin (Bisno 1991). financial cost incurred to the state in New Zealand in 1985 and showed that the country spent $2 million to treat 5,625 RF and RHD patients.' This represents an average cost of $355 per case Figure 10-4. Prevention of Rheumatic Fever and per year. Hospital care, including heart surgery, represented 87 Rheumatic Heart Disease percent and ambulatory treatment 13 percent of the total expenditure. The average hospital stay was twenty-four days Percentage that will develop the next stage for RF, twenty-one days for RHD, and twenty-seven days for surgery (Neutze 1988b). ( ) Number of cases The indirect costs of RHD are quite high because it is primar- ily adolescents and young adults who are disabled or die during GASP 3 -* RF 70 -* RHD 35 -+ Cardiac their most productive years. Mild carditis results in 10 percent + failure disability and moderate carditis 25 percent disability for thirty tj (25) years. Deaths from RHiO are due to progressive heart failure and .ouii occur most often between the ages of twenty-five and thirty- vaccination Prophylaxis four, on average twenty years after the initial RF attack. In addition, RtHD causes approximately 50 percent disability dur- ing the twenty years preceding death. Thus, one can estimate Primary Secondary Rehabilitation the disability resulting from mild carditis to be three disability- prevention prevention (heart surgery) adjusted life-years (DALYs) and the disability from moderate (penicillin) (penicillin) carditis to be six DALYs. Severe carditis causes the loss of twenty DALYs as a result of premature death and the loss of an addi- Source: Authors, trom epidemiological data. Rheumatic Heart Disease 229 Vaccine Development diagnosed. It targets children who have had an initial attack of RF and consists of monthly injections of benzathine penicil- A vaccine could prevent GASP, but vaccine development has lin to prevent recurrences ofGASP that might trigger new bouts been slow because initial vaccine candidates induced the pro- of RF. Secondary prophylaxis should be given to those under duction of antibodies that cross-reacted with the heart valves. the age of twenty-four for at least five years. A vaccine is unlikely to become available for several more years. The advantage of secondary prophylaxis is that it targets a Recombinant DNA technology now provides the means to much smaller population, because only 3 percent of GASP develop synthetic vaccines. An important first step toward the episodes are followed by an initial attack of RF. Rheumatic fever development of a synthetic vaccine was to discover the struc- cases are usually diagnosed when the patients seek treatment. ture of M proteins and to learn the sequence of amino acids of In some instances, rheumatic carditis has been actively different M proteins. M proteins encompass hypervariable and searched for among schoolchildren. Health care workers can conserved regions. Hypervariable regions differ among differ- be trained to note heart murmurs and to refer these children ent serotypes, but conserved regions are common to all sero- for further investigation. types. Protection against group A streptococci could be The greatest challenge of secondary prevention is to ensure induced by antibodies directed to some conserved regions long-term compliance with monthly injections of benzathine common to all M proteins (Fischetti 1991). penicillin for several years (Gordis, Lilienfeld, and Rodriguez 1969b; WHO Study Group 1988). In Hamilton, Australia, for Primary Prevention instance, hospital contact was mairntained by less than half of a group of RF patients during a ten-year period. Secondary Primary prevention consists of a single injection of benza- prevention programs therefore require the establishment of a thine penicillin (or oral penicillin) each day for ten days to registry of RF/RHD patients and good coordination at various treat GASP effectively and prevent the occurrence of RF. levels of the health system to ensure the proper follow-up of Difficulties in finding cases and in establishing the diagnosis all RF patients. Of GASP limit the effectiveness of this strategy. Primary Substantial experience now exists in many developing prevention has been successful in industrial countries where countries. The World Health Organization implemented GASP can be readily diagnosed. In many developing coun- secondary RF prevention programs in Barbados, Cyprus, tries, however, GASP cannot be diagnosed because laboratory Egypt, India, Iran, Mongolia, and Nigeria, as well as in facilities required to identify group A streptococci are not several Latin American countries. In 1970 the cost of the available, and rapid antigen detection tests do not yet have program was estimated to be $325 per month, which was the sensitivity required for a reliable diagnosis. Even where considerably less than the cost of maintaining pediatric beds GASP can be diagnosed, the finding of cases is limited by at $1,260 per month (Hassell and Stuart 1974). The success biological factors, because GASP causes no symptoms in 30 of the programs nevertheless differed among countries. to 50 percent of those who later develop RF. Children at Compliance was considered adequate if fewer than 30 per- greatest risk live in poor, overcrowded conditions and often cent of benzathine penicillin injections were missed. In have only limited access to health services. Barbados the introduction of RF identification cards and In the absence of proper GASP diagnosis, the targeting vigorous follow-up ensured compliance by 89 percent of all required for effective primary prevention is impractical. patients. In other countries, prophylaxis was effective in Several hundred episodes of pharyrigitis would have to be only 80 percent of the cases. The WHO study also demon- treated to prevent a single death from intractable cardiac strated that patients who had only occasional or no penicil- failure (table 10-4). Twenty percent of all pharyngitis epi- lin prophylaxis at all spent on average six times as much sodes are due to group A streptococci. Thus 333 out of 1,000 hospitalization time as those on full prophylaxis (Strasser episodes of pharyngitis would be GASP and would lead to 10 1985). The importance of community participation was fur- cases of RF (3 percent), 7 cases of RHD (70 percent), and 2 or ther demonstrated in Soweto, South Africa, where regular 3 deaths from intractable cardiac failure (35 percent). Be- attendance to clinic-based prophylaxis increased from 17 per- cause children remain at high risk between the ages of five cent to 38 percent seven months after a community outreach and fifteen and commonly have pharyngitis at least twice a program was established (Edington and Gear 1982). In Thai- year, penicillin treatment would be required approximately land the National Control Program for RF/RHD was developed twice a year for ten years to prevent RF effectively. There- by an initial working group and was effectively implemented fore, even though a single injection of benzathine penicillin once it had been approved by the government. Ongoing eval- costs only $3, averting one death would cost $40,920, or uation provides the information required to make the neces- $1,049 per DALY (table 10-9). sary changes (Vongprateep, Dharmasakti, and Sindhavanonda 1988). Secondary Prevention Cost estimates were based on the following assumptions: a benzathine penicillin prophylaxis once each month for five Long-term penicillin prophylaxis is the cornerstone of RF pre- years and one hospitalization of twenty-four days for each vention in developing countries in which GASP cannot be initial RF attack to avert death among 34 percent of patients. 230 Catherine Michaud, Jorge Trejo-Gutierrez, Carlos Cruz, and Thomas A. Pearson Under these assumptions, the cost of secondary prophylaxis to Once the diagnosis has been established, the management avert one death is $5,520, or $142 per DALY. is straightforward: Tertiary Prophylaxis * Eradication of group A streptococci from the throat by a course of penicillin. Tertiary prophylaxis can reverse the progression of intractable * Acetylsalicylic acid (aspirin) or, in serious cases ofcardi- cardiac insufficiency by means of surgical repair or replacement tis, corticosteroid treatment to suppress the accompanying of damaged heart valves. Surgical interventions are important inflammatory response. Response to acetylsalicylic acid and in developing countries because children and young adults corticosteroid is so good that it should not be administered often seek treatment when they already suffer from severe until the diagnosis is confirmed. cardiac failure. The natural history of the disease demonstrates , . , ,, . , , . , ~~~~* Bed-chair rest in the hospital until cardiac manifesta- that surgical treatment is the only effective method in those instances (Antunesand others 1987).tions subside, and mobilization when the acute phase reac- instances (Antunes and others 1987). tat aertmdt oml Different types of surgical interventions exist to repair dam- tants have returned to normal. aged heart valves: a mechanical valve prosthesis or biological * In rare instances of severe carditis, cardiac valve replace- valves are inserted in the heart to replace damaged heart ment or valvuloplasty might be considered. valves; valvuloplasty conserves and repairs damaged valves * Long-term penicillin prophylaxis. rather than replacing them. Mechanical prostheses cause a high incidence of thromboembolic events which are poten- Chronic Rheumatic Heart Disease tially lethal, and anticoagulant therapy is required to avoid them; many biological valves degenerate early as a result of The initial management of the patient with RHD is directed at fibrosis and calcification, leading to early cardiac failure in the control of heart failure. Cardiac arrhythmias are rare in young patients (Abid and others 1989). Thus despite immense young patients but increase with age. Cardiotonic (usually strides in the perfection of materials and design, the ideal valve digitalis) is the appropriate drug to control arrhythmias and remains elusive. Consequently, the desirability of preservation increase myocardial contractility. Diuretics are usually given of heart valves has become more appreciated. Valvuloplasty to control fluid retention associated with heart failure. The causes very few thromboembolic events. Monthly benzathine most important therapeutic decision in RHD is the timing of penicillin prophylaxis is required after surgery to prevent GASP surgical intervention, a costly tertiary prophylaxis requiring and to avoid damage to the new cardiac valves (Antunes and tertiary-care centers. others 1987). In all cases, with or without surgery, long-term penicillin The cost-effectiveness of those procedures needs careful prophylaxis and prophylaxis of bacterial endocarditis before evaluation because so many children and young adults suffer other surgical procedures must be ensured. today from debilitating or intractable cardiac failure. Richard Bumgarner found that in China, out of 410,000 children Priorities suffering from RHD, 80,000 need valvular replacement (Bum- gamer, personal communication). The mean cost of tertiary Secondary prophylaxis is the most cost-effective approach to prophylaxis to avert one death is $12,750, or $425 per DALY prevent RF and RHD in developing countries. It is important to (table 10-9). Tertiary prophylaxis is an attractive option de- take into account the level of training of health care providers spite high intervention costs. and the coverage of the health system while planning and implementing secondary prophylaxis in each country. Elements of Case Management Strategy Priorities for Resource Allocation The experience gained thus far has led to the formulation of clear guidelines for the clinical management of cases of r and Recent experiences in several developing countries have dem- RHD. onstrated that secondary prophylaxis is a cost-effective ap- proach. In addition to providing penicillin, it is important to Acute Rheumatic Fever provide careful training of health care personnel in diagnosing heart murmurs and to ensure close follow-up of patients with RF The patient who has even a suspicion OfRF should be hospitalized and RHD by sensitizing health care providers and communities to for diagnosis and initial treatment. In most secondary-level the need for monthly penicillin injections. Where a good health hospitals the diagnosis can be established according to Jones care infrastructure is already in place, prevention of RHD can be criteria. Other diagnostic procedures usually include chest included in national health plans with little added cost. In areas x-ray, blood cell count, erythrocyte sedimentation rate, with little health coverage, populations at highest risk-school- C-reactive protein, and ASO titer. A limitation in developing age children of low socioeconomic groups enrolled in school in countries is the availability of a microbiological laboratory for large urban areas-could receive special attention and secondary throat culture and exclusion of infective endocarditis. prophylaxis through the school system. Rheumatic Heart Disease 231 Priorities for Research Hosier, D. M., J. M. Craenen, D. W. Teske, and J. J. Wheller. 1987. "Resur- gence of Acute Rheumatic Fever." American Journal of Diseases of Children. Further research is crucial to reduce the present burden of 141:730. disease due to RF and RHD. New tools are needed to diagnose Hutt, Michael S. R.. 1991. "Cancer and Cardiovascular Disease." In R. G. Feachem and D. T.Jamison, eds., Disease and Mortality in Sub-Saharan Africa. pharyngitis due to group A streptococci in developing coun- New York: Oxford University Press. tries and to produce a vaccine. Research is also important to Irwig, L. M., B. Porter, T. D. Wilson, L. D. Saunders, Lucy A. Wagstaff, N. monitor changes in the epidemiology of RF and RHD and to Liesch, S. G. Reinach, M. S. Makhaya, and J. S. Gear. 1985. "Clinical assess the effectiveness of interventions. Competence of Pediatric Primary Health Care Nurses in Soweto." South African MedicalJournal 67:92-95. * Biomedical research is necessary to unravel fully the "Jones Criteria (Revised) for Guidance in the Diagnosis of Rheumatic Fever." pathogenesis of RHD and to develop a simple diagnostic test 1984. Circulation 69:203A-208A. and a vaccine. Jones, T. D. 1944. "The Diagnosis of Rheumatic Fever." JAMA 126:481-84. o Epidemiologic research isimportanttosseKaplan, Edward L., and Milton Markowitz. 1988. "Rheumatic Fever in the * Epidemiologic research Is important to assess the magni- UnitedStates:NoLongeraDiseaseofthePast."New ZealandMedicalJournal tude of the problem in different geographic areas and to 101:402-4. monitor trends. MacDonald, K. T., and A. C. Walker. 1989. "Rheumatic Heart Disease in * Finally, operations research is essential to evaluate the AboriginalChildren in theNorthemTerritory." MedicaljournalofAustralia effect of interventions. 150:503-5. McLaren, M. J., D. M. Hawkins, H. J. Koomhof, K. R. Bloom, D. M. Bramwell-Jones, E. Cohen, G. E. Gale, K. Kanarek, A. S. Lachman, J. B. Lakier, W. A. Pocock, and J. B. Barlow. 1978. "Epidemiology of Rheumatic Hearr Disease in Black School Children of Soweto, Johannesburg." Britsh References MedicalJournal 3:474-78. Maharaahj, B., R. B. Dyer, W. P. Leary, D. D. Arbukle, T. G. Armstrong, and Abid, F., N. Mzah, F. El Euch, and M. Ben Ismail. 1989. "Valve Replacement D. J. Pudifin. 1987. "Correspondence: Screening for Rheumatic Heart in Children under 15 Years with Rheumatic Heart Disease." Pediatric Disease amongst Black School Children in Inanda, South Africa." Jounal Cardiology 10: 199-204. of Tropical Pediamcs 33:60-61. Anabwani, G. M., A. B. Amoa. and A. K. Muita. 1989. "Epidemiology of Majeed,H.A.,N.N.Khan, M.Dabbagh,andothers. 1981."AcuteRheumatic Rheumatic Heart Disease among Primary School Children in Western FeverduringChildhoodinKuwait.TheMildNatureoftheAttack."Annals Kenya." Internanonal Journal of Cardiology 23:249-52. of Tropical Pediatrics 1:13-20. Antunes, M. J., M. P. Magalhaes, P. R. Colsen, and R. H. Kinsley. 1987. Majeed, H. A., F. A. Khuffash, D. C. Sharda, S. S. Farwana, A. F. el "Valvuloplasty for Rheumatic Mitral Valve Disease, a Surgical Challenge." Sherbiny, and S. Y. Ghafour. 1987. "Children with Acute Rheumatic Journal of Cardiovascular Surgery 94:44-56. Fever and Acute Poststreptococcal Glomerulonephritis and Their Behrman, Richard E., and Victor C. Vaughan, eds. 1983. Nelson Textbook of Families in a Subtropical Zone: A Three Year Prospective Compara- Pediatrics, twelfth edition. Philadelphia: W. B. Saunders Company. tive Epidemiological Study." International Journal of Epidemiology 16: Bisno, A. L. "Group A Streptococcal Infections and Acute Rheumatic Fever." 561-67. 1991. New EnglandJournal of Medicine 325:783-93. Markowitz, Milton. 1970. "Eradication of Rheumatic Fever. An Unfulfilled Caims, L. M. 1988. "The Immunology of Rheumatic Fever." New Zealand Hope." Circulation41:1077-84. MedicalJournal 101:388-91. . 1983. "Rheumatic Fever." In Richard E. Behrman, and Victor C. Denny, F. W. 1987. "T. Duckett Jones Memorial Lecture: T. Duckett Jones Vaughan, eds., Nelson Textbook of Pediatrics, twelfth edition. Philadelphia: and Rheumatic Fever in 1986." Circulation 76:963-70. W. B. Saunders Company. Edington, M. E., and J. S. Gear. 1982. "Rheumatic Heart Disease in Soweto- . 1985. "The Decline of Rheumatic Fever: Role of Medical Intervention. A Program for Secondary Prevention." South Afnican Medical Journal 62: Lewis W. Wannaker Memorial Lecture." Journal of Pediatrics 106:545-50. 523-25. -. 1988. "Evolution and Critique of Changes in the Jones Criteria for Fischetti, V. A. 1991. "Streptococcal M Protein." Scientific Amnencan, June, the Diagnosis of Rheumatic Fever." New Zealand Journal of Medicine 101: 58-65. 392-94. Goldstein, ., P. Rebeyrotte, J. Parlebas. 1968. "Isolation from Heart Valves of Massell, B. F., C. G. Chute, A. M. Walker, and G. S. Kurland. 1988. "Penicillin Glycopeptides which Share Immunological Properties with Streptococcus and the Marked Decrease in Morbidity and Mortality from Rheumatic Fever hemolyticus Group A Polysaccharides." Nature 219:866-8. in the United States." New EnglandJournal of Medicine 318:280-86. Gordis, Leon. 1985. "T. Duckett Jones Memorial Lecture. The Virtual Disap- Neutze, J. M. 1988a. "Rheumatic Fever and Rheumatic Heart Disease in the pearance of Rheumatic Fever in the United States: Lessons in the Rise and Westem Pacific Region." New Zealand MedicalJournal 101:404-6. Fall of the Disease." Circulation 72:1155-62. . 1988b. "The Third Intemational Conference on Rheumatic Fever Gordis, Leon, Abraham Lilienfeld, and Romeo Rodriguez. 1969a. "Studies in and Rheumatic Heart Disease." New Zealand MedicalJournal 101:387. the Epidemiology and Preventability of Rheumatic Fever-1. Demographic Padmavati, S., and Vijay Gupta. 1988. "Reappraisal of the Jones Criteria: The Factors in the Incidence of Acute Attacks." Journal of Chronic Diseases Indian Experience." New Zealand MedicalJournal 101:391-92. 21:645-54. Patarroyo, M. E., R. J. Winchester, Alberto Vejerano, Allan Gibofsky, .1969b. "Studies in the Epidemiology and Preventability of Rheumatic Femand Chalem, J. B. Zabriskie, and H. G. Kunkel. 1979. "Association of Fever-11. Socio-economic Factors and rhe Incidence of Acute Attacks." a B-Cell Alloantigen with Susceptibility to Rheumatic Fever." Nature Journal of Chronic Diseases 21:655-66. 278:173-74. Hassell T. A., and K. L. Stuart. 1974. "Rheumatic Fever Prophylaxis:AThree Potter, E. V., Mauri Svartman, Isahak Mohammed, Reginald Cox, Theo Year Study." British MedicalJournal 2:39-40. Poon-King, and D. P. Earle. 1978. "Tropical Acute Rheumatic Fever and 232 Cathenne Michaud, Jorge Trejo-Gutierrez, Carlos Cruz, and Thomas A. Pearson Associated Streptococcal Infections Compared with Concurrent Acute Intermountain Areaof the United States." New Englandjournal of Medicine Glomenulonephritis."Journal of Pediatrics 92:325-33. 316:421-26. Sanyal, Shvamal. K., M. K. Thapar, S. H. Ahmed, Vijaya Hooja, and Promila Vigneron, Emmanuel. 1989. "The Epidemiological Transition in an Overseas Tewari. 1974. "The Initial Attack of Acute Rheumatic Fever During Territory:DiseaseMappinginFrenchPolynesia."SocialScienceandMedicine Childhood in North India. A Prospective Study of the Clinical Profile." 29:913-22. Circulation 49:7-12. Vongprateep Choompol, Duangsuda Dharmasakti, and Kamol Sindha'a- Sharper, A. G. 1972. "Cardiovascular Disease in the Tropics-1. Rheumatic nonda. 1988. "The National Program and the Control of Rheumatic Heart Heart." British MedicalJournal 3:683-86. Disease in Two Project Areas of Thailand." New Zealand Medical Journal Strasser, Toma. 1985. "Cost-Effective Control of Rheumatic Fever in the 101:408-10. Community." Health Policy 5:159-64. Wald, Ellen R., Barry Dashefsky, Cindy Feidt, Darleen Chiponis, and Carol Taranta, Angelo. Jeta Torosdag, Julius D. Metrakos, Wanda Jegier, and Irene Byers. 1987. "Acute Rheumatic Fever in Westem Pennsylvania and the Uchida. 1959. "Rheumatic Fever in Monozygotic and Dizygotic Twins." Tristate Area." Pedianics80:371-74. Circulation 20:778. WHO (World Health Organization). 1980a. "Community Control of Rheu- Vasloo, M. B., and B. Reichart. 1987. "The Feasibility of Closed Mitral matic Fever in Developing Countries: 1. A Major Public Health Problem." Valvotomy in Pregnancy." Journal of Thoracic and Cardiovascular Surgers WHO Chronicles 34:336-45. 93:675-79. 1.980b. "Community Control of Rheumatic Fever in Developing Coun- Veasy, L. George, S. E. Wiedmeier, G. S. Orsmund, H. D. Ruttenburg, M. tries: 2. Strategies for Prevention and Control." WHO Chronicles 34: 389-95. M. Boucek, S. j. Roth, V.F. Tait, J. A. Thompson,]. A. Daly, E. L. Kaplan, - . 1988. "Rheumatic Fever and Rheumatic Heart Disease." Study Group and H. R. Hill. 1987. "Resurgence of Acute Rheumatic Fever in the Report. Technical Report 764. 11 Tuberculosis Christopher Murray, Karel Styblo, and Annik Rouillon Tuberculosis is an ancient disease that has long been a signif- primary complex. The primary complex has two components, icant public health challenge in the world and remains a one in the lung and the other in the corresponding lymph node significant health problem in developing countries. In the last or nodes. In most cases, both the primary pulmonary lesions century, tuberculosis was responsible for nearly one in ten and lesions in lymph nodes heal spontaneously, leaving behind deaths in Europe (Preston, Keyfitz, and Schoen 1972). There a focus of a few "dormant" bacilli that can be reactivated and is reliable evidence that irrespective of its magnitude, the cause clinical disease at any moment during an individual's tuberculosis problem in industrial countries has been decreas- lifetime. Before the development of allergy and immunity, ing for at least the last forty years, since the introduction of some bacilli escape from the primary lesions into the blood antituberculosis chemotherapy. In many industrial countries, stream and set up blood-borne foci in other parts of the body, a steady decrease in mortality from tuberculosis in the pre- for example, in the kidneys, ends of long bones, spine, or brain. chemotherapy era was observed from the turn of this century In newborns and small children the infection progresses either if not before (Frost 1937; Styblo 1986). The elimination of in the primary site or metastatic foci, and serious forms of tuberculosis in most industrial countries will not be substan- tuberculosis may develop, in particular, miliary tuberculosis tially influenced by acquired immunodeficiency syndrome and tuberculous meningitis. These forms of tuberculosis also (AIsS) because of the low prevalence of tuberculous infection occur in adolescents and adults but much less frequently than in subjects age twenty to fifty years in whom infection from the in newborns and small children. human immunodeficiency virus (HIV) is most frequent (Styblo Two to six weeks after the primary infection, the body's 1989). In developing countries, however, tuberculosis contin- immune system develops a certain level of cell-mediated im- ues to be an important problem and there appears to have been munity to M. tuberculosis antigens. This leads to the formation virtually no tendency for tuberculosis to eliminate itself in the of granulomas-a type of histological pattern-around the absence of intensive control measures. Unlike in industrial focus of the bacilli. When these areas become calcified, they countries, HIV infection will result in a considerable increase of may be detected on a chest x-ray as a calcified primary com- tuberculosis cases in those developing countries where both plex. (If the calcified lesions of the primary complex in the lung tuberculous and HIV infections are prevalent. Tuberculosis and the lymph node are too small, they may not be seen on an remains, therefore, one of the top priorities for action in x-ray.) Clinical disease, however, may occur weeks to years developing countries, because tools exist to diagnose and cure after the primary infection with the bacillus, although about infectious cases of tuberculosis and thus to decrease transmis- 80 percent of all cases occur during the first two years after sion of tuberculous infection. infection (Sutherland 1968). The probability of progressing The epidemiology of tuberculosis is complex and a certain from a primary infection to clinical tuberculosis is discussed knowledge of the natural history of tuberculosis is required in more fully below. The key aspect of the natural history of order to discuss the policy options. Tuberculosis is caused by tuberculosis is that infection may lead, in a relatively small the bacillus Mycobacterium tuberculosis, which in most cases proportion of infected persons, to clinical disease at a later attacks the lungs. Infection is most commonly transmitted date. Consequently, the process of elimination of tuberculosis from persons with pulmonary tuberculosis to other persons, in in a community is very slow, because a certain risk of latent particular when coughing or sneezing. The most important infections developing into active tuberculosis (endogenous exception to the airbome route of infection is infection of the exacerbation) cannot be completely prevented. digestive tract through contaminated milk containing Myco- Four important diagnostic strategies are used to detect tu- bacterium bovis from cows suffering from tuberculosis, which berculous infection and clinical disease. First, a recently or causes a disease clinically similar to tuberculosis. remotely infected person, whether or not he or she has clinical One or more bacilli reaching the lung tissue can cause a disease, develops a certain degree of cell-mediated immune nonspecific inflammatory response, which may result in a response to M. tuberculosis antigen. An intradermal injection 233 234 Christopher Murray, Karel Styblo, and Annik Rouillon of tuberculin (preferably purified protein derivative) will cause negative they are much less infectious.) The above two cate- an induration in forty-eight to seventy-two hours. This skin gories are sometimes labeled infectious or open tuberculosis test (Mantoux test) permits a relatively easy detection of the and noninfectious tuberculosis, respectively (India, Ministry prevalence of tuberculous infection in any population. The of Health and Family Welfare 1986). The distinction between tuberculin test does not distinguish, however, between recent sputum smear-positive tuberculosis and other tuberculosis is and remote infections or between an infection caused by M. particularly important when considering the policy options for tuberculosis and one caused by M. bovis or by some other tuberculosis control and prevention. mycobacteria. In spite of these limitations, tuberculin sensitiv- Tuberculosis without detection and the institution of ade- ity surveys in a representative sample of a population are one quate treatment is highly fatal-specific studies will be re- of the mainstays of tuberculosis epidemiology. The tuberculin viewed below. Because mycobacteria are able to survive within test, however, has a limited value for the diagnosis of clinical host lesions as persisters (dormant bacilli), treatment is long tuberculosis. If the test is positive, it does not distinguish and requires, in smear-positive cases, the combination of at between infection and disease; if it is negative it does not least two drugs in the initial intensive phase. Length of treat- always exclude disease. Studies by Canetti (1939, 1972) indi- ment ranges from six to eighteen months. cate that most patients who have been infected and have a pos- itive skin test maintain viable bacilli within their bodies. Tuberculosis Incidence and Mortality Second, detection by microscopy of acid-fast bacilli (nearly always identical with tubercle bacilli) in sputum and other In the following section, we outline the empirical and epi- specimens (for example, gastric washings) is the most import- demiological basis for estimating tuberculosis incidence and ant tool to detect highly infectious cases of tuberculosis. There mortality. is strong evidence (Rouillon, Perdrizet, and Parrot 1976; Styblo 1984) that those patients whose sputa contain sufficient bacilli Tuberculosis Incidence to be detected by microscopy are highly infectious. These cases are referred to as "smear-positive." To put tuberculosis in the proper perspective we need to know Third, the culture of specimens for mycobacteria detects, in the number and the age distribution in new cases of tubercu- about four to six weeks, tubercle bacilli in sputum containing losis which develop in a community each year, as well as the insufficient bacilli to be detected by microscopy. These cases number and the age distribution of patients who die from are then classified as sputum smear-negative but "culture- tuberculosis each year. Health information systems in devel- positive" pulmonary tuberculosis. Patients whose sputum is oping countries are too incomplete to provide meaningful smear-negative and culture-positive or culture-negative are information on the incidence or mortality of tuberculosis several times less infectious than smear-positive cases. (Styblo and Rouillon 1981). We are forced to estimate the Fourth, in smear- and culture-negative patients (particu- burden of tuberculosis indirectly by using several epidemiolog- larly in children and young adults) diagnosis of tuberculosis is ical parameters. These include the average annual risk of made on the basis of clinical examination and interpretation tuberculous infection and the incidence of smear-positive pul- of chest x-ray. monary tuberculosis, the proportion of all cases of tuberculosis Extrapulmonary tuberculosis is diagnosed by, in a number of that are smear positive, and case-fatality rates for smear- cases, bacteriology (in patients with tuberculous meningitis, positive tuberculosis and other tuberculosis. lymphadenitis, genitourinary tuberculosis, and the like) or by histology of biopsy material. Depending on the site of infec- ANNUAL AVERAGE RISK OF TUBERCULOUS INFECTION. Tubercu- tion, roentgenologic and other special examinations are re- losis epidemiologists have used skin tests to measure the prev- quired to diagnose extrapulmonary tuberculosis. It is important alence of infection in communities. A technique has been to stress that extrapulmonary tuberculosis is either noninfec- developed for converting this information on prevalence of tious or the degree of infectivity is very low. The natural history of tuberculosis illustrates that patients Ta le1. EstimatedRisksofTuberculosis suffering from smear-positive pulmonary tuberculosis are the Infection in Developing Countries 1985-90 main source of infection. For the rest of this chapter, therefore, (pectcent) tuberculosis will be divided into two categories: (a) sputum smear-positive tuberculosis, which will be referred to as smear- Risk of Annual decrease positive tuberculosis; and (b) other tuberculosis, which in- Area infection in risk cludes pulmonary tuberculosis, in which the sputum is smear Sub-Saharan Africa 1.50-2.50 1-2 negative, and extrapulmonary tuberculosis. Because children North Africa and westem Asia 0.50-1.50 4-5 rarely suffer from sputum smear-positive tuberculosis, most Asia 1.00-2.00 1-3 cases of tuberculosis in children will be included in the cate- South America 0.50-1.50 2-5 gory "other tuberculosis." (If children are smear positive, they Central America and the are highly infectious sources of infection. If they are smear Caribbean___________1_3 negative and culture positive or smear negative and culture Source: Based on Cauthen, Pio. and ten Dam 1988. Tuberculosis 235 Figure 1, 1. Relationship between Annual Risk detection rates can be only a fraction of the respective true of Infection and Incidence of Smear-Positive incidence rates. Tuberculosis Prevalence of smear-positive cases is of limited value as an epidemiological index because it largely depends on the quality Incidence (per 100,000 U.S. dollars) of chemotherapy of smear-positive cases and the extent and 500 quality of case finding. (In industrial countries, prevalence may be substantially lower than incidence, especially in countries 400 where a six-month course of treatment is given to patients. In developing countries, prevalence may be several times higher 300 - than incidence if treatment results are poor and the case- detection rate is low.) For these very same reasons, prevalence 200 - 0 may be an important indicator for management of a national tuberculosis control program, but estimates of prevalence de- 100 - pend on too many locally specific parameters to be made here for regions or for the developing world as a whole. 1 2 3 4 5 6 7 8 The relationship between the annual risk of infection and the incidence of smear-positive tuberculosis can provide one Risk of infection (percent) of the only means of estimating the incidence of smear-positive tuberculosis (Styblo 1985). Styblo examined the relationship Source: Authors. between the annual risk of infection and incidence of smear- positive pulmonary tuberculosis using a variety of data sources tuberculous infection into a series of annual risks of tuberculous from the developing and industrial world. We have recom- infection (Styblo, Meijer, and Sutherland 1969; Sutherland puted this relationship using only the results of a series of 1976). If several tuberculin surveys of the same population surveys sponsored by the World Health Organization in devel- have been made at different times (using similar techniques oping countries and data from the Netherlands before chemo- and testing a representative sample of subjects of the same age therapy was widely available. We must note that for some of not vaccinated with BCG Ibacille Calmette-Guerin]), the level these surveys data are available on the prevalence of smear- of and percentage decrease in the risk of infection can be positive tuberculosis, not the incidence. In such cases, we estimated. Techniques have been developed to estimate, if the derived the incidence rates by using the historical observation pattem of the annual risk of infection by age is assumed, the that the prevalence of smear-positive tuberculosis was usually level and time trend in the annual risk of infection from a twice the incidence in the communities without widespread single tuberculin survey (Sutherland 1976). The annual risk of institution ofchemotherapy (Holm 1970). In these developing infection tells us the probability that any individual will be countries,therelationshipbetweentheannualriskof infection infected or reinfected with M. tuberculosis in one year. This and incidence of pulmonary smear-positive tuberculosis was measure has become the standard indicator of the tuberculosis linear. A least squares regression line (figure 11 -1) gives an burden in a community (Leowski 1988). Since the 1950s many different tuberculin sensitivity surveys in developing countries have provided us with an Table 1-2. Estimated Incidence of Smear -Positive approximate picture of the annual risk of infection in differ- Tuberculosis in Developing Countries, 1985-90 ent regions of the developing world. Our best estimates, Cases based on a recent review of survey data on the annual risk Incidence of infection, are presented in table 11-1. The annual risk of Area Low Midpoint High (per 100,000) tuberculous infection is probably highest in Sub-Saharan Sub-Saharan Africa, followed closely by Asia. For comparison the annual Africaa 342,921 591,445 839,970 117 risk of infection in the Netherlands in 1985 was 0.012 North Africa and percent (Styblo 1989). western Asia 52,592 145,640 238,687 54 Asia 1,141,877 2,298,393 3,454,909 79 INCIDENCE OF SMEAR-POSITIVE TUBERCULOSIS. Incidence of t America 57937 160,440 262,943 54 smear-positive pulmonary tuberculosis is one of the two key and the Carib- epidemiological indexes (the other being the average annual risk bean 30,022 83,138 136,266 54 of tuberculous infection) for evaluation of the overall tuberculosis Total 1,625,349 3,279,056 4,932,775 79 situation. Lack of data on smear-positive tuberculosis cases in developing countries makes it difficult to convey the enormity of Note: Based on annual risk of infection for each region presented in table the tuberculosis problem to the public health community. It is not 11-1, 1990 popuIlation, and incidence of thirtv-nine to fifty-nine cases per II OO,Q0 population for each I percent annual risk of infection. possible readily to obtain reliable information on incidence of a. Includes cases attributable to dual Hiv/tuberculosis infections. smear-positive tuberculosis in developing countries because case- Source: Authors. 236 Christopher Murray, Karel Stybkl, and Annik Rouitlon Figure 11 -2. Age Distribution of Smear-Positive given annual risk of infection will vary substantially between Tuberculosis in Four Sub-Saharan Tuberculosis communities. Because the tuberculosis control program in Programs Tanzania is well organized and captures most of the tuberculo- sis cases, the age distribution from Tanzania will be used as Percent representative of the developing world. In figure 11 -2 we show the age distribution of smear-positive tuberculosis in Tanzania for 1985-87 (Chum and others 1988), Malawi for 1989, Mo- 30 _ zambique for 1989, and Benin for 1989. The pattern is remark- 25 - ably similar in these four countries, all of which have good programs and case registration. It is important to note that 20 BCG coverage in Tanzania was roughly 50 percent in 1983-87 (Bleiker and others 1987), based on scar examination in the 15 National Tuberculin Survey in Tanzania carried out on 80,000 10 schoolchildren from twenty regions selected at random from 1983 to 1987, which is below the officially reported average 5 for the developing world (UNICEF 1988). Thus any effect 0-I * 214 [ [ I w * * W W 10 l such BCG coverage may have on preventing tuberculosis in 0 0-14 15-24 25-34 35-44 45-54 55-64 65+ children is partially represented in the age distribution; be- cause world coverage is probably higher than in Tanzania, the Age (years) estimate for the incidence of smear-positive tuberculosis in children based on this age distribution may be slightly high. Clearly, smear-positive cases are relatively rare in children; Tanzania _ Mozambique Malawi _ Nicaragua smear-positive tuberculosis is concentrated in adults-more than 80 percent of cases occur between the ages of fifteen and Source: Tanzania: Chum and others 1988; other countries: government fifty-four, according to the data from these four countries. registry data. INCIDENCE OF OTHER FORMS OF TUBERCULOSIS. Estimates of the incidence of smear-negative pulmonary and extrapulmonary estimate of 49 cases of smear-positive tuberculosis per 100,000 for every 1 percent annual risk of infection. The 95 percent confidence interval for the coefficient is 39 to 59.1 Figure 11-3. Smear-Positive Tuberculosis as a Using the estimates of the risk of infection for different Proportion of All Cases of Tuberculosis, by Age, regions in table 11-1 and the confidence interval for the United States (1985-87) and Norway (195 1-72) relationship between incidence of pulmonary smear-positive tuberculosis and the risk of infection, we have calculated the low and high estimates of the incidence of smear-positive Percent tuberculosis for different regions (table 11 -2). The midpoint of 70 the confidence interval of the estimates of smear-positive 60 - incidence is 3,208,000 cases, or an incidence of 77 per 100,000 in the developing world. These must be viewed as only crude so _ estimates, which nevertheless illustrate the continuing magni- 40 - tude of the tuberculosis problem. 30 - AGE DISTRIBUTION OF SMEAR-IPNSITIVE TUBERCULOSIS. The age 20 distribution of incidence is important in determining the effect 10 on public health of smear-positive tuberculosis and the most 0 appropriate means of preventing or controlling tuberculosis. From the historical record of industrial countries and epidemi- ological models, the age and sex distribution of incidence Age (years) appears to change as the annual risk of infection declines. Because most developing countries have annual risks of tuber- culous infection between I and 2 percent, we propose to use the age distribution of the incidence of smear-positive tuber- United States Norway culosis from a developing country with an annual risk of infection in this range. There is no reason to believe that the Soure: United States: CDC (unpublished data); Norway: Tuberculosis epidemiology and thus the age distribution of incidence for a Surveillance Research Unit (TSRU ; unpublished data). Tuberculosis 237 Figure 1 1-4. Estimated Age Distribution of smear-positive tuberculosis in Tanzania and the age-specific Tuberculosis in the Developing World, 1990 ratios of other to smear-positive in the United States, we have derived the rough estimate of the age distribution of other Percent tuberculosis shown in figure 11-4. Although the assumptions underlying these estimates of other tuberculosis may be chal- 30 lenged on many grounds, we believe it is preferable to make 25 - some objective attempt to estimate the age distribution of smear- negative and extrapulmonary tuberculosis in developing coun- 20 - tries because it is an important input to policy decisions. 15 _ Our estimations imply that there are 1.22 cases of smear- negative and extrapulmonary tuberculosis for every case of 1 0 - smear-positive tuberculosis in developing countries with an _ _ _ n annual risk of infection between I and 2 percent and an overall age distribution similar to Tanzania. Low and high estimates 0 of the number of new cases of smear-negative and extra- 0-14 15-24 25-34 35-44 45-54 55-64 65+ pulmonary tuberculosis for each region in the developing world are provided in table 11-3. For all types of tuberculosis Age (years) combined, the data in table 11-4 indicate that the incidence of tuberculosis exceeds 260 per 100,000 in Sub-Saharan Africa. Smear-positive Other HIV-ASSOCIATED TUBERCULOSIS INCIDENCE. The close relation- ship between HIV infection and clinical tuberculosis that has been widely observed will substantially affect the predicted Source: Authors, incidence of clinical tuberculosis in regions with high levels of HIV seropositivity. Using the most recent estimates of country- tuberculosis are also needed. These forms of tuberculosis are specific seroprevalence provided by the World Health Orga- particularly difficult to quantify because the main diagnostic nization, we estimate that there are approximately 4.9 million tool used in developing countries, sputum microscopy, does HiV-seropositive patients in Sub-Saharan Africa. Using esti- not detect these cases. Because the diagnosis of extrapulmo- mates of the prevalence of tuberculosis infection in the region, nary tuberculosis is often based on clinical criteria, no survey we arrive at the approximate figure of 2.1 million cases of dual data are available to estimate the relation between the risk of HIV and tuberculosis infections. Individuals with dual infec- infection and other tuberculosis. In the past, estimates of tions have much higher rates of breakdown from infection to smear-positive tuberculosis have simply been doubled to pro- clinical disease (Selwyn and others 1989). A range for the vide a figure for other tuberculosis (Styblo and Rouillon 1981; annual breakdown rate in dually infected individuals of 5 to Leowski 1988). The distribution of total cases between the 10 percent has been used to estimate that an additional categories sputum smear-positive and other tuberculosis can- 105,000 to 210,000 cases of tuberculosis occur each year in not be accurately established. Whereas smear-positive tuber- Sub-Saharan Africa. These extra cases have been included in culosis and tuberculosis positive by culture can be objectively tables 11-2 through 11-6. If seroprevalence continues to rise, determined, the number of culture-negative cases detected the tuberculosis burden attributable to the HIV epidemic will depends on various factors, such as whether or not mass min- also rise. For the estimates in this chapter, we assume the same iature radiography is used to find cases (this method was clinical spectrum between smear-positive and other tubercu- extensively employed in Europe in the 1950s, 1960s, and losis for Hiv-positive patients (Chaisson and Slutkin 1989). 1970s), the criteria used for activity in asymptomatic cases, age group, and so on. We will assume that within each age group, Tuberculosis Mortality using the same diagnostic approach, the percentage of cases that are sputum smear-positive and the percentage of other This section discusses death rates of tuberculosis and their age cases should be the same independently of the overall annual distribution. risk of infection. The proportion of all tuberculosis cases in the United States and Norway that are smear-positive by age CASE-FATALITY RATES. In order to calculate tuberculosis mor- is shown in figure 11-3 (Galtung Hansen 1955; and personal tality from the estimates of incidence derived above, we need communication from the Centers for Disease Control in to estimate the case-fatality rate. Without appropriate chemo- Atlanta (CDC) in 1989). Because the data set for the United therapy, tuberculosis is highly fatal. Two types of sources States is larger and no mass miniature radiography was used on provide information on the relationship between incidence a large scale, we will use the ratio of cases of other tuberculosis and tuberculosis mortality: data from before chemotherapy was to smear-positive tuberculosis within each age group in the available in industrial countries and survey data from southern United States. Using the age distribution of the incidence of India. First, Drolet (1938) investigated the relation between 238 Christopher Murray, Karel Styblo, and Annik Rouillon Table 11-3. Estimated Incidence of Other Formns Table 11-5. Estimated Cases of Tuberculosis of Tuberculosis in Developing Countries, 1990 Detected and Case-Fatality Rates in Developing Cases Countries, 1990 Incidence Case-fatality rates' Area Low Midpoint High (per 100,000) (percent) Percentage of Sub-Saharan Cases total cases Africa' 418,363 721,563 1,024,763 143 Area detected actually detected Low High North Africa and westem Asia 64,162 177,680 291,198 66 Sub-Saharan Africa 325,132 25 39 47 Asia 1,393,090 2,804,039 4,214,989 96 North Africa and South America 70,683 195,737 320,791 66 westem Asia 222,686 69 26 29 Central America Asia 2,572,809 50 32 37 and the Caribbean 36,627 101,429 166,231 66 South America 221,856 62 28 32 Total 1,982,925 4,000,448 6,017,972 96 Central America _________________________________________________ and the Car-ibbean 62,054 34 38 45 Note: Based on the relationship between smear-positive tuberculosis and Total 3,404,537 47 33 38 other forms of tuberculosis in the United States by age, combined with the age _ distribution of smear-positive tuberculosis in Tanzania. Note: Based on assumption that 15 percent of patients receiving standard a. Includes cases attributable to dual Hlv/tuberculosis infections. chemotherapy die; this is a conservative assumption. Source: Authors. Source: Authors. the tuberculosis mortality rate and reported incidence in se- at five years and 73.3 percent at ten years. Berg found that even lected American cities from 1915 to 1935. He found that the fifteen to nineteen years after diagnosis, smear-positive pa- estimated case-fatality rate for all types of tuberculosis in tients had mortality rates five times higher than the general Detroit and New Jersey was 58.8 percent and 54.9 percent, population of the same age. Second, a five-year study of the respectively. The calculated case-fatality rate varied little dur- natural history of tuberculosis in Bangalore, India, found that ing that twenty-year period. Similar case-fatality rates for all 49 percent of smear-positive and culture-positive patients forms of tuberculosis were recorded in European countries- whose tuberculosis was detected on the first round of the survey Denmark, 1925-34: 51.2 percent (Lindhart 1939); Norway, were dead within five years (Olakowski 1973; National Tuber- 1925-44: 50.6 percent (Galtung Hansen 1955); and England culosis Institute, Bangalore 1974). Mortality was concentrated and Wales, 1933-35: 49.1 percent (Drolet 1938). The most in the first year and a half, during which 30 percent of the detailed study is from Berg (1939), who followed 6,162 smear- patients died. The death rate among new cases dropped in the positive patients for periods of up to twenty years. After two second and third rounds (32.4 percent over three and one-half years, 40.1 percent had died; deaths increased to 60.7 percent years in the first round and 33.9 percent over two years in the second round) because all round 2 patients received isoniazid Table 1-4. Estimated Incidence of All Form.s for one month. Taken together, these data suggest that, with- OfTublerculosis 1-4. EstimatedlIncidence Countout treatment, from 50 to 60 percent of tuberculosis patients of Tuberculosis in Developing Countries, 1990 will die. Cases The case-fatality rate for smear-positive patients is thought Incidence to be higher than for all forms of tuberculosis combined. Area Low Midpoint High (per 100,000) Rutledge and Crouch (1919) followed 1,229 patients who had Sub-Saharan 761,284 1,313,008 1,864,733 260 smear-positive tuberculosis and found that 66 percent of them Africa' were dead within four years. Lindhart (1939) found in Den- North Africa and mark that 66 percent of patients who were bacteriologically western Asia 116,754 323,320 529,885 120 positivedied.Berg'sresultsforsmear-positivepatientsprovide Asia 2,534,966 5,102,432 7,669,898 174 the most direct evidence on the higher case-fatality rate of South America 128,619 356,177 583,734 120 Central America smear-positive tuberculosis. A higher mortality of smear- and the positive patients has also been shown in the study in southern Caribbean 66,649 184,567 392,485 120 India (Olakowski 1973).Ofthe 126 bacillarypatientsdetected Total 3,608,272 7,279,504 10,950,735 175 in round 1, the death rate in the culture-positive and smear- negative group (62 patients) was 45.2 percent at five years and Note: Based on annual risk of infection for each region presented in table 53.1 percent in the smear-positive group (64 patients) for the 11-1, 1990 population, and incidence of thirtv-nine to fifty-nine cases per 100,000 population for each I percent annual risk of infection and also the same period. Case-fatality rates must also be expected to vary relationship between smear-positive tuberculosis and other forms of tubercu- between communities as a result of other factors, such as losis mn the United States by age, combined with the age distribution of nutrition and concurrent infections. Still, the above studies smear-positive tuberculosis in Tanzania. a. Includes cases attributable to dual Hlv/tuberculosis infections. provide a rough indication of the likely range in case-fatality Source: Authors. rates from tuberculosis. If the case-fatality rate for smear- Tuberculosis 239 positive tuberculosis is higher than for all forms of tuberculosis, specific countries to reflect newly registered cases.2 In addition, then the case-fatality rate of other tuberculosis on average the highest number of cases reported in the last ten years has must be somewhat lower. In other tuberculosis, however, some been adjusted upward by 20 percent for those regions with an forms, such as tuberculous meningitis, will cause 100 percent active private sector to try to account for those cases that are or very high case-fatality rates if the patients receive no treat- detected in the private sector but are not reported to the ment. For the rest of this chapter, we will assume that the government; in Asia, where data for some large countries may case-fatality rate for smear-positive tuberculosis is 60 to 70 include a large number of retreatment cases, we have not percent; for other tuberculosis as a whole, 40 to 50 percent; adjusted the figures by 20 percent. and for all forms combined, 50 to 60 percent. Separate estimates for the percentages of smear-positive and other cases are needed. The primary mode of case detection TUBERCULOSIS DEATH RATES IN DEVELOPING COUNTRIES. The varies across regions; for example, in Sub-Saharan Africa, tuberculosis death rates in developing countries cannot be as sputum microscopy is the main tool, whereas in China, much high as the incidence rates and a case-fatality rate of 50 to 60 greater emphasis is placed on chest radiography. We will percent imply, because a significant proportion of cases are assume arbitrarily that 50 percent of cases detected are smear detected by existing health services and the patients receive positive and 50 percent are other tuberculosis. This mix be- treatment. For all those cases that are estimated to be detected tween smear-positive and other tuberculosis cases is probably and receive treatment, we assume the case-fatality rate is an underestimate for Sub-Saharan Africa and an overestimate reduced to 15 percent after five years. For example, in the East for Asia. The detection rate of the various forms of tuberculosis African and British Medical Research Council survey in Kenya and the likely range of case-fatality rates discussed above can the case-fatality rate for patients receiving standard chemo- be combined to estimate the tuberculosis death rates from therapy was 13 percent after twelve months (East African and smear-positive tuberculosis and other tuberculosis. British Medical ResearchCouncil 1977,1979). In many coun- In table 11-6 we show estimated deaths each year from all tries, however, the case-fatality rate may be over 15 percent for forms of tuberculosis based on the calculations of the tubercu- those receiving chemotherapy, after five years of follow-up, losis death rates discussed above. The wide confidence inter- making the following estimates of mortality conservative. vals reflect the cumulative uncertainty in the parameters of the Estimates of the percentage of new cases that are detected estimation procedure. The midpoints of the confidence inter- and the patients treated are based on the number of cases of vals give a total number of deaths from tuberculosis in the tuberculosis detected that are reported by countries to the developing world of almost 2.7 million. Tuberculosis, there- World Health Organization (table 11-5; WHO 1988). Because fore, accounted for approximately 6.9 percent of all deaths in reporting is extremely variable, these estimates are based on the developing world in 1990 (United Nations 1986).3 the highest number of cases reported by each country for any year in the last decade. This basis is justified by the assumption AGE DISTRIBUTION OF TUBERCULOSIS DEATHS. To estimate the that year-to-year variation in the number of cases reported, age distribution of tuberculosis deaths, we must take into which can be greater than an order of magnitude, is due more to incomplete reporting of health service activities than to Figure 1 -5. Estimated Distribution of Deaths changes in the epidemiology of tuberculosis. The number of from Tuberculosis in the Developing World, 1990 new cases reported from some programs has been confused with the total number registered at the end of the year, which Percent includes old cases. We have adjusted the country estimates for 25 Table 11-6. Estimated Incidence of All Forms of Tuberculosis in Developing Countries, 1990 20 - Cases Incidence 15 Area Lou: Midpoint High (per 100,000) Sub-Saharan 10 - Africa' 300,604 585,591 870,578 116 North Africa and westem Asia 29,881 90,960 152,039 34 Asia 811,303 1,824,756 2,838,208 62 SouthAmerica 35,915 110,548 185,180 37 O - - - - - - --- Central America l l l and the Caribbean 25,217 79,966 134,715 52 0-14 15-24 25-34 35-44 45-54 55-64 65+ Total 1,202,920 2,691,820 4,180,720 65 Age (years) a. Includes deaths attributable to dual HIv/tuberculosis infections. Source: Authors. Source: Authors. 240 Christopher Murray, Karel St-vblo, and Annik Rouillon Table 11-7. Distribution of Tuberculosis Deaths in children under age fifteen, which is within the lower range by Age in Three European Countries before the for the three industrial countries in table 11-7. The compara- Availability of Anti-Tuberculosis Chemotherapy tively lower value may be a result of the higher BCG coverage (percent) in Tanzania now than in these countries at the time. Variation Czechoslovakia Norway Netherlands in the age pattern of tuberculosis deaths highlights the tenta- Age (years) 1940 1931 1941 1951 1931 1941 1951 tive nature of the estimates presented here. The basic conclu- - - - - - sion, however, that tuberculosis is concentrated in the adult 0-14 11.7 11.8 10.3 8.0 24.0 19.4 13.6 age groups, appears to be robust. 15-24 22.0 30.6 25.4 10.8 22.4 20.3 12.8 As implied in this discussion the age patter of tuberculosis 25-34 18.7 25.9 25.4 24.4 20.8 20.7 16.9 deaths shifts toward higher ages as the annual risk of infection 35-44 14.0 14.5 9.6 13.2 11.7 13.1 12.8 declnssingdtarom ther Uie d Ste s which easben 45-54 12.5 7.7 9.6 13.2 7.7 9.6 11.6 declines.UsingdatafromthelUnitedStateswhichhasbeen 54-64 11.4 5.0 6.6 13.2 6.3 8.2 13.4 adjusted to the 1990 age structure of the developing world, we 65+ 9.7 4.5 6.5 13.4 7.1 8.7 18.9 demonstrate in figure 1-6 how the mean age ofdeath increases Risk of as the risk of infection declines. The number of deaths in infection 5.5 - - - 3.7 1.8 0.5 children declines faster than the annual risk of infection; this relationship will become important when we consider the - Not available, cost-effectiveness Of BCG. Note: Based on age-specific mortality rates tor each country and the esti- mated population age structure for the developing world in 1990. a. 1938 figure. Trends in Incidence and Mortality Source: Authors. We have estimated cases and rates of tuberculosis in the year consideration the age distribution of new cases and the relation 2015 using the midpoints of the ranges of the annual risk of between case-fatality rates and age. Clearly, the relation is infection in table 11-1, population projections, and the rates complex; for example, the death rates may also vary by age of decline in the annual risk of infection, also reported in table because certain age groups or sexes may be more likely to seek 11 - I (table 11 -8). These estimates are based on the assumption treatment and be cured. For example, comparing the distribu- that the rates of decline in the annual risk of infection observed tion of smear-positive patients by age and sex between Malawi between 1970 and 1985 will continue into the future. In other and Tanzania, it is evident that women in Tanzania are much words, the projections are based on the assumption that the less likely to seek care than men. The reduction of female socioeconomic changes and tuberculosis control activities that case-detection rates by sex bias in access to care is probably caused the decline in the risk of infection in the last two quite widespread, especially in South Asia. Tuberculosis case- decades will continue at the same rate. Such projections suffer fatality rates tend to increase steadily at older ages (Berg 1939; Styblo 1984). We have derived from the Berg data the relation Figure 11-6. Shifting Age Structure of Tuberculosis between age-specific case-fatality rates with some data from Deaths in the United States, 1937 Styblo (1984) on mortality in the age group zero through four- teen.4 Figure 11-5 provides the crude estimates of the age Percent pattern of tuberculosis deaths in a country with an annual risk 80 of infection of 1 to 2 percent, where the probability of detec- tion is equal for smear-positive tuberculosis across all age 70 - groups and equal for other tuberculosis across all age groups. 60 - This estimated pattern can be compared with the age distri- bution of tuberculosis deaths in Westem countries when the 50 annual risk of infection was similar to that now seen in the 40 developing world. The age distribution of tuberculosis deaths adjusted to the age structure of the developing world in 30 Czechoslovakia, Norway, and the Netherlands (Tuberculosis 20 Surveillance Research Unit 1966) is illustrated in table 11-7. The percentage of deaths in children under fifteen ranged from 10 approximately 10 to 20 percent. However, overall, the tuber- 0 culosis death rates in children were considerably lower in the 1937 1942 1947 1952 1957 Netherlands than in Czechoslovakia, even at higher risks of Age (Years) infection. Clearly, there are other variables that are significant determinants of the reported age distribution of tuberculosis r J 0-14 15-24 M 25-44 M 45+ death rates. One explanation may be the high rates of M. bovis infection in the Netherlands at the time. According to our Note: Adjusted for age structure changes. estimates for Tanzania, I I percent of tuberculosis deaths occur Source: Authors. Tuberculosis 241 Table 11-8. Projected Cases and Deaths for all Of avoidable adult deaths, 26 percent are probably due to Forms of Tuberculosis in 2015 tuberculosis. Cases The consequences of adult death from tuberculosis on chil- Smear- dren and other dependents can also be great. Studies have Area positive Other Total Deaths shown that, when a mother dies, her children suffer higher rates of mortality (Greenwood and others 1987). One can Sub-SaharandAfric& 1,270,366 1,549,846 2,820,212 1,257,791 speculate that similar relationships exist for patemal death. North Africa and westem Asia 128,607 156,900 285,507 80,288 Several studies from industrial countries have shown that Asia 1,873,615 2,285,810 4,159,424 1,487,316 tuberculosis is concentrated in lower socioeconomic groups, South America 98,667 120,373 219,040 67,965 those households least able to cope with the burden of tuber- Central America and culosis. Pryer (1989) found that in households in which one the Caribbean 79,980 97,575 177,555 76,919 parent suffers from a serious debilitating disease, such as tuber- Total 3,451,235 4,210,504 7,661,738 2,970,279 culosis, children are two and one-half times more likely to be Note: These projections are based on the following assumptions: (a) the severely malnourished. Because tuberculosis deaths are con- current rate of decline in the annual risk of infection will continue over the centrated in the segment of the population that is economi- next twentv-five years except in Sub-Saharan Africa, where it will not change cally most productive, the economic cost of tuberculosis in lost because of the Hiv epidemic; (b) the percentage of cases detected will remain production must be greater than that of a disease that exclu- the same in each region; (c) the percentage of patients treated with standard and short-course chemotherapy, and thus the population cure rate, will remain sively affects children or the elderly. constant. Source: Authors. Prevention from all the same limitations as any projection of current Before discussing specific measures to prevent or treat tuber- trends. For Sub-Saharan Africa, we have assumed no net culosis in developing countries, we will summarize the ratio- increase in the seroprevalence of Hiv and no net decline in the nale for tuberculosis control programs in countries with a low annual risk of infection in making these projections to repre- prevalence of HIV infection. The presence of a significant sent the potential contribution of the HIV epidemic. This is an proportion of the population infected with HIV may change the extremely conservative assumption that probably underesti- strategy for control. mates the contribution of Hlv-associated tuberculosis to the * Unlike many other infectious diseases seen in develop- total tuberculosis caseload and the real potential for an in- ing countries, tuberculosis can be controlled with existing crease in the risk of infection in the Hiv-negative population, tools because the infectious agent is almost exclusively given an increasing number of sputum-positive patients. Ac- cording to these conservative assumptions, tuberculosis will noninf eth iose remain a significant problem in all developing world regions nonmnectious. referred to in table 11-8. In Africa, population growth alone * The detection of infectious, particularly smear-positive, will lead to an absolute increase in the number of cases. cases of pulmonary tuberculosis and their cure are the key to effective prevention and control of the disease, both in Social and Economic Costs industrial and developing countries. In addition, detection and treatment of cases reduce suffering and if adequately There are few if any studies of the actual costs or consequences applied, very much lower the death rate of tuberculosis. of tuberculosis on the family, community, or economy in devel- * Because a balance exists in developing countries be- oping countries. The special burden of ill health and death tween the tubercle bacillus and people in the absence of caused by tuberculosis, however, follows from the age distribu- human-made interference (that is, case finding and chemo- tion of its incidence. Although morbidity and mortality in any therapy), any reduction in the sources of infection will age group have significant social and economic costs, the death inevitably improve the epidemiological situation. If all or of adults in their prime, who are the parents, community nearly all smear-positive cases of pulmonary tuberculosis leaders, and producers in most societies, cause a particularly diagnosed at present in any developing country could be onerous burden. The incidence of tuberculosis is concentrated rendered noninfectious, the risk of tuberculous infection in adults age fifteen through fifty-four. For example, whereas would immediately start to fall. A decrease in the annual the overall incidence in Africa of tuberculosis in 1985 is esti- risk of tuberculous infection of 4 percent or more would not mated to be 260, in adults it is approximately 390 per 100,000. only result in a decrease in the incidence rate of the disease One of the greatest costs to society and the economy from but would also outweigh increases in the population; conse- tuberculosis is mortality. It has been estimated that there are quently the absolute number of smear-positive cases would just under 10.6 million deaths in adults age fifteen through fall as well. A 5 percent decrease in the risk of infection each fifty-nine in the developing world (Murray and Feachem 1990). year would ensure that the tuberculosis problem in a given Of these, our figures suggest that approximately 18.5 percent community or country would halve itself about every four- are due to tuberculosis. Not all these deaths are preventable. teen years. 242 Christopher Murray, Karel Styblo, and Annik Rouillon * Reliable diagnostic tools that enable detection of the prevalence of HIV infection among mothers. The WHO Ex- great majority of smear-positive cases of pulmonary tuber- panded Programme on Immunization, which is responsible for culosis and highly efficient chemotherapy regimens that can the program of vaccination against six selected childhood cure nearly all discovered cases of tuberculosis are available. diseases in the world, has been continuing BCG vaccination of There are three main strategies for preventing tuberculiosis: newborns and small children, including those whose mother is known to be or suspected of being infected with HIV. As of the BCG vaccination, chemoprophylaxis, and decreasing sources of t.m of wrtn,eiec.ean icnlsv eadnh infection through case treatment. Each will be discussed int rate of adverse reactions after BCG immunization among asymp- turn. tomatic HIv-infected individuals. The vaccine should be with- held from individuals with symptomatic HIV infection (WHO BCG Vaccine 1987). The effect of mass BCG vaccination on the epidemiological The bacillus of Calmette and Guerin (BCG) was developed in situation of tuberculosis was overestimated until the mid- 1921. Since that time it has become one of the most widely 1970s (Styblo and Meijer 1976). As mentioned earlier, tuber- used yet controversial vaccines. Although BCG coverage has culosis is largely transmitted by persons with sputum been up to now quite high on average compared with other smear-positive pulmonary tuberculosis. From the age distribu- immunizations, the effectiveness of BCG in preventing tuber- tion of smear-positive patients, it is clear that even complete culosis in adults remains controversial. From clinical trials BCG coverage can have little effect on the annual risk of conducted in the United Kingdom and in the United States it infection. Total coverage with BCG, however, will have a was found that BCG was up to 80 percent effective (Aronson, significant effect on tuberculosis mortality in children, if BCG Aronson, and Taylor 1958; Great Britain Medical Research is 40 to 70 percent effective, as we have assumed. Based on the Council 1972). Important vaccine trials in southern India, assumptions discussed above, complete coverage could reduce however, revealed no effectiveness of BCG (Tuberculosis Pre- total tuberculosis mortality by 4 to 7 percent. The vaccine will vention Trial 1979; Tuberculosis Prevention Trial, Madras most likely have very limited effect on the remaining 90 1980). Reports from a variety of prospective trials in the percent and more of tuberculosis mortality. Evidently, the industrial world and more recent case-control studies in devel- expansion of BCG coverage alone cannot or should not be the oping countries state effectiveness ranging from 0 to 80 percent sole means employed to control tuberculosis in any community. (Clemens, Chung, and Feinstein 1983; Smith 1987). Many explanations and theories have been advanced to Cost-Effectiveness of BC(G explain this variance, including differences in strains of BCG, infections with other mycobacteria, and differences in suscep- For two principal reasons, generalizable estimates of the cost- tibility resulting from factors such as nutritional status (Fine effectiveness of BCG cannot be made. First, there may be 1989). Although there is no consensus on the effectiveness of substantial differences in the computed average and marginal BCG, we will assume that BCG is between 40 and 70 percent costs of BCG programs, depending on the program considered. effective in preventing tuberculosis in children age zero Second, the cost-effectiveness of BCG is inversely proportional through fourteen when given at birth. Some would argue that to the annual risk of infection. BcG given at birth may protect beyond fifteen years; there When more than one vaccine is given at the same time, is, however, no evidence of this, especially in developing average costs for delivering each particular immunization are countries. often calculated by dividing the cost per client contact by the The BCG vaccine is given as early as possible in life, prefera- number of vaccinations received. Thus the difference between bly at birth, in the vast majority of developing countries. marginal costs and average costs for a BCG program will depend Serious consideration might also be given to "indiscriminate on whether BCG is delivered in an independent campaign, in a (re)vaccination" (that is, without prior tuberculin testing) at contact with mother and child, or along with other immuni- older ages, irrespective of vaccination at birth. Depending on zations such as the first DPT (diphtheria-pertussis-tetanus) vac- the feasibility of coverage, BCG (re)vaccination could be given cination. The Expanded Programme on Immunization has not, to children entering and leaving school, pregnant women unfortunately, collected data on how BCG is delivered in each during prenatal care visits to health facilities, or to the general country. We conclude that the marginal cost-effectiveness of population during routine contacts with health workers. For expanding BcG will necessarily depend on the location and example, tetanus toxoid is now considered by many to be an timing of vaccination in a particular country. integral component of prenatal care; RCG could be delivered at As the annual risk of infection declines, if all else remains the same time for only a small increase in the total cost. The the same, the cost of immunizing all newboms does not actual effect of BcG (re)vaccination at older ages has not been change. The benefits of BCG immunization in cases or deaths studied; there seems little reason to believe that it would be averted, however, will decline inversely to the risk of infection. harmful, however, and it may have some beneficial effect. For example, as the risk of infection declines from 2 percent to Still, we must realize that vaccination of newborns with BcG I percent, the cost per death averted will more than double. is a problem in those developing countries where there is a high The increase in the cost per death averted is greater than the Tuberculosis 243 decline in the risk of infection because the age distribution of that delivery of BCG to adults could be feasible at the same deaths also shifts away from children as the risk of infection average or marginal costs as its delivery to infants. declines (see figure 11-6). The expected relation between the risk of infection and the cost per death averted by BCG is Chemoprophylaxis illustrated in figure 11-7. In only one study has an attempt been made to cost a BCC Secondary prevention of clinical tuberculosis can be accom- program and estimate its effect in a developing country. Bar- plished by treating patients with tuberculosis infections. Che- num, Tarantola, and Setaidy (1980) estimated the cost of moprophylaxis is applied either to freshly infected so-called operating a BCG program alone and also the marginal cost of tuberculin converters or to those who have been infected with adding a BCG program to an existing DPT program. Their esti- virulent tubercle bacilli in the more distant past. The latter mates of deaths averted were based on local incidence and either do or do not have abnormalities in the lungs on x-ray. case-fatality rates of tuberculosis and an assumed effectiveness Tuberculin converters undoubtedly represent a very reward- for BCG of 50 percent. Using their original data, we have ing group in terms of chemoprophylaxis results and thus che- recalculated the cost per discounted death averted in U.S. moprophylaxis policy has been adopted as a routine procedure dollars.5 Deaths prevented by BCG vaccination now occur over in a number of low-prevalence countries. Mass chemoprophy- the next fourteen years; these are discounted to present value laxis of converters is impossible, however, since their identifi- for comparison with interventions that avert deaths in the cation depends on repeated tuberculin tests of the population. current time period. The cost in 1986 U.S. dollars per death However, a selective search for converters in high-risk groups, discounted at 3 per cent was $644 for the BCG program alone such as close family contacts of smear-positive sources, is a and $144 for the marginal BcG program. At the time in Indo- feasible altemative. Asdiscussed later, 6 to 10percent of recent nesia, survey data suggested that the risk of infection was infections evolve into clinical tuberculosis. In developing approximately 3 percent: regional surveys reported annual countries, where large percentages of the population have been risks of infection of between 2 and 4 percent (Cauthen, Pio, infected, the International Union against Tuberculosis and and ten Dam 1988). It must be stressed that these estimates of Lung Disease (IUATLD) recommends chemoprophylaxis only cost-effectiveness do not take into consideration the potential for all non-BCG-vaccinated children age five years or under, benefits of BCG in reducing leprosy (Fine 1989). with no symptoms of tuberculosis. In children with symptoms, With no evidence at all on the efficacy of indiscriminate standard treatment should, of course, be given. adult (re)vaccination with BCG, it is difficult to discuss the Chemoprophylaxis in tuberculin-positive subjects who cost-effectiveness of adult BCC vaccination. Because tubercu- have not developed clinical tuberculosis would reduce the losis mortality is concentrated in the young adult ages, revac- number of sources of infection, if given for six to twelve cination, if it proved to be as effective as the vaccination of months. In most developing countries, this group is very large, infants, would be more cost-effective. This, of course, assumes and resources would be far better directed to the detection of cases and to treatment. Still, chemoprophylaxis might play a very important role both in industrial and developing coun- Figure 11-7. Cost-Effectiveness of BCG and Case tries in subjects with the dual Hlv and tuberculous infections Treatment as a Function of the Annual Risk but without clinical and bacteriological signs of tuberculosis. of Infection Research on HIV chemoprophylaxis is under way in several Sub-Saharan African countries. Cost per death averted (U.S. dollars) Studies in industrial countries have found cost-effectiveness ratios per case averted to be greater than $7,000 for a twenty- 2,500 four-week regimen (Snider, Caras, and Koplan 1986). Without accurate data to review the cost-effectiveness of chemoprophy- 2,000 - lax is in developing countries, we can only make some compar- isons with the costs per case treated. Because only 6 to 10 1,500 - percent of those who have recently become skin-test positive BCG develop clinical diseases, 10 to 16.7 recent tuberculin-positive 1,000 patients must be given chemoprophylaxis to prevent one case of tuberculosis, assuming prophylaxis is 100 percent effective. 500 In tuberculin-positive patients as opposed to new converters, Case treatment the ratio would be one or two orders of magnitude higher 0 I I I I because the long-term breakdown rate is only 25 to 40 per 0.5 1 1.5 2 2.5 3 3.5 100,000 per year. The drug costs for chemoprophylaxis are lower than for treatment, but the costs of administration, Risk of infection (percent) screening, transport, delivery, and monitoring would be similar. Thus, chemoprophylaxis is unlikely to be more cost- Souree: Authors. effective in developing countries than case treatment of 244 Christopher Murray, Karel StNblo, and Annik Rouillon patients presenting with symptoms suggestive of tuberculosis A certain percentage of these new infections or reinfections as discussed later. One exception may be in children under five caused by a smear-positive person will in turn break down and exposed to an adult with active smear-positive pulmonary lead to clinical tuberculosis. Reference is made to three reports tuberculosis. of newly infected persons to determine the percentage that developed clinical tuberculosis: the British Medical Research Decreasing Sources of Infection Council study (Sutherland 1976) found that 8.1 percent of converters developed clinical tuberculosis within fifteen years; The transmission of tuberculosis appears to take an extremely in Saskatchewan, 6.4 percent of recently infected individuals regular and stable course in comparison with most other infec- developed clinical tuberculosis within a few years after primary tious diseases such as malaria, schistosomiasis, or cholera. Each infection (Barnett, Grzybowski, and Styblo 1971); and a Tu- infectious or smear-positive person infects many others each berculosis Surveillance Research Unit study of European data year. The number of new infections caused each year by a found that 6.0 percent of converters developed bacillary tuber- person with smear-positive tuberculosis can be estimated from culosis in five years (Sutherland 1968)). For the purposes of survey data on the number of new infections and the preva- modeling transmission, we will assume that from 6 to 10 lence of smear-positive tuberculosis. It has been estimated from percent of new infections will eventually develop some form data from developing and industrial countries that an undiag- of clinical tuberculosis.6 In figure 11-8 we show that the new nosed and untreated smear-positive source of tuberculous in- infections could lead to 100 cases of smear-positive tuberculo- fection would infect on average between ten and fourteen sis and 122 cases of smear-negative or extrapulmonary tuber- persons per year (Sutherland and Fayers 1975; Styblo 1984). culosis. The transmission cycle would then repeat itself over Each smear-positive person continues to excrete the bacillus and over. for an average of two years, thus leading to the well-known 2:1 The steady state illustrated in figure 11 -8 is a close approx- ratio of prevalence to incidence (Styblo 1984). A person imation of reality in mostofthedevelopingworld. Data on the with smear-positive tuberculosis will be responsible for ap- annual risk of infection summarized in table 11-1 showed that proximately twenty to twenty-eight new infections before the annual decline in the risk of infection for Africa and Asia either dying or becoming smear negative. Figure 11-8 is an was between 1 and 3 percent. Population in these regions is illustration in a schematic form of the nature of tuberculosis also growing at an annual rate of I to 3 percent, so the absolute transmission. number of cases of smear-positive tuberculosis remains nearly Figure 11-8. Tuberculosis Transmission Schematic Infections 2,400 2,400 2,400 Smear-positive cases 1 100 100 Other cases 1 1 1 Deaths 50 50 5 50 50 50 Source: Authors. Tuberculosis 245 constant. In other words, each smear-positive case of tubercu- Figure 11 -9. Age-Specific Tuberculosis Death Rates losis must lead on to approximately one more smear-positive in the United States, 1900-50 case after a round of transmission. The best way to prevent tuberculosis, therefore, is to interrupt the transmission cycle. Deaths (per 100,000 population) As early as 1961, Crofton (1962) realized that chemotherapy 350 for smear-positive patients, which rapidly renders them non- infectious, is the best way to reduce the transmission of the 300 disease. 250 \ N \ Risk Factors 200 .\ / /9001 0 . _ . The history of tuberculosis in industrial countries clearly 150 / , 192--- - --- - - demonstrates an important role for socioeconomic change in 100 - , ._ the decline of tuberculosis. In figure 11-9 we show how the -- --.-- 19S01----- age-specific tuberculosis mortality rates in the United States e _ = declined from 1900 to 1950, before chemotherapy was avail- 0 , , able. This decline appears to have been due to a decrease in 1-4 5-14 15-24 25-34 35-44 45-54 55-64 65-74 75-84 85+ transmission, because the case-fatality rates remained constant Age (years) in this period (Drolet 1938). Reduction in transmission may have been a result of improvements in housing, nutritional standards, general health, and perhaps most important the policy, instituted at the turn of the century, of isolating infec- tious sources of tuberculosis in sanatoriums. During the first between HIV and tuberculosis infections, particularly in coun- four decades of this century the annual risk of infection in most tries where both infections are prevalent, appear more and industrial countries was falling about 3 to 5 percent per year more clearly. In several African countries, a considerable in- (Styblo 1984). crease in newly discovered cases has already been documented With the introduction of chemotherapy, tuberculosis mor- (for example, Tanzania and Malawi). Tuberculosis is a fre- tality rates declined at a faster rate, approaching 10 to 12 quent presenting symptom of HIV infection and AIDS in these percent. In some developing countries that have reliable reg- and other countries. istration of vital statistics, such as Chile, tuberculosis mortality The WHO Global Programme on AIDS, the WHO tuberculosis was fluctuating and high before 1945, after which it declined unit, and the IUATLD have initiated studies on the study on the precipitously. The most famous example of the effect of che- various facets of the interactions of these two diseases, partic- motherapy is in the Eskimos. In populations in Canada and ularly, the epidemiological effect ofi4rv infection on the overall Greenland, the risk of infection before 1950 approximated 25 tuberculosis situation. Researchers suspect not only that HIV percent per year. After aggressive case detection and treatment infection will increase the incidence of tuberculosis in individ- were instituted, the annual risk of infection declined by 17 to uals already carrying the tubercle bacillus, as a result of the 20 percent per year. Tuberculosis mortality has followed a decrease in immunity, but that the excess sources of infection similar precipitous decline (Johnson 1973; Grzybowski, will result in an increase in the annual risk of tuberculous Styblo, and Dorken 1976). infection in the country. The experience of industrial countries, and some disadvan- The two other specific potential risk factors for developing taged groups in these countries, indicates that although socio- tuberculosis are mining and associated silicosis and malnutri- economic changes can reduce the transmission of tuberculosis, tion. The association between silicosis and tuberculosis has widespread use of chemotherapy can greatly accelerate the long been noted (Brink, Grzybowski, and Lane 1960). This decline in tuberculosis. In many parts of the developing world relationship may explain in part the high incidence of tuber- tuberculosis is declining at a rate of I to 2 percent per year. culosis in southern Africa and the Altiplano of South America, Improved case detection and case treatment could realistically where a significant proportion of the adult male population accelerate that decline to 5 to 10 percent per year. works in mining and where there is an elevated prevalence of Three specific risk factors for developing tuberculosis de- silicosis (De Beer 1984). Although the validity of this hypoth- serve note. First, for the last few years we have been witnessing esis has not been rigorously tested, the association between the strongest risk factor for developing tuberculosis in individ- silicosis and high incidence of tuberculosis is accepted by many uals remotely or recently infected with tubercle bacilli-HIv authors. infection. The mechanism is easy to understand: the decrease Malnutrition has been associated with increased inci- in immunity caused by HIV infection results in the flaring up of dence of and deaths from tuberculosis. During World War this virulent agent, the tubercle bacillus. II, tuberculosis rates increased in European countries af- Tuberculosis is thus one of the diseases in developing coun- fected by the war, particularly in some special groups, such tries most influenced by the HIV pandemic. The interactions as in camps (Cochrane 1948) and in the Warsaw Ghetto 246 Chnstopher Murray, Karel St-blo, and Annik Rouillon (Schechter 1953). There has been, however, no careful, con- infected byaperson with smear-positive tuberculosis before its trolled demonstration of this association, because crowding, detection, this number may be four or five in developing recirculating air, and poor sanitation seem to be at least as countries, because of a higher number of close contacts. No important. Conversely, despite improvements in nutritional contacts will be infected after the start of adequate chemother- intake, the case-fatality rate for tuberculosis remained un- apy. The validity of these assumptions depends on local con- changed in the United States, England, Denmark, and other ditions, cultural perception of disease, access to care, and the industrial countries from the turn of the century until the effectiveness of health services. introduction of chemotherapy. Improvement in nutritional Regardless of the technology used, active case detection is status may alter the probability of those who are infected more expensive per case detected because the yield of tuber- developing clinical tuberculosis, and it may decrease the culosis per patient screened is lower. For example, if the breakdown rate from infection to disease. As a pragmatic incidence of smear-positive tuberculosis is 100 per 100,000 means of preventing tuberculosis, however, improving the people, then the sputum of more than 1,000 people would have general nutritional status of the population holds little prom- to be screened to detect one case of smear-positive tuberculo- ise, unless it is combined with an efficient program of case sis, provided it is the general population that is being screened. detection and treatment. If specific high-risk groups were identified, the yield would clearly be higher. For comparison, the use of sputum micros- Curative Care copy to screen patients who present with cough in Tanzania yields one patient in ten with smear-positive tuberculosis. The subject of curative care can naturally be divided into Another argument against active case detection is that persons tuberculosis detection and chemotherapy. We will address actively identified as being infected may be less likely to each of these in turn, highlighting the policy options. comply with long drug regimens. Clearly, they did not yet consider their health to be impaired enough to seek treatment. Case Detection Moreover, a proportion of smear-negative persons with few or no clinical symptoms cure spontaneously and in a number of There are two main issues in detecting cases of clinically cases the disease is in regression (Styblo and others, 1969; significant tuberculosis: active as opposed to passive detection Meijer and others 1971; National Tuberculosis Institute, Ban- strategies and the choice of diagnostic technology. "Active galore 1974). In developing countries, active case finding was detection" means an attempt to screen the population at large studied by the Kenyan and British Research Councils in the or target populations, such as military recruits, for evidence of late 1970s and early 1980s. These studies have yielded seven tuberculosis. "Passive detection" means screening and diagnos- reports, and the conclusion in the last study is that a patient ing only those patients who visit a health service provider suffering from symptoms suggestive of pulmonary tuberculosis because of symptoms suggestive of tuberculosis. In the 1950s nearly always seeks medical advice from a health unit, usually and 1960s, the choice between active and passive detection in several times. In many instances, however, health workers at industrial and developing countries was a controversial topic the peripheral level do not think of tuberculosis and do not (Styblo and others 1969; Meijer and others 1971; WHO 1974; examine the sputum themselves or do not refer the patient to Toman 1979; Styblo and Meijer 1980). In the last two decades, the nearest microscopy center for sputum examination for a consensus for passive case detection of tuberculosis in all tubercle bacilli. In many developing countries, public trans- countries has developed-both WHO and the IVATLD advocate portation is rudimentary; even if available, it is not always this policy. affordable to poor people. Moreover, the Kenyan studies have Three assumptions underlie the wide acceptance of passive shown that active case finding, except in health units, is not case detection as the primary strategy in tuberculosis control. feasible. First, 90 percent of patients with smear-positive pulmonary The second issue in case detection is the choice of technol- tuberculosis have objective symptoms, such as cough, fever, ogy. At present, the main options are sputum microscopy, loss of weight, sputum, or hemoptysis. These symptoms de- sputum culture, and radiology. To illustrate the yield and likely velop quite soon after the onset of the disease, prompting the cost of case detection using microscopy (Ziehl-Nielsen stain), patient to seek medical advice. Second, the great majority of we shall examine data from the National Tuberculosis and sputum smear-positive tuberculosis cases develop in a shorter Leprosy Programme in Tanzania. In that country, one in seven period of time than the shortest feasible interval between two people who are suspected of having tuberculosis and are mass radiography survey rounds. That is why smear-positive screened is identified as having smear-positive tuberculosis. tuberculosis cases were detected outside (and usually earlier Normally, three smears are conducted on each patient. The than) the periodic case-finding campaigns conducted by the cost of supplies and reagents alone for these thirty smears is regular health services.Third, appropriate diagnostic and cura- $2.50. A microscopist can examine about twenty sputa per day. tive care ought to be physically, socially, and economically The effective cost per case detected using sputum microscopy available. Most infections, before chemotherapy is instituted, in Tanzania is $5.46, including the depreciated cost of the would therefore occur within the family. Whereas in industrial microscopes. In Tanzania, three sputa are examined to increase countries it is estimated that two to three persons would be test sensitivity; the increased sensitivity achieved with the Tuberculosis 247 third smear is in fact small and could be sacrificed to reduce 1989). If these technologies yield new tools that can be inex- the cost. pensively applied in developing countries, passive case detec- Sputum culture is used to diagnose pulmonary tuberculosis tion may be improved, especially for smear-negative and in those patients who produce too few bacilli to be detected on extrapulmonary tuberculosis, which cannot be diagnosed by a smear, to confirm sputum microscopy, and to characterize the sputum microscopy. Active case detection in some high-risk type of mycobacterium. Because culture takes several weeks to groups would perhaps become feasible. yield results, it is not useful as a primary diagnostic tool in A limited number of interventions are available to improve developing countries. For retreatment cases, however, culture the effectiveness of passive case detection. The factors that and sensitivity may be very important to determine the most would be most important in improving such effectiveness are cost-effective drug regimen. a high cure rate of diagnosed cases and a friendly relationship There are at least two different roles for chest radiography between the treating health staff and the patient. Public edu- in the diagnosis and treatment of tuberculosis. First, for diag- cation can increase general awareness of the symptoms of nosing smear-positive tuberculosis, chest radiography can be tuberculosis and encourage those suspected of having it to seek used to identify a group with a much higher probability of being medical advice. Improved diagnostic skills of primary health smear positive. The resulting yield on sputum microscopy can care providers, transport of sputum or a patient to a microscopy be increased and many fewer total smear examinations under- center, and availability of x-ray facilities can also improve the taken. In areas where the prevalence of smear-positive tuber- detection of both smear-positive tuberculosis and other tuber- culosis is low, the increased yield may be important for culosis. Finally, if diagnosis and adequate treatment are free, as maintaining the quality of sputum examinations. Unfortu- recommended by wHo and IUATLD, more patients will seek early nately, x-rays are not 100 percent sensitive in detecting tuber- care. culosis, so an initial screening with chest radiography will decrease the total yield as compared with sputum culture-for Treatment example, in Bangalore, x-ray was 87 percent sensitive. Second, chest radiography is essentially the only available tool for use The first antituberculosis drug, streptomycin, became avail- in the periphery of most developing countries for the diagnosis able in the early 1940s. In 1952, three antituberculosis drugs of smear-negative tuberculosis. Sputum culture, although the were available (streptomycin; para-aminosalicylic acid, or PAS; gold standard for smear-negative diagnosis, takes too long and and isoniazid) which were able to cure virtually all patients is too difficult to implement in the periphery of most develop- however severe their disease, provided that their bacilli were ing countries. The role of chest radiography depends on the initially sensitive to the above drugs. Such results were desirability of detecting and treating smear-negative tubercu- achieved in Edinburgh in 569 cases as far back as 1953 and losis; this subject will be discussed later in relation to cost 1954 and in 2,506 cases treated in 1955-56 (Crofton 1961). effectiveness. Since then a variety of chemotherapeutic agents have been The cost per case of tuberculosis detected by x-ray depends developed. The six drugs recommended by WHO and the IUATLD largely on three factors. First is the prevalence among symp- and most commonly used in developing countries for tubercu- tomatic patients presenting with x-rays suggestive of tubercu- losis are isoniazid, streptomycin, thiacetazone, ethambutol, losis at health services. This can reportedly vary from one in rifampicin, and pyrazinamide. These drugs are used in a host two in China to a more realistic rate of one in four or five in of combinations for different durations (table 1 1-9). Tanzania. The second factor is the cost of x-ray machines, Despite the availability of powerful and potentially effective which are expensive capital investments. The depreciated antituberculosis drugs, tuberculosis treatment programs in capital cost per patient screened depends on how much the most developing countries have not been very successful. machine is used. A district-level machine used for the diagno- Overall cure rates for most national programs in poor develop- sis of many diseases is likely to be less expensive per patient ing countries are below 50 percent. Evidently, the "standard" than an underused machine dedicated to the detection of chemotherapy (isoniazid, streptomycin, and thiacetazone) tuberculosis. Considerations of depreciated capital cost will recommended by the wiio Expert Committee on Tuberculosis require x-rays to be used at a level in the health services that (WHO 1974) for use in developing countries is presently, and has an adequate patient load. On the basis of hypothetical cost probably will be in the future, beyond the organizational calculations, the cost per case of tuberculosis detected using resources of many of them. This was clear to Canetti more than chest radiography varies from $6 to $10 in China and Tanza- thirty years ago. As the principal reporter to the Panel on nia, assuming a caseload of 5,000 x-rays per year on a machine Eradication of Tuberculosis he stated: "On the global level, that costs $50,000 and lasts ten years. More research is needed and among the efforts required to make some headway towards on defining the true average and marginal costs of deploying tuberculosis eradication, an absolute priority stands out im- chest radiography in various conditions for the diagnosis of peratively: to develop chemotherapeutic methods adapted to smear-negative pulmonary tuberculosis. the conditions prevailing in underdeveloped countries" New diagnostic technologies based on the enzyme-linked (Canetti 1962). immunoabsorbent assay or DNA probes for mycobacterial DNA In the 1960s and 1970s, experience showed that Canetti was or RNA are currently being investigated (Daniel 1989; Bloom right. In many poor developing countries in Africa, many parts 248 Christopher Murray, Karel St.blo, and Aninik Rouillon Table 11-9. Examples of Anti-tuberculosis factors today in nearly all contexts is the cure rate, which Chemotherapy Regimens Used in Developing decisively influences the remaining two factors. Countries One determinant of the cure rate is the biological effective- Regimen Duration (months) ness of WHO standard and short-course chemotherapy given under ideal conditions of 100 percent compliance. With short- New smear-positive standard course chemotherapy, after two months of treatment 80 to 90 2SH/10TH 12 percent of smear-positive pulmonary cases will have converted 2SH/ I OsH7 1 to sputum-negative status. Under WHO standard therapy, after 2 -1,H/10S,H1 1 2 two months 50 percent will remain smear positive. The "per- Short-course manent" cure rate is a more important aspect of the treatment 2SHRz/4HR or 2EHRZ4HR 8 regimens. In figure 11-10 we show the percentage of patients 2HRZ/4HR 6 who will remain or become smear positive, say, two years after 2Rz/4HiR3 6 the start of the (first) treatment (with no retreatment during the first two years), provided that chemotherapy is discon- 2STH/10TH 1e tinued at each point in time. We shall refer to them as 2SHRz/6-1H 8 "failures.)" (Patients who remained or became smear positive Retreatmet 8and died during the first two years will also be referred to as Rerreatment 6 failures. Under short-course chemotherapy (for example, 2SHRZE/1HRZE/5H R3E 8 2SHRZ/6TH [see table 11-9]), about 40 percent of those who 2RLZ5Hdiscontinue chemotherapy at two months may be failures Note: s = steptomycin I gm; H isoniazid 300 mg; R = rifampicin 450/600 compared with approximately 10 percent of those who com- mg; z = pyrazinamnide 1500/200 mg; E = ethambutol 25 mg/kg; T = thiacetazone plete six months of short-course chemotherapy. Under stan- 150 mg. Subscripts indicate the number of times each week drugs are given dardechemotherapyo(orhexample,seschemotherathe failuretrat during intermittent therapy. dard chemotherapy (for example, 2STH/10TH), the failure rate Source: Authors. in patients who discontinue standard chemotherapy after two months may reach 65 to 70 percent, and in those who complete of Southeast Asia, and certain parts of Latin America, signs of six months it might be approximately 50 percent. The failure improvement in the epidemiological situation of tuberculosis rate begins to drop significantly on WHO standard chemo- were the exception, despite widespread attempts at disease therapy only after six months. By twelve months, under ideal control. The most important reason for the failure of such conditions of 100 percent compliance, approximately 10 per- control programs was the low cure rate. As Canetti postulated, cent will become failures if treatment is stopped. unless a large increase in the cure rate for smear-positive Since standard and short-course chemotherapy both give pulmonary tuberculosis can be achieved, there will be no high cure rates and do not lead to secondary resistance in marked improvement in the tuberculosis problem in many developing countries in the foreseeable future. Figure 11-10. Patients Failing Therapy after Although there are many interesting issues in tuberculosis Two Years of Follow-up, as a Function of Months treatment, in this discussion we will stress the choice between of Chemotherapy WHO standard chemotherapy regimens that last from twelve to eighteen months and use fewer and cheaper drugs (isoniazid, Failures (percent) streptomycin, and thiacetazone), and short-course chemother- apy that lasts from six to eight months and uses multiple and 100 more expensive drugs (rifampicin and pyrazinamide). To com- pare these two strategies with chemotherapy, we must examine 80 the relative effectiveness of each and the relative costs of each. Because of the great diversity in effectiveness and costs be- Standard tween countries, the emphasis will be on the key determinants 60 chemotherapy of the effectiveness and costs of the two regimens. It should be stressed that the regimen with a higher cure rate leads to a more 40 - rapid reduction in the risk of tuberculosis infection and the incidence of active tuberculosis. 20 S tu chemotherapy\ Effectiveness of Chemotherapy 0 l l l l l 2 4 6 8 10 12 14 The effectiveness of standard and short-course chemotherapy depends on three main factors: the cure rate, acquired drug Treatment received (months) resistance, and the effect on the trend of the risk of tuberculous infection. Without question, the most important of these Source: Author. Tuberculosis 249 controlled clinical trials, compliance is the most important combined tablets daily for at least another two months to determinant of the cure rate in national tuberculosis programs. achieve 90 percent sputum conversion. There is a vast and detailed literature on compliance in general In all probability, the patient's perception of the effective- and on tuberculosis in particular (WHO Tuberculosis Chemo- ness of treatment and the balance between discounted future therapy Centre 1963; Haynes, Taylor, and Sackett 1979; Fox costs and benefits of treatment are also important determi- 1983a, 1983b; Chaulet 1987; Reichman 1987). Many of the nants of compliance. In Tanzania and other IUATLD-assisted factors that one might expect would influence patients' com- national tuberculosis programs, it has been observed that both pliance with antituberculosis drug regimens, such as the sever- the perceived effectiveness of treatment and individual and ity of side effects, have not been empirically observed. There group education of patients during the initial intensive phase is a clear consensus, however, that the duration of treatment of short-course chemotherapy positively affected compliance adversely affects compliance (Haynes 1979). Moodie (1967) during the continuation phase. in unusual circumstances in Hong Kong found that most Other possible determinants of compliance include the noncompliers dropped out in the first three weeks; but all other number of medications taken at each time, the number of doses studies have observed a steady dropping out over time (East per week, and the cost of therapy to the patients. Combination African and British Medical Research Council 1977, 1979). tablets of isoniazid and thiacetazone and isoniazid and rifam- Improved net compliance in shorter regimens is an important picin have been in use in national tuberculosis programs of advantage of short-course chemotherapy over standard che- many developing countries for several years. Conversely, motherapy. Given the relapse rate as a function of months of intermittent standard chemotherapy (streptomycin and isoni- treatment discussed above, in a situation where patients con- azid) has never been used on a large scale in developing tinue to drop out over time, short-course chemotherapy will countries. In India, it has been shown that intermittency leads have a higher total cure rate. to increased irregularity of compliance (Pamra and Mathur Anotherdeterminantoftuberculosischemotherapycompli- 1973). Also Blackwell (1979) could not validate the expected ance is the degree of supervision of treatment. A spectrum relationship between reduced number of doses and improved exists, from giving supplies of drugs for multiple months to compliance. The advantages and disadvantages of intermit- patients all the way to hospitalization for the entire duration tent standard chemotherapy will not be addressed further here. of treatment. Between these extremes, a wide variety of super- The common sense notion that increasing costs both in time vision strategies are possible, including daily patient visits to and money will decrease compliance has been confirmed in health centers, health visitors contacting patients in the home, most studies (Haynes 1979). To maximize compliance, tuber- periodic urine tests to monitor compliance, and hospitalization culosis chemotherapy should be free and the spatial and tem- for the first two months of treatment. Although increased poral ease of access to treatment should be improved. When supervision increases compliance in most settings (Haynes alternative treatments are available in the private and public 1979), increased supervision also means increased cost. The sector, patients may initiallyprefertopayfortherapy perceived balance of this tradeoffwill depend on the specific institutional as better, but when funds run out they may switch to the public and cultural characteristics of each community. For example, sector (Uplekar and Shepard 1991). This mixing of different in Madras, in areas where most of the population has ready drug regimens will tend to increase the failure rate and the access to health centers, entirely ambulatory care has been probability of secondary resistance. successful (Tuberculosis Chemotherapy Centre, Madras 1959; The second factor determining the effectiveness is the Dawson and others 1966). On the other hand, in many parts development of drug resistance. Under ideal conditions, of rural Sub-Saharan Africa, the only way to guarantee daily such as in many clinical trials in patients with sensitive supervision of chemotherapy may be to hospitalize patients for bacilli, the cure rates for both standard and short-course the first two months of chemotherapy; this has been the chemotherapy are over 95 percent. In patients infected with experience in seven African countries (Tanzania, Kenya, Mo- tubercle bacilli that are isoniazid resistant, the cure rate zambique, Malawi, Benin, Senegal, and Mali) (Styblo and with total compliance is greatly reduced (Shimao 1987). Chum 1987). Isoniazid resistance is already a significant problem in many The rationale for hospitalizing patients to ensure close su- developing countries (Kleeberg and Boshoff 1980). A sys- pervision of the initial intensive phase is much greater in tematic application of short-course chemotherapy referred short-course chemotherapy than in WHO standard chemother- to above (2SHRz/6TH) in new smear-positive cases makes it apy. Two months of short-course chemotherapy will convert virtually impossible to select for a bacillus resistant to all smear-positive sputum into smear-negative sputum in about 90 four drugs. Decreased development of resistance means that percent of patients, and in another two to four weeks, in the short-course chemotherapy is a substantially more effective remaining 10 percent. Even if they stop taking drugs one or long-term strategy for tuberculosis control than standard two months after they leave the hospital, many will not relapse. chemotherapy. It has to be stressed that acquired (and in In Tanzania, approximately 50 percent of smear-positive pa- contacts of the index cases, primary) resistance to both tients enrolled in WHO standard chemotherapy remain smear isoniazid and rifampicin results in incurability of the major- and culture positive at two months. For standard chemother- ity of such cases in developing countries, with serious con- apy, it is crucial to continue to take isoniazid and thiacetazone sequences for elimination of tuberculosis. 250 Christopher Murray, Karel Styblo, and Annik Rouillon Finally, tuberculosis is maintained in the community by the Figure 11 -11. Tuberculosis Program Costs transmission of the bacillus from smear-positive patients to susceptible hosts. Short-course chemotherapy converts most Cost patients to smear negativity faster than standard chemother- apy and more effectively because of higher cure rates and higher compliance. Therefore, fewer patients transmit the bacillus to new hosts. Short-course chemotherapy will thus _ lead to a more rapid reduction in the risk of infection and incidence of clinical tuberculosis. This transmission effect for Cost Variable costs the treatment of smear-positive tuberculosis has a significant effect on the choice of chemotherapy. Fixed tuberculosis program costs Costs of Chemotherapy The cost of any tuberculosis control program is made up of the Other fixed costs costs of many components, including drugs, staff, transporta- tion, training, and hospitalization. Although drugs form a considerable portion of the budget, probably from 20 to 40 percent, they are not the only cost. Cost differences between Soure: Authors. short-course and standard chemotherapy, however, are attrib- utable to the costs of drugs and hospitalization. The choice of and so on. Those with these forms should be enrolled in suppliers, such as UNIPAC, Chinese pharmaceutical firms, or short-course chemotherapy (patients with tuberculous menin- European companies, will have a substantial bearing on the gitis should also be given rimactazid in the continuation phase costs of standard and short-course chemotherapy. Likewise, of treatment). For treatment of cases in which the sputum or the size of the drug purchases have a significant influence on culture fails to convert in the first round of treatment, the drug cost. Without a single agency or group that can evaluate the costs are particularly high, because these patients harbor tu- routine quality of antituberculosis drugs produced by different bercle bacilli, frequently resistant, in developing countries, to manufacturers, it is difficult to choose the most cost-effective isoniazid and/or streptomycin. Many of them have to be treated supplier or suppliers. In general, the short-course regimen used with short-course chemotherapy for retreatment cases, which in IUATLD national tuberculosis programs is approximately $20 should ideally contain three drugs to which the bacilli are to $25 more per patient than standard chemotherapy, depend- sensitive. A retreatment regimen includes, as a rule, rifampicin ing on the supplier. and pyrazinamide. In the IUATLD-assisted national tuberculosis Another potential source of cost differences between treat- programs the following regimen is used: 2sHRzE/IHRzE/5H3R3E3 ment regimens is the level and intensity of supervision. Both for patients resistant to isoniazid or 2SHRZE/1 HRZE/5TH for standard and short-course chemotherapy should be given patients sensitive to isoniazid. In programs that are committed whenever possible on an entirely ambulatory basis. In some to treating all patients that present for care, retreatment rural areas, however, where the population is without ready must also be considered in examining short-course and access to health facilities, daily regimens may have to be standard chemotherapy. Because failure rates are higher for delivered in district hospitals to maintain acceptable compli- standard chemotherapy, more resources would have to be ance and cure rates. The experience of the national tubercu- devoted to retreatment of these patients. losis programs in Tanzania, Malawi, and Mozambique has indicated that hospitalization during the intensive phase of Cost-Effectiveness chemotherapy is indeed necessary in many areas. Not only will this improve compliance, but expensive and valuable drugs The cost-effectiveness of treating smear-positive tuberculosis can be better accounted for in these conditions. Because two will be addressed first. In general, the cost per death averted months of short-course chemotherapy can permanently cure directly and indirectly will be lowest for smear-positive tuber- more than 60 percent of patients as compared with standard culosis, higher for other tuberculosis, and highest for retreat- chemotherapy, which cures only 30 percent, the higher cost of ment cases. Although this statement may run counter to hospitalization may be more justified for short-course regimens intuitive notions of the clinical costs of treating each type of in some circumstances. tuberculosis, the rationale is based on the effect of interrupting This discussion has thus far been implicitly restricted to the transmission, as explained more fully below. treatment of smear-positive tuberculosis. Once other forms of Few studies have examined the cost-effectiveness of tuber- tuberculosis have been identified, treatment costs for other culosis treatment in developing countries (Feldstein, Piot, and tuberculosis should be similar to standard chemotherapy ex- Sundaresan 1973; Bamum 1986; Joesoef, Remington, and cept for serious forms of smear-negative tuberculosis, such as Tjiptoherijanto 1989; Murray, Styblo, and Rouillon 1990). miliary tuberculosis, tuberculous meningitis, Pott's disease, The authors of two of these investigations reported that per Tuberculosis 251 case cured short-course chemotherapy was more cost-effective. Table 11 -11. Estimated Average Incremental Cost They did not, however, report figures on the cost per death per Patient Treated in Low- and Middle-Income averted. To fill the gap in information on the cost-effectiveness Countries of short-course and WHO standard chemotherapy, the national (dollars) tuberculosis programs in Malawi, Mozambique, and Tanzania GDP Short-course Short-course Standard Standard have been studied by Murray and others (1991) and DeJonghe per capita hospitala ambulatoryb hospitat ambulatoryd and others (1992). Before detailing the costs per case treated, some unit cost 250 181 70 159 48 definitions are needed. Program costs in these three countries 500 296 87 274 64 can be divided into three components. The first is variable 750 411 104 389 82 costs, which are a direct function of the number of patients 1,000 526 122 504 100 treated and include costs such as drugs, reagents for diagnosis, 1,250 641 139 619 117 and food during hospitalization. The second component is the 1,500 756 156 734 134 fixed costs associated with the tuberculosis program itself, such a. Shorc-course chemotherapy with sixty days of hospitalization during the as the salaries of district and regional tuberculosis coordinators, intensive phase. capital costs of vehicles, and administrative costs of the tuber- h. Short-course chemotherapy with daily supervision during the intensive culosi unit The hird omponnt is he fied cots inurred phase. culosis unit. The third component is the fixed costs incurred c. Standard chemotherapy with sixty days of hospitalization during the through use of the primary health care infrastructure, such as intensive phase. clinics and district hospitals. These three types of costs are d. Standard chemotherapy with daily supervision during the intensive illustrated in figure 1 1 - 1 1. Three unit costs can also be defined. pSource. Atithrs. Marginal costs are here defined as the average variable costs per case; average incremental costs are variable costs plus the fixed tuberculosis program costs per case; and average costs are In table 11-10 we provide the estimated average, average total costs, including the fixed costs outside the tuberculosis incremental, and marginal costs per case treated under short- program per case. course, standard, and retreatment regimens for smear-positive cases with and without hospitalization. These costs cannot easily be generalized to developing countries that have sub- Table 1-IO. Estimated Costs per Case Treated stantially higher incomes per capita than those in Malawi, in Malawi, Mozambique, and Tanzania Mozambique, and Tanzania, whose gross domestic product (1989 U.S. dollars) (GDP) per capita is under $300. Some treatment costs require Treatment and type of cost Malawi Mozambique Tanzania foreign currency or are intemationally traded goods; other costs are local costs that can be paid in local currency and are Short-course chemotherapy not traded commodities. By separating the external costs from with hospitalizationa the domestic costs, we can generate more representative esti- Average cost 160 217 174 mates of the cost of treating patients in countries with different Average incremental cost 99 155 127 incomes per capita. The extemal component of the cost is Marginal cost 69 140 101 assumed to be the same in all countries, whereas the domestic Standard chemotherapy component is assumed to be proportional to GDP per capita. In with hospitalization table 11-11 we give our best estimates of the cost of chemo- Average cost 91 73 72 therapy in countries with different levels of income.7 Notably, Average incremental cost 71 54 63 Marginal cost 42 40 37 the cost of chemotherapy with hospitalization increases much mote rapidly than ambulatory strategies as GDP per capita Ambulatory short-course increases. In other words, chemotherapy with hospitalization Average cost 66 55 50 is relatively more affordable in low-income countries. Average incremental cost 45b 36 41 The benefits of chemotherapy can be divided into the direct Marginal cost 19 18 15 benefits for the patient of cure and a reduced death rate and Re treatment chemotherapy the indirect benefit of reduced transmission. A life table for with hospitalization the prognosis of smear-positive pulmonary tuberculosis based Average cost 209 323 252 on the most detailed study of the prognosis of pulmonary Average incremental cost 141 232 182 tuberculosis by Berg (1939) is provided in table 11-12. The Marginal cost 97 206 146 Bangalore epidemiological study confirms the general case-fa- a. Hospitalization for sixty days during the intensive pliase of chemo- tality rates inadevelopingcountry (Olakowski 1973; National therapy. Tuberculosis Institute, Bangalore 1974). Foreach program, the h. Hypothetical estimate based on measured COStS; ambulatory therapy is cohort results of chemotherapy have been used to construct an not actually provided. alternative life table of the fate of cases treated by the program. c. Hospitalization for ninety days. Source: Authors. Comparison of the treatment life table and natural progression 252 Christopher Murray, Karel Styblo, and Annik RouiUon Table 11 -12. Life Table for Untreated, Smear- economic terms, the depth of the margin is much greater with Positive Pulmonary Tuberculosis short-course chemotherapy than with standard chemotherapy. Population There are also several other unquantified benefits to short- Year after Population alive at Deaths excreting bacillus at course chemotherapy. Rates of secondary resistance with short- diagnosis beginning of year during year beginning of year course chemotherapy are much lower. And the cost of 0 100,000 28,596 100,000 retreating failures has not been built into the comparison. 1 71,404 11,564 51,334 With standard chemotherapy many more patients will require 2 59,840 9,771 30,928 the expensive retreatment regimens. The benefits of short- 3 50,070 5,705 18,605 course and standard chemotherapy are summarized in table 4 44,364 5,055 11,851 11-14. 5 39,309 3,545 7,549 We do not arrive at equally robust conclusions concerning 6 35,764 3,225 4,938 the appropriate role of hospitalization with short-course che- 7 32.538 2,074 2,230 motherapy. Clearly, ambulatory chemotherapy is much 9 28,522 1,818 1,463 cheaper per patient treated. Experience in Tanzania and Mo- 10 26,704 - 985 zambique has shown that in urban areas with good health - No ialservice facilities high cure rates can be achieved with ambula- - Not available. Note; Based on Berg's study of 6,162 cases. By convention, the radix of the tory short-course chemotherapy. In the rural areas, these same starting population is set at I 00,000. programs have not been successful in employing ambulatory Source: Berg 1939; authors. life table allows us to quantify the marginal benefits of treat- Table 11-13. Estimated Average Incremental Unit ment. A model based on the principle that in an untreated Costs per Case Cured and per Death Averted in population one case of tuberculosis will lead to one case of Malawi, Mozambique, and Tanzania smear-positive tuberculosis in the future has been constructed. (1989 U.S. dollars) Transmission benefits have been counted for four transmission Treatment and type of cost Malawi Mozambique Tanzania cycles, which occur over the next eighteen and one-half years. Short-course chemotherapy Deaths averted and years of life saved have been discounted at with hospitalization 3 percent. In the model, it is assumed that passive case detec- Per case cured 165 232 202 tion will lead to diagnosis after three months of symptoms and Per direct death averted 200 267 236 that during the first three months before diagnosis the rate of Per total deaths averted 38 57 47 transmission is 50 percent higher than normal. This captures Per year of life saved 1.7 2.6 2.1 the increased rate of transmission to close household contacts Standardchemotherapy during the period before diagnosis. The false positive diagnosis with hospitalization rate has been studied in Tanzania and is less than 5 percent. Per case cured 215 301 270 The study results are based on an assumed false positive rate of Per direct death averted 187 272 227 5 percent for all three programs. Per total deaths averted 54 76 68 The cost-effectiveness ratios for short-course and standard Per year of life saved 2.4 3.4 3.1 chemotherapy with and without hospitalization during the Ambulatory short-course intensive phase are summarized in table 11-13. Four ratios are chemotherapy provided for each intervention: the cost percasecuredPer case cured 107 81 101 provided for eachnintervention: the cost per case cured; the Per direct death averted 130 94 117 cost per direct death averted; the cost per total death averted, Per total deaths averted 25 20 23 which includes deaths averted due to decreased transmission Per year of life saved 1.1 0.9 1.1 over the next eighteen and one-half years; and the cost per year Ambulator- standard of life saved, including transmission benefits. Three conclu- chemotherapy sions follow from the cost-effectiveness ratios. First, chemo- Per case cured 111 82 107 therapy for smear-positive tuberculosis is extremely cost- Per direct death averted 96 74 90 effective. The average incremental cost per year of life saved Per total deaths averted 28 21 27 ranges from $1 to $4. Second, short-course chemotherapy is Per year of life saved 1.3 0.9 1.2 preferable to standard chemotherapy in virtually all situations. Note: For Malawi, the estimates for standard chemotherapy with hospital- The ratios in table I I -13 show that short-course chemotherapy Ization and ambulatory short-course and standard chemotherapy are not based is cheaper than standard chemotherapy for virtually all indi- on actual program results. The costs are based on estimates of the likely cost of ambulatory chemotherapy, and the results of treatment are the average cators of cost-effectiveness. The absolute difference in the cost results achieved in Tanzania and Mozambique. per unit benefit is not large, but the cost-effectiveness ratios The results for ambulatorv treatment are based on the overail results of the do not tell the whole story. The cure rate with short-course progam for each countrv, not on specific results of ambulatory chemotherapv. TheyXae applicable onl y to those urban areas where high compliance can be chemotherapy in all three countries is approximately 25 per- maintained with daily supervised chemotherapy in the intensive phase. centage points higher than with standard chemotherapy. In Source. Authors. Tuberculosis 253 Table 11-14. Costs and Benefits of Short-Course culosis cases diagnosed only by x-rays that would, if left un- Chemotherapy and Standard Chemotherapy Based treated, progress to smear-positive tuberculosis. on National Tuberculosis Programs of Malawi, The Bangalore epidemiological study provides one of the Mozambique and Tanzania few sources for estimating these parameters (Olakowski 1973). Standard Short-course Of 304 persons considered to have active or probably active Parameter chemotherapy chemotherapy tuberculosis according to radiological findings but who were smear and culture negative, a total of 13 percent became Average incremental cost per year... of life saved with hospitalization bacteriologically positive during five years of observation. If (U.S. dollars) 3.00 2.00 half of these were smear positive, that would be only 6.5 percent of those whose x-ray was suggestive of tuberculosis who went on to become infectious smear-positive patients. This Percent of cases requiring percentage is the product of the specificity of the original x-ray retreatment (percent) 30 10 diagnosis and the probability of true smear-negative persons Source: Authors. progressing on to become smear positive. If diagnosis was only 50 percent specific, then 13 percent of the true smear negatives would have progressed on to become smear positive. This chemotherapy. Because the determinants of compliance are should be taken as the minimum estimate because smear- complex and often locally specific, we cannot make general negative, culture-positive patients were excluded from the conclusions. We can, however, estimate the marginal cost per analysis. Short-course chemotherapy trials in Hong Kong have patient cured through hospitalization for any given percentage shown in a much more medically sophisticated setting that 56 point increase in the cure rate purchased through hospitaliza- percent of patients whose x-ray was suggestive of tuberculosis tion during the intensive phase. In figure 11-12 we show, for went on to develop bacteriologically positive or clinically Malawi, Mozambique, and Tanzania, the relation between the active disease during a period of sixty months (Hong Kong marginal cost per case cured and the absolute percentage point Chest Service/Tuberculosis Research Centre, Madras/British increase in the cure rate achieved through hospitalization. We Medical Research Council 1984). show that once the cure rate is increased by as much as 10 to In the Bangalore study, although the percentage becoming 15 percentage points, hospitalization becomes relatively inex- bacteriologically positive was low, the death rate of those pensive per marginal patient cured. In middle-income coun- whose x-ray was suggestive of tuberculosis was 30.9 percent tries, where the cost of hospitalization increases much more over five years as compared with approximately 50 percent in than ambulatory chemotherapy, the increase in the cure rate bacteriologically confirmed cases. The mortality rate in the would have to be substantially higher to achieve the same former cases was well over twice the baseline death rate in the marginal cost per patient cured. study population. In countries with poorly trained microscopists or frequent atypical mycobacteria infections, the predictive value positive Figure 11 -12. Marginal Cost per Case Cured with of sputum positivity could be lower than 95 percent. The Hospitalization and Short-Course Chemotherapy potential of wasting scarce resources on patients without tu- in Three Countries berculosis puts a high premium on training health workers and microscopists to diagnose tuberculosis correctly. Costs (U.S. dollars) Chenotherapy for Snear-Negative Pulnonany Tuberculosis 4.5 The cost-effectiveness of chemotherapy for smear-negative 4.0 pulmonary tuberculosis is much more difficult to assess. The _\\ criteria both for diagnosis and effective therapy are less objec- 3 tive. A series of studies (see Toman 1979 for review and 3.0 Mozambique discussion) have shown that there is substantial variation in the x-ray diagnosis of active tuberculosis both between observ- 2.5a ers and by the same observer seeing the same film at different times. Cost-effectiveness can be discussed only in hypothetical 2.0 - terms and using realistic values from a variety of studies for the key parameters. There are four main determinants of the 1.5 l l l l l l cost-effectiveness of chemotherapy for smear-negative pulmo- 5 10 15 20 25 30 35 40 nary tuberculosis: the predictive value positive of x-ray diag- Absolute increase in cure rate (percent) nosis; the case-fatality rate of untreated smear-negative cases or cases suggested by x-ray; the effective cure rate of chemo- Note: Costs are drawn on a log scale. therapy; and, perhaps most important, the percentage of tuber- Source: DeJonghe 1993. 254 Christopher Murray, Karel Styblo, and Annik Rouillon Using the average incremental unit cost for ambulatory Figure 11 13. The Ratchet Effect therapy in Malawi, Mozambique, and Tanzania and a cheap short-course regimen for smear negatives ($16 per course), we Annual risk of infection (percent) will calculate the hypothetical cost-effectiveness. We will 1.6 assume a predictive value positive of 50 percent for active disease detected on x-ray. The true value will be locally specific 1.4 No program and could well range between 25 and 75 percent. For true 1.2 smear-negative cases, we will assume a case-fatality rate of 40 1.0 Temporary program percent. Because the regimen proposed is ambulatory and the 0.8 symptoms in smear-negative patients are often less severe, we will assume that the effective cure rate would be on the order 0.6 of 50 percent. With this set of assumptions, the cost per death 0.4 - Continued program averted is $450, or nearly ten times the cost per death averted 0.2 - of short-course chemotherapy with hospitalization for smear- 0_l_l_l_l_l_ l_l_l_l_l_l_l_ l positive patients and twenty times the cost of ambulatory 0 2 4 6 8 10 12 14 16 18 20 22 24 short-course chemotherapy. Year If a percentage of smear-negative cases do not progress to become smear-positive cases, then the costs of treating smear- negative patients are nearly an order of magnitude greater than Source: DeJonghe 1993. the costs of treating smear-positive patients. On the basis of the Bangalore data, however, we expect that at least 10 to 1 5 previous levels. The number of people infected by a smear- percent would progress to become smear-positive patients. positive patient is a function of the social patterns of inter- Treating smear-negative patients that go on to become smear action and household structure, not of the overall risk of positive cuts out the prediagnosis transmission that cannot be infection. The last line shows the rapid decline in the risk of affected with chemotherapy for those who are smear positive. infection if investments in chemotherapy are maintained. This prediagnosis transmission bonus accounts for nearly one- Each investment, even if only temporary, has a permanent fifth of total transmission. If 15 percent of the cases progress to effect of ratcheting down the annual risk of infection and thus become smear positive, the cost per death averted by treating incidence. smear-negative patients is reduced to $185 and $155 if 20 With the HIV epidemic, it is possible that the baseline percent become infectious. This is still three and one-half to decline in the annual risk of infection in Sub-Saharan Africa eight times more expensive than treating smear-positive pa- may not persist and may even be reversed. In this scenario, tients. In comparisonwith many other health sector interven- although the annual risk of infection will not be ratcheted tions, this is relatively inexpensive per death averted or year down by investments in chemotherapy, there will be a persist- of life saved. If the predictive value positive of x-ray diagnosis ing benefit. The two lines labeled "No program" and "Tempo- can be increased to 70 percent, the cure rate increased to 65 rary program" in figure 11 -13 would increase slowly at the same percent, and 20 percent of cases go on to be smear negative, rate, maintaining a permanent difference in the ultimate risk the cost per death averted could be as low as $85. of infection, even many years into the future. The Ratchet Effect BCG and Case Treatment Investments in chemotherapy for smear-positive tuberculosis One would like to compare the two main interventions for are relatively secure as compared with investments in other tuberculosis control: BCG and case treatment. They are, how- infectious diseases. An illustration of how chemotherapy pro- ever, not truly comparable because even complete BCG cover- grams can ratchet down the incidence of tuberculosis is pre- age at birth will affect only 4 to 7 percent of mortality. Case sented in figure 11-13. The top line shows the slow decline in treatment is absolutely necessary to reduce the other 90 per- the annual risk of infection in the absence of an effective cent and more of mortality. How does the cost-effectiveness program. After twenty-five years the annual risk of infection of expanding BCG coverage compare with expanding case- will only be reduced by 20percent. Agood tuberculosis control treatment activities? The cost per death averted can be com- program should be able to reduce the annual risk of infection pared directly using the studies mentioned earlier. Some may by at least 6 percent per year-for example, since the 1950s, object that a death between the ages of 0 and 14 represents a the annual risk of infection in the West has been declining by greater loss of years of life than a death at age 35. If we choose about 10 percent per year. If after seven years, as shown in to examine discounted years of life lost, however, it will not figure 11-13, the program collapses, then the annual risk of significantly alter the comparison. A death at age 7, the infection will revert to its baseline rate of 1 percent decline. midpoint for deaths averted by BCG, represents, at a 3 percent There is no reason to expect, as with malaria, schistosomiasis, discount rate, 29.7 years of life lost, whereas a death at age 34, or hookworm, that the risk of infection will increase back to the average age of a tuberculosis death, represents 23.4 years Tuberculosis 255 at a similar discount rate. Therefore, we can examine the tool, especially if it also prevented tuberculosis in already cost-effectiveness of the two interventions using the cost per infected individuals. Fine (1989), however, has pointed out death averted, bearing in mind that discounted years of life lost that, for moral and technical reasons, it will be difficult to test would change the relationship by less than 20 percent. appropriately the effectiveness of any new vaccines. Research The cost per death averted through tuberculosis chemo- is also needed to explore the most appropriate role for chemo- therapy should change little as the risk of infection in a prophylaxis in developing countries. community declines. If all else remains the same, the only change would be the slight increase in the cost of detection DIAGNOSIS. Development of new tools for the rapid diagnosis as more cases of cough would have to be screened per case of tuberculosis would substantially improve case detection. of tuberculosis detected. This does not hold true for any Research into serological or sputum diagnosis that can be immunization, including BCG. The cost of immunizing all deployed inperipheralhealthfacilities indevelopingcountries infants will not change as the risk of infection declines, but should be a priority. the benefits in terms of deaths averted will decline propor- tionately to the risk of infection. In other words, the cost CHEMOTHERAPY. Development of new, shorter acting, cheap per death averted through BCG must be inversely propor- drugs would help address two important issues in tuberculosis tional to the risk of infection. In figure 11-7 we show two control: compliance and cost. Although opportunities exist for hypothetical curves for the cost per death averted as a developing new drugs (Sensi 1989), relatively little research is function of the risk of infection. The curve for the cost- under way. Another possibility that seems worth exploring is effectiveness of BCG is estimated on the basis of a single data the use of depot preparations and four drug combination pills point for Indonesia and the average cost per death averted which could solve compliance problems. for short-course chemotherapy with hospitalization during the intensive phase is based on data from Malawi, Mozam- PROGRAM DESIGN.There is an urgent needforoperational and bique, and Tanzania. Although the data are clearly weak, health economics research on strategies for tuberculosis con- the principle is clear. At low annual risks of infection, case trol. Some key issues have been highlighted in this chapter: treatment is a substantially more cost-effective strategy than what is the tradeoff between the cost of supervision and the expanding BCG coverage. At higher risks of infection, the improvement in compliance, taking the existing infrastructure costs of both interventions are of the same order of magni- into consideration? What is the cost-effectiveness of altema- tude. This curve should not be interpreted to mean that tive diagnosis strategies? These and many other issues need to countries with low risks of infection should curtail BCG be addressed in an organized fashion. immunization activities. The discussion so far provides no insight into the savings from cutting back an existing activ- HIV AND TUBERCULOSIS INTERACTIONS. The interaction be- ity as opposed to the potential reduction in benefits. This tween HIV and tuberculosis has not been fully addressed in this discussion does not imply that the policy choice in tubercu- chapter. It appears that immune-suppressed patients with HIV losis control is between BecG and case treatment. Some have a high probability of developing clinical tuberculosis. In combination of the two is likely to be desirable in many Central and East Africa, tuberculosis programs are already countries. It does, however, indicate that BCG becomes rel- reporting an increase in the number of cases of tuberculosis. atively less attractive as the risk of infection declines. The effect of any HIv-tuberculosis interaction on the annual risk of infection for the rest of the population is not known. Research Priorities Epidemiological study of these relationships has just begun and should be considered a priority for research. This discussion of tuberculosis leads naturally to some general recommendations for tuberculosis research. These can be di- Major Operational Conclusion vided into six areas. This review of tuberculosis can be summarized in six main EPIDEMIOLOGY. The wide confidence intervals In the esti- mates of incidence, prevalence, and mortality highlight the points. need for epidemiological research. Many countries require * The magnitude of the tuberculosis problem is simply basic information on incidence and mortality rates and their staggering. Our estimates suggest that 2.7 million people die distribution by age and socioeconomic status in order to estab- from tuberculosis each year. This is probably more than from lish the importance of tuberculosis as a health sector priority. any other single pathogen. The burden of tuberculosis ex- For those countries that do not register vital statistics, new tends beyond morbidity; the annual incidence of new cases survey techniques based on the verbal autopsy may provide the of all forms of tuberculosis is over 7.3 million in the devel- tools with which tuberculosis mortality can be quantified. oping world. Tuberculosis is unique among the main killers of the developing world in that it afflicts nearly all age PREVENTION. Because of the uncertain and variable effective- groups. Many children die from tuberculous meningitis and ness of BCG, a new effective vaccine would be an important miliary tuberculosis. But the greatest burden of tuberculosis 256 Christopher Murray, Karel Srt-blo, and Annik Rouillon incidence and mortality is concentrated in adults age fifteen The combination of the enormous burden of the disease, to fifty-nine. These are the parents, workers, and leaders of years of neglect, the existence of effective interventions, and society. This heavy toll of the care givers makes tuberculosis the availability of one of the most cost-effective -interventions a unique problem. must make tuberculosis one of the highest priorities for action * In at least the last decade and a half, tuberculosis has been and research in international health. ignored by much of the intemational health community. Shimao (1989) has outlined the decline of the human and institutional capacity to address the tuberculosis problem Notes over the last decades, which is but one symptom of a general 1. The Ilinear relationship between the annual risk of infection and the lack of priority attached to tuberculosis action and research. incidence of smear-positive tuberculosis will not hold at low annual risks of Another example is the study done by the Institute of infection. As the annual risk of infection declines, the percentage of cases Medicine (1986) of vaccine development priorities for the resulting from endogenous reactivation that are related to past levels of the developing world. The institute classified diseases into three annual risk of infection (during the last fifty to ninety years) rather than to levels of priority for research on vaccines. Whereas leprosy current levels will increase. received significant attention, tuberculosis was not even 2. There are two other problems with the interpretation of reported newly mentioned in the lowest priority group. Clearly, focusing registered cases. For most countries, no distinction is provided between smear- mentiond in te lowet prioity grop. Clerly, fcusing positive cases and other cases. For countries in Latin America, the Middle East, intemational attention on tuberculosis is the necessary first and China, where a substantial portion of diagnosis is through x-ray, the step if more resources are to be directed to combatting the numbers detected can be misleading. For example, in China, the widespread disease. use of x-ray for diagnosis and the poor quality of microscopy means that only 10 to 20 percent of cases detected are smear positive. Many of the undetected Existing diagnostic technology and chemotherapeutic smear-positive cases are probably diagnosed as smear negative with tuberculo- agents can be used effectively in developing countries to cure sis suggested by the x-ray. Many of the putative smear-negative cases, however, tuberculosis. The lUATLt)-assisted national tuberculosis pro- are probably misdiagnosed or overdiagnosed. Thus, if the total number of cases grams (for example, those in Malawi, Mozambique, and detected is divided by the estimated incidence, we will substantially overesti- Tanzani) haveshown hat shrt-couse cheottierpy can mate case-detection rates. On the basis of discussions with national programs, we have ad)usted the cases reported by China to reflect the likely overdiagnosts be applied on a national scale with excellent results. Cure of smear-negative cases; data presented in Fox (1990) have been used to adjust rates approaching 90 percent, even taking into consider- the Indian data. ation problems with compliance, can be achieved in the 3. The midpoint of the confidence interval estimates is not equal to the most difficult circumstances. expectation of the interval. When two 95 percent confidence intervals are multiplied, the resulting interval is actually much larger. In addition, the * Short-course chemotherapy and BCG immunization (in expectation is slightly lower than the interval midpoint. countries with high risks of infection) are some of the most 4. If the age-specific case-fatality rate for ages twenty-five through twenty- cost-effective health interventions available in the health four is assumed to be indexed at 1, then the other rates are: zero through armamentarium. The analysis of the programs in Malawi, fourteen, 0.9; fifteen through twenty-four, 1.15; twenty-five through thirty- Mozambique, and Tanzania has shown that treating smear- four, l.0;thirty-fivethroughforty-four, 1.07;forty-fivethroughfifty-four, 1.15: positive tuberculosis costs $20 to $57 per death averted. The fifty,five through sixty-four, 1.65; and sixty-five and over, 2.5. The ratios for ages fifteen through sixty-five and over are based on Berg (1939); the ratio for cost per discounted year of life saved is therefore $1 to $3. ages zero through fourteen is based on case registration and tuberculosis There are few interventions that are as cost-effective as mortality data for London during 1933-34 in Styblo (1984). tuberculosis case treatment. s. All dollar amounts are 1986 U.S. dollars. 6. The studies cited refer to the risk of developing clinical tuberculosis soon * On the basis of country-by-country estimates, taking after primary infection. What about the risk of persons infected with tubercle into consideration the estimated incidence and current bacilli developing clinical tuberculosis, with or without a fresh reinfection? levels of case detection and treatment, we estimate that Because it is not possible to detect reinfection with tubercle bacilli by tuber- $150 million in extra resources is needed to treat 65 percent culin testing, it cannot be discovered directly whether or not exogenous of smear-positive patients in low-income countries and 85 reinfection is important in the development of tuberculosis in an adult. It is percent of smear-positive cases in middle-income devel- evident that in countries with low annual risks of infection, tuberculosis in elderly and old persons is predominantly a result of endogenous exacerbation oping countries with short-course chemotherapy. Of this among those remotely infected with tubercle bacilli. In developingcountries, $150 million, approximately $70 million in foreign cur- exogenous reinfection seems to play an important role in developing active rency is required to address the problem of smear-positive tuberculosis in the adult population, because 0.5 to 2.5 percent or more of tuberculosis. previously infected individuals are annually reinfected with tubercle bacilli, as was the case in industrial countries some two to four decades ago (Canetti * Evidence has accumulated that the interaction between 1972; Jancik and Styblo 1976). Strong evidence for the latter is the rapid HIV and tuberculosis may significantly exacerbate the epide- decline in tuberculosis incidence in Eskimos over the space of twenty years, miological situation of tuberculosis. The potential rise, as a not only in children and young adults but also in elderly and old people, when result of this interaction, in the risk of infection in Africa aggressive case detection and adequate chemotherapy was introduced and other regions, depending on the spread of iv, makes (Grzybowski, Styblo, and Dorken 1976). 7. This method is a modification of that of Bamum and Greenberg in our operational conclusions about tuberculosis all the more chapter 21, this collection, who have calculated unit costs bv the percentage pressing. of cot per capita. External costs or the costs of internationally traded goods Tuberculosis 257 whether they are domesticallv produced or not will not vary in proportion to Clemens, J. D., J. J. H. Chung, and A. R. Feinstein. 1983. "The BCO Contro- car per capita. Local costs of nontraded goods, most notably labor, will in all versy. A Methodological and Statistical Reappraisal." JAMA 249(17): probability change in proportion to cGr per capita. The distinction between 2362-69. external and domestic costs not only leads to different estimates of unit costs Cochrane, A. L. 1948. "Tuberculosis among Prisoners of War in Germany." but can alter the relative cost of different interventions. As discussed in the Bnnsh Medicaljournal 2:656. text, the cost of an intervention with a higher domestic percentage of total Crofton, J. 1962. "The Contribution of Treatment to the Prevention of cost will rise at a fasrer rate than the coist of an intervention with a higher Tuberculosis." Buletin of the International Union against Tuberculosis 32:643- percentage of extemal costs. Thus in a low-income country one intervention 53 with a large domestic component may be cheaper than an altemative with a large extemal component. The relative cost-effectiveness rank could, how- Daniel,T. M. 1989. "Rapid DiagnosisofTuberculosis: LaboratoryTechniques ever, reverse as income per capita rises. Applicable in Developing Countries." Review's of Infectious Diseases 2(sup- plement 2):S47 1-S478. Dawson, J. J. Y., S. Devadatta, Wallace Fox, S. Radakrishna, C. V. Rama- krishnan, P. R. Somasundavam, H. Stott, S. P. Tripathy, S. Vehi. 1966. "A References 5 Year Study of Patients with Pulmonary Tuberculosis in a Concurrent Comparison of Home and Sanatorium Treatment for One Year with Isoni- Aronson, J., C. Aronson, and H. Taylor. 1958. "A 20-Year Appraisal of BCG, aznd plus PAS." Bulletn of the World Health Organization 34:533-51. Vaccination in the Control of Tuberculosis." Archives of Internal Medicine DeJonghe, E., C. ]. L. Murray, H. G. Chum, D. S. Nyangulu, A. Salamao, and 101:881-93. K. Styblo. 1991. "Cost-Effectiveness of Chemotherapy for Sputum Smear- Barnett, G. D., S. Grzybowski, and K. Styblo. 1971. "Present Risk of Develop- Positive Pulmonary Tuberculosis in Malawi, Mozambique, and Tanzania." ing Active Tuberculosis in Saskatchewan According to Previous Tubercu- WHO, Tuberculosis Unit, Geneva. Forthcoming 1993 in Internationaljournal lin and X-Ray Status." Bulletin of the Internanonal Union against Tuberculosis of Health Management 8. 45:51-74. Drolet, G. J. 1938. "Present Trend of Case Fatality Rates in Tuberculosis." Bamum, H. N. 1986. "Cost Savings from Alternative Treatments for Tuber- American Review of Tuberculosis 37:125-51. culosis." Social Science and Medicine 23:847-50. East African and British Medical Research Council. 1977. "Tuberculosis in Bamum, H. N., D. Tarantola, and 1. F. Setaidy. 1980. "Cost-Effectiveness of Tanzania: A Follow-Up of a National Sampling Survey of a Drug Resistance an Immunization Programme in Indonesia." Bulletn of the World Health and Other Factors." Tubercle 58:55-78. Organization 58:499-503. . 1979. "Tuberculosis in Kenya: Follow-Up of the Second National Berg, G. 1939. "The Prognosis of Open Pulmonary Tuberculosis. A Clinical- Sampling Survey and a Comparison with the Follow-Up Data from the First Statistical Analysis." ActaTuberculosea Scandinavica, supplementum IV:vui- (1964) National Sampling Survey." Tubercle 60:125-49. viii, 1-207. Feldstein, M. S., M. A. Piot, and T. K. Sundaresan. 1973. "Resource Allocation Blackwell, B. 1979.."The Drug Regimen and Treatment Compliance." In R. Model for Public Health Planning. A Case Studv of Tuberculosis Control." Blackwell, B. 1979. "The Drug Regimen and Treatment CompEliance." In R. B. Haynes, D. W. Taylor, and D. L. Sackett, eds.,Compliance in Health Care. Bulletin of the World Health Organization 48 (supplement): 1-1 10. Baltimore: Johns Hopkins University Press, 1979. Fine, P. E. M. 1989. "The BCG Stora: Lessons from the Past and Implications Bleiker, M. A., H. J. Chum, S. J. Nkinda, and K. Styblo. 1987. "Tanzania for the Future." Reviews of Infectious Diseases 2(supplement 2):S353-S359. National Tuberculin Survey, 1983-1986." In Tuberculosis and Respiratory Fox, W. 1983a. "Compliance of Patients and Physicians: Experience and Diseases. Proceedings of the 26th IBAT (Internanonal Union against Tuberculo- Lessons from Tuberculosis." Part I. Bntish MedicalJournal 287:33-35. sis) World Conference on Tuberculosis and Respirators Diseases. Tokyo: Pro- . 1983b. "Compliance of Patients and Physicians: Experience and fessional Post-Graduate Services. Lessons from Tuberculosis." Part 2. British MedicalJournal 287:101-5. Bloom, B. 1989. "An Ordinary Mortal's Guide to the Molecular Biology of .1 990. "Tuberculosis inIndia: Past, Present,andFuture."Indianjournal Tuberculosis." Bulletin of the International Union against Tuberculosis and Lutig of Tuberculosis 37:175-213. Disease. 64:50-58. Frost, W. H. 1937. "How Much Control ofTuberculosis?" AmencanJournalof Brink, G. C., S. Grzybowski, and G. B. Lane. 1960. "Silicotuberculosis." Public Health 27:759-66. Canadian Medical Association Journal 94:999. Galtung Hansen, 0. 195 5. Tuberculosis Mortality and Morbidity and Tuberculin Canetti, G. 1939. Les riinfectons tuberculeuses latentes du poumon. Paris: Vigot Sensitivit in Norway. WHO EURO-84/1 5. Copenhagen: World Health Orga- Editions. nization. .1962."L'`radicationdelatuberculosedanslesdifferentspays,compte Great Britain Medical Research Council. 1972. "BCG and Vole Bacillus tenu des conditions existantes (problemes th6oriques et solutions pra- Vaccines in the Prevention ofTuberculosis in Adolescence and Early Life." tiques)." Bulletin of the International Union against Tuberculosis 2:608-42. Bulletn of the World Health Organization 46:3 71-85. 1972. "Endogenous Reactivation and Exogenous Reinfection. Their Greenwood, A., B. M. Greenwood, A. K. Bradley, K. Williams, F. Shenton, Relative Importance with Regard to Development of Nonprimary Tuber- S. Tulloch, P. Bvass, and F. S. J. Oldfield. 1987. "A Probpective Survey of culosis." Bulletin of the International Union against Tuberculosis 47: the Outcome of Pregnancy in a Rural Area of The Gambia, West Africa." 116-22. Bulletin ofthe World HealthOrganization 65:636-43. Cauthen, G. M .. A. Pic, and H. G. ten Dam. 1988. Annual Risk of Tuberculous Grzybowski, S., K. Styblo, and E. Dorken. 1976. "Tuberculosis in Eskimos." Infection. WHorTB/88.154 Geneva: World Health Organization. Tubercle 57(supplensent 41:1-58. Chaisson. R. E. and G. Slutkin. 1989. "Tuberculosis and Human Immuno- liavnes, R. B. 1979. "DeterminantsofCompliance: The Disease and Mechan- deficiency Vinis Infection." Journal of Infectious Diseases 159: 96-1 00. ics ofTreatment." In R. B. Haynes, D. W. Taylor, and D. L. Sackett, eds.. Chaulet, P. 1987. "Conipliance with Anti-tuberculosis Chemotherapy in Comphlance in Health Care. Baltimore: Johns Hopkins University Press, Developing Countries." Tubercle 68:19-24. 1979. Chum, H. J., K. Styblo, M. R. A. van Cleef, and M. Gunzareth. 1988. "Further Haynes, R. B., D. W. Taylor, and D. L. Sackett. 1979. Compliance in Health Development in Tanzania National Tuberculosis and Leprosy Programnme." Care. Baltimore: Johns Hopkins University Press. Paper presented at the IUATLD (Intemational Union against Tuberculosis Holm, J. 1970. "Our Enemy: The Tubercle Bacillus." InternationalTuberculosis and Lung Disease) Conference, Dubrovnik. Digest 5 (IL ATLD special publication.) 258 Chnstopher Murray, Karel St-vblo, ard Annik Rouillon Hong Kong Chest Service/Tuberculosis Research Centre, Madras/British 1989. "A Prospective Study of the Risk of Tuberculosis among Intravenous Medical Research Council. 1984. "A Controlled Trial of 2-month, 1mug Users with Human Immunodeficiency Viis Infection-" New England 3-month, and 12-month Regimens of Chemotherapy for Sputum Smear- Journal of Medicine 320:545-50. Negative Pulmonary Tuberculosis." American Review of Respiratory Diseases Sensi, P. 1989. "Approaches to the Development of New Antituberculosis 130:23-28. Drugs." Reviews of Infectious Diseases 2(supplement 2):S471 -S478. India, Ministry of Health and Family Welfare. 1986. Health Atlas of India. Shimao, T. 1987. "Drug Resistance in Tuberculosis Control." Tubercle 68(sup- Directorate General of Health Services, New Delhi. plement 2):5-15. Institute of Medicine. 1986. New Vaccine Development Establishing Prioriries. . 1989. "Institutional Capacity for Disease Research and Control: Vol. 2. Diseases of Importanice in Developing Countries. Washington D.C.: Tuberculosis." In M. Reich and E. Marui, eds., InternatonalCooperationfor National Academy Press. Health; Problems, Prospects, and Priorities. Dover, Mass.: Aubum House. Jancik, E. H., and K. Styblo. 1976. "Die Problematik der postprimaren Smith, P. G. 1987. "Case-Control Studies of the Efficacy of BCG against mykobakteriellen Superinfektion-Versuch einer epidemiologisch- Tuberculosis." In Tuberculosis and Respiratory Diseases. Proceedings of the klinischer Sicht." In Brecke: Fortbildung in Thoraxkrankheiten, vol. 7. Stutt- 26th IUAT (International Union against Tuberculosis) World Conference on gart: Hipokrates Verlag. Tuberculosis and Respiratory Diseases. Tokyo: Professional Postgraduate Joesoef, M. R., P. L. Remington, and P. Tjiptoherijanto. 1989. "Epidemiolog- Services. ical Mlodel and Cost-Effectiveness Analysis of Tuberculosis Treatment Snider, D. E., 0. J. Caras, and J. P. Koplan. 1986. "Preventive Therapy with Programmes in Indonesia." lnternationaljournal of Epidemiology 18:174-79. Isoniazid. Cost-Effectiveness of Different Durations of Therapy." JAMA Johnson, M. W. 1973. "Results of Twenty Years of Tuberculosis Control in 255(12):1579-83. Alaska." Health Services Reports 88:247-54. Stilwell, J. A. 1976. "Benefits and Costs of the Schools BCG Vaccination Kleeberg, H. H., and M. S. Boshoff. 1980. "A World Atlas of Initial Dnug Programme." Bintsh MedicalJournal l:1002-4. Resistance." Report to the Scientific Committee on Bacteriology and Stvblo, K. 1980. "Recent Advances in Epidemiological Research inTubercu- Immunology of the Intemational Union against Tuberculosis. 1osis." Advances in Tuberculosis Research 20:1 -63. Leowski, J. 1988. Global Status of Tuberculosis Control and ILs Prospects. WHO/ .1984. "EpidemiologyofTuberculosis."In G.Meissnerandothers,eds., TB/88.156. Geneva: World Health Organization. Infektionskrankheiten und ihre Erreger: Mykobaktena und mykobakteriellen Lindhart, M. 1939. The Statistics of Pulmonary Tuberculosis in Denmark, 1925- Krankheiten, vol. 4. Jena: Gustav Fischer Verlag. 1934. A Statistical Investigation on the Occurrence of Pulmonary Tuberculosis . 1985. "The Relationship between the Risk of Tuberculosis Infection in theperiod 1925-1934, WorkedOuton the Basis of the Danish National Health and the Risk of Developing Infectious Tuberculosis." Bulletin of the Interna- Service File of Notified Cases and of Deaths. Copenhagen: Ejnar Munksgaard. tional Union against Tuberculosis 60:117-19. Meijer, J., G. D. Bamett, A. Kubik, and K. Styblo. 1971. "Identification of .1986. "Tuberculosis Control and Surveillance." In D. C. Flenley and Sources of Infection." Bulletin of the International Union against Tuberculosis T. L. Petty, eds., Recent Advances in Respiratory Medicine, vol. 4. Edinburgh: 45:5-50. Churchill Livingstone. Moodie, A. S. 1967. "Mass Ambulatory Chemotherapy in the Treatment of - 1988. "The Relationship between the Annual Risk of Tuberculosis Tuberculosis in a Predominantly Urban Conimunity." American Review of Infection and the Incidence of Smear-Positive Pulmonary Tuberculosis." Respiratory Diseases 95:384-97. Unpublished. Murray, C. J. L., E. DeJonghe, H. G. Chum, D. S. Nyangulu. A. Salamao, and -. 1989. "Overiew and Epidemiologic Assessment of the Current K. Styblo. 1991. "Cost-Effectiveness of Chemotherapy for Pulmonary Tu- Global Tuberculosis Situation with an Emphasis on Control in Develop- berculosis in Three Sub-Saharan Countries." Lancet 338:1305-8. ing Countries." Reviews of Infectious Diseases, II (Supplement 2):S339- Murray, C.J. L., and R. G. Feachemn. 1990. "Adtilt Mortality in the Developing 46. World." Transactions of the Royal Society of Tropical Medicine. Styblo, K., and H. J. Chum. 1987. "Treatment Results of Smear-Positive National Tuberculosis Institute, Bangalore. 1974. "Tuberculosis in a Rural Tuberculosis in the Tanzania National Tuberculosis and Leprosv Pro- Population of India: A Five-Year Epidemiological StUdy." Bulletin of the gramme: Standard and Short-Course Chemotherapy." In Tuberculosis and World Health Organizanon 51:47 3-88. Respiratory Diseases. Proceedings of the 26th IUAT (International Union against Olakowski. T. 1973. Assignment Report on a Tuberculosis Longitudinal Survey, Tuberculosis) World Conference on Tuberculosis and Respirators Diseases. National Tuberculosis Insttute, Bangalore WHO Project: India 0103. Published Tokyo: Professional Postgraduate Services. in Geneva. sEA!TB/129. World Health Organization, Regional Office for Styblo, K., D. Dankova, J. Drapela, J. Galliova, 7. Jezek, J. Krivanek, Kubik, South East Asia. M. Langerova, andJ. Radkovsky. 1969. "Epidemiological and Clinical Study of Tuberculosis in the District of Kolin, Czechoslovakia." Bullenn of the Panira, S. P., and G. P. Miathur. 1973. Indian JounTial of Tuberculosis 20:108. World HealthOrganization 37:819-74. Preston, S. H., N. Keyfit7, and R. Schoen. 1972. Causes of Death: Life Tables fo Naioa Pouatos Ne Yok'einrPes Stv'blo, K., and)l. Mweijer. 1 976. "Impact of 8cC Vaccination Programmes in Children and Young Adults on the Tuberculosis Problem." Tubercle 57: Pryer, J. 1989. "When Breadwinners Fall Ill: Preliminary Findings from a Case 17-.43. Study in Bangladesh." IDS Bullenn 20(2):49-57. . 1980. "The Quantified Increase of the Tuberculous Infection Rate in Rechman, L. B. 1987. "Compliance in Developed Nations." Tubercle 68:25- a Low Prevalence Country to be Expected if the Existing MMR Programme 29. were Discontinued." Bulletin of the Internanonal Union against Tuberculosis Rouillon, A., S. Perdrizet, and R. Parrot. 1976. "Transmission of Tubercelle 55:3-8. Bacilli: The Effects of Chemotherapy." Tubercle 57:275-99. Styblo, K., J. Meijer, and 1. Sutherland. 1969. "The Transmission of Tubercle Rutledge, C. J. A., and J. B. Crouch. 1919. "The Ultimate Results in 1,694 Bacilli: Its Trend in a Human Population, Tuberculosis Surveillance Re- Cases of Tuberctilosis Treated at the Modem Woodmen of America Sana- search Unit (TSRU)" Bulletin of the International Union against Tuberculosis torium." American Review of Tuberculosis 2:755463. 42. Schechter, IM. 1953. "Health and Sickness in Times of Starvation and Food Styblo, K., and A. Rouillon. 1981. "Estimated Global Incidence of Smear Shortage." Harofe Haivri 2:191. Positive Pulmonary Tuberculosis. Unreliability of Officially Reported Fig- Selwyn, Peter A., Diana Harterl, Victor A. Lewis, Elhe E. Schoenbaum, Sten ures on Tuberculosis." Bulletin of the Internatonal Union against Tuberculosis H. Vermund, Robert S. Klein, Angela T. Walker, Gerald H. Friedland. 56:118-26. Tuberculosis 259 Sutherland, 1. 1968. The Ten-Year Incidence of Clinical Tuberculosis Followting . 1980. "Trial of BCG Vaccines in South India for Tuberculosis Preven- 'Conversion' in 2,550 Individuals Aged 14 to 19 Years. Tuberculosis Surveil- tion." Indian Journal of Medical Research 72(supplernent): 1-74. lance Research Unit (TSRU) Progress Report. The Hague. Tuberculosis Surveillance Research Unit. 1966. Progress Report. The Hague. .1976. "Recent Studies in the Epidemiology of Tuberculosis Based on UNICEF (United Nations Children's Fund). 1988. State of the WZorld's Children. the Risk of Being Infected with Tubercle Bacilli." Advances in Tuberculosis Oxford: Oxford University Press. Research l19:1-63. United Nations. 1986. World Population Prospects. Estimates and Projections as Sutherland, L., and P. M. Fayers. 1975. "The Association of the Risk of Assessed in 1984. New York. Tuberculosis Infection with Age." Bulletin of the International Union against WHO (World Health Organization). 1974. Expert Committee on Tuberculosis, Tuberculosis 50:70-81. Ninth Report. Technical Report Series 552. Geneva. Toman, K. 1979. Tuberculosis Case-Findingand Chemotherapy. Geneva: World . 1987. "Statement from Consultation on Human Immunodeficiency Health Organization. Virus (HIV) and Routine Childhood Immunization." Weekly Epidemiological Tuberculosis Chemotherapy Centre, Madras. 1959. "A Current Comparison Record 62:297-99. of Home and Sanatorium Treatment of Pulmonary Tuberculosis in South . 1988. "Reported Annual Incidence of Tuberculosis 1974-1987." India." Bulletin of the World Health Organization 21:44-51. Geneva. Mimeo. Tuberculosis Prevention Trial, Madras. 1979. "Trial of Vaccines in South WHO (World Health Organization)/Tuberculosis Chemotherapy Centre. 1963. India forTuberculosis Prevention: First Report." Bulletin of the WorldHealth "Drug Acceptability in Domiciliary Tuberculosis Control Programmes." Organization 57:819-27. Bulletin of the World Health Organization 29:627-39. 12 Leprosy Myo Thet Htoon, Jeanne Bertolli, and Lies D. Kosasih Leprosy has been referred to as one of the oldest diseases known whereas the paucibacillary form (i, TT, and BT) is generally less to humankind. The earliest written records describing true infectious and is bacteriologically negative. leprosy come from India and there it was known as kushta. It is Onset of leprosy is usually gradual, and the first signs may believed that from India it spread eastward to China and Japan not be apparent for quite some time after infection. The and then westward to the lands bordering the Mediterranean. insidious onset and uncertain time of exposure make it difficult During the fourteenth century, the leprosy epidemic reached to calculate the exact incubation period. The average incuba- its peak in Europe. During the height of the epidemic there tion period of leprosy is estimated to be two to four years, may have been as many as 2,000 hospices in France alone to though it may vary from nine months to twelve years (WHO care for the victims of leprosy (Browne 1985). The influence 1985b). of the Crusades on the spread of leprosy in Europe and the Before the days of modern chemotherapy for leprosy, the effect of the "Black Death" (which wiped out a third of the duration of illness was lifelong for some cases. Even with the population of that continent in 1349) on the decline of the discovery of dapsone, lepromatous cases were treated for life. leprosy epidemic has long been debated. For reasons still With the recent discovery of newer drugs, the duration of unknown, Europe was rid of the disease by the early nineteenth illness has been drastically reduced. The World Health Orga- century. The last known indigenously contracted leprosy case nization now recommends that duration of treatment for a in Britain was diagnosed in 1798 (Browne 1985). This decline paucibacillary case be six months and for a multibacillary case, of leprosy prevalence in Europe even before the advent of two years or until the skin smear becomes negative (WHO effective treatment and a century before the discovery of 1982), which may be on average three and one-half years. Even sulfone drugs has been mainly attributed to the changing though the present therapy still takes months and sometimes socioeconomic environment brought about by the industrial years to cure a patient, when compared with previous treat- revolution. ments, it has greatly improved the prospects for a quick cure. Leprosy is an infectious disease caused by Mycobacterium Since patients rarely die of leprosy, the case-fatality rate is leprae. The host response to this infectious agent depends on negligible (Walsh [1988] estimates it is 0. I percent). A few may the cellular immune mechanism. A person with a good cellular die from complications of septic wounds or from severe reac- immune response who develops leprosy is more likely to have tions. This low case-fatality rate has made the disease appear the milder tuberculoid type, whereas those with weak or lack- to be less serious, especially when priorities are being set for ing cellular immunity are likely to develop the lepromatous health care by politicians and health administrators according type (Bloom and Godal 1983). to mortality rates and public outcry. Leprosy tends to be given The clinical spectrum of leprosy varies from a single benign a low priority, but the disability and economic loss it causes, hypopigmented skin patch that may heal spontaneously to along with the social and psychological problems the patient widespread damage to nerves, bones, eyes, muscles, and kid- and his or her family suffer, make it more than an insignificant neys. Long-standing disease may produce severe mutilation of disease. the face and extremities, making the psychological trauma of The disability suffered by leprosy patients is a secondary leprosy victims at least as important as the physical suffering. consequence of nerve damage caused by infection with M. The Ridley and jopling classification system (Ridley 1974) leprae. The resulting anesthesia makes leprosy patients vul- defines the immunological spectrum of leprosy in clinical and nerable to accidental injury of anesthetic tissue. In 1970 histological terms, dividing it into indeterminate (i), tuber- WHO formulated a disability grading system which catego- culoid (TT), borderline tuberculoid (BT), borderline (BB), bor- rized disability from such injury into three grades. The derline lepromatous (BL), and lepromatous (LL) types. The system has since been simplified. It is still a three-grade multibacillary form of the disease (BB, BL, and LL) is highly system (grades 0, 1, and 2 rather than grades 1, 2, and 3) infectious and strongly positive on bacterial examination, with two separate sets of criteria, one for the hands and feet, 261 262 Myo Thet Htoon, Jeanne Bertolli, and Lies D. Kosasih and one for the eyes. The new grade 2, however, includes the appropriate denominator, the total number of cases of the previous grades 2 and 3. disease. Data on reaction incidence are usually from clinic The current grading system is as follows. For the hands and sources, but many undetected cases exist which are not feet, the criteria for grade 0 are no anesthesia and no visible counted in clinic data. deformity or damage. A grade 1 condition is indicated by the There was initial concern that with the introduction of new presence of anesthesia, but no visible deformity or damage, bacteriocidal drugs in multidrug therapy, more patients might whereas grade 2 is indicated by the presence of visible defor- have reactions due to the release of antigens from the killed mity ordamage. Each hand and foot is to be assessed and graded bacilli. But the results of studies of reaction incidence are separately. "Damage" in this context includes ulceration, encouraging. In a study by Boerrigter, Ponnighaus, and Fine shortening, disorganization, stiffness, and loss of part or all of (1988), the rates of the more serious type I reactions range from the hand or foot. For the eyes, criteria for grade 0 are no eye 76.9 per 1,000 person-years to 43.6 per 1,000 person-years in problems due to leprosy and no evidence of visual loss. Grade paucibacillary cases treated with the WHO multidrug therapy I is indicated by the presence of eye problems due to leprosy regimen. (Reactions are discussed later in this chapter.) Other but with vision remaining at six-sixty or better or the ability secondary complications of leprosy infections include tissue to see well enough to count fingers at 6 meters. Visual impair- necrosis, plantar ulceration, secondary bacterial cellulitis and ment is classified as grade 2 when vision is worse than six-sixty osteomyelitis, and progressive absorption of the digits of the or the patient is unable to count fingers at 6 meters. "Eye hands and feet. problems due to leprosy" include corneal anesthesia, lagoph- thalmos, and iridocyclitis. Each eye is to be assessed and Distribution and Risk Factors classified separately (WHO, Regional Office of the Westem Pacific 1988). Although leprosy is one of the oldest diseases known to The percentage of untreated patients who are disabled may humankind, very little is known about its natural history. reach 50 percent if the less serious forms (grade I ) of anesthesia The mode of transmission is still not known with certainry. arising from peripheral nerve trunk involvement are consid- The isolation of bacilli from loosened squamous epithelium ered. If only the more serious forms (currently grade 2) of of intact skin and from nasal washings of untreated lepro- disability are considered, the percentage is about 32 percent matous cases has led to the hypothesis that the portals of exit (WHO 1980). Among the newly diagnosed patients, the per- are ulcers on the skin and the respiratory tract (WHO 1985b). centage who are disabled (all grades) ranges from 2.7 percent The two widely accepted modes of transmission are direct in India (Mittal 1991) to 16percent insome areasofMyanmar skin-to-skin contact and droplet dispersion, but there is still (formerly Burma) (Myint, Htoon, and Shwe 1992). Although no clear-cut evidence to support their role in the spread of multidrug therapy has reduced the occurrence of disability, the infection. proportion of treated patients who are disabled remains high in some areas because many patients are diagnosed after irre- Age Distribution versible nerve damage has occurred. Early diagnosis and treat- ment are thus important in reducing the proportion ofdisabled The incidence of leprosy is bimodal. The first peak occurs patients. Another reason that disability may remain common between the ages of ten and twenty years and the second peak is the failure of leprosy control programs to incorporate the use between the ages of thirty and fifty (Dominguez and others of simple technologies for disability prevention as well as 1980). Clinical leprosy may occur in infants, but it does so patient education. It must be noted, however, that despite rarely and only in children of those with the disease. rigorous efforts, even regarding early diagnosis and treatment, some patients will develop disability. Sex Distribution The World Health Organization has estimated that two- thirds of the leprosy cases in the world are still unregistered, so In most endemic countries both the prevalence and the inci- there may be as many as 2 million undetected, partially dis- dence of leprosy is higher in males than in females. The higher abled patients (grade 2) in addition to the registered cases. It occurrence in males has been attributed to the greater mobility is estimated that 250,000 of those with leprosy are blind (vision of males, which provides greater opportunity for exposure and three-sixty). This figure increases if a visual acuity of less than contact with infectious cases. It may also be attributable to six-sixty is considered (Courtwright and Johnson 1988). failure to detect disease in females because of social attitudes, The clinical entity of leprosy represents a spectrum of im- which result in less thorough examination of females by health munological reactions to infection by M. Ieprae. Some patients workers. In a cohort study conducted in Myanmar from 1964 have skin patches only (uncomplicated leprosy), and others to 1976 in a population of 61,000 people living in highly have skin lesions complicated by immunological reactions endemic areas, the prevalence of leprosy for males was 42.2 ranging from mild to serious, producing such conditions as per 1,000 and that of females was 32.6 per 1,000. The neuritis, iritis, and increased inflammatory response in theskin incidence rate for males and females was 7.0 per 1,000 per lesions. Very little is known about the incidence of reactions year and 4.9 per 1,000 per year, respectively (Dominguez in leprosy patients because of the problems in defining the and others 1980). Leprosy 263 Host Factors onset of leprosy or lead to deterioration of a patient's clinical condition. These factors need to be studied in more detail to Until recently, humans were considered to be the only hosts gain further information. Recent studies in Ethiopia suggest and sources of M. leprae (Walsh and others 1977), but now that there is a relationship between pregnancy and the onset there is evidence that armadillos are naturally infected with an or reactivation of the disease (WHO 1985b). organism which is indistinguishable from M. leprae. There is little evidence that these animals transmit the infection to Current Prevalence and Trends humans, because armadillos are not found in areas of the world in which leprosy is hyperendemic. Estimatingthe numberof leprosycases in the world is adifficult That genetic factors as well as environmental factors have task because the projections have to be based on the registered a role in the pathogenesis of leprosy is demonstrated by both cases. These data may not be kept up to date, may contain twin and family studies which indicate that host genetics misdiagnosed cases, and may be affected by incomplete report- influence the type of disease that develops after infection (WHO ing. With consideration of these potential inaccuracies, it has 1985b). been estimated that the prevalence of leprosy in the world is 5.5 million cases. The geographic distribution of cases in the Contact Status six WHO regions is shown in appendix 12A, table 12A-1. Of the 3.7 million registered cases in all the WHO regional areas, The household contacts of persons with leprosy are at a greater Southeast Asia has the largest estimated number of patients, risk of infection than nonhousehold contacts. And household 2.7 million. In this region, the majority of the cases (2.5 contacts of those with lepromatous leprosy have a higher risk million) are in India. of infection than household contacts of those with non- There is no uniform distribution of either the disease itself lepromatous leprosy. A study conducted in the Philippines or the various clinical forms. At the subcontinental level, the found that the attack rate in nonhousehold contacts was 0.83 following patterns are recognized (WHO 1985b; see table 12A-l) per 1,000 person-years of observation. In contrast, incidence * Tropical and subtropical belt of Africa and southern Asia. rates in household contacts of nonlepromatous and leproma- This is considered to be the original source of leprosy; in tous cases were 1.6 and 6.23 per 1,000 person-years of obser- these areas the disease can be traced back at least 2,500 years vation (WHO 1985b). A similar study in Myanmar found that and remains endemic today. the cumulative incidence among nonhousehold contacts was 5.9 per 1,000 population per year and 21.9 per 1,000 per year * Mediterranean basin. Leprosy has probably been present among household contacts of lepromatous cases. For border- in this region for 2,000 years, and still persists today, al- line cases it was 10 per 1,000 per year and for indeterminate though prevalence of the disease is low and declining. and tuberculoid cases it was 7.6 per 1,000 per year (Lwin and * Northern Europe. Leprosy was widespread in this area others 1985). 1,000 years ago, reaching as far north as the Arctic Circle; Leprosy has been generally associated with poverty, and however, from the thirteenth century, the prevalence of the crowding may facilitate transmission (Noordeen 1985). Eco- disease has declined progressively. The decline observed in nomic development has been proposed as a reason for the some populations of North America is in some ways compa- decline of leprosy prevalence in Europc and Japan. Consistent rable to the decline that occurred in northern Europe. with this hypothesis is the twenty-year lag in the decline of * South and Central America. Leprosy was introduced into leprosy (relative to the rest of the country) that was observed this region from Europe and Africa, and the disease remains in the Okinawa prefecture in Japan, which had the slowest rate endemic today, although the incidence is declining in Ven- of economic development in the country. These shifts in risk ezuela. factors appear to occur independently of any leprosy control . Pacific.slandsandAustralia.Leprosyhasbeenintroduced activity, because they have been observed in countries with into several island populations during the last 200 years, and strict or relaxed isolation policies and in the absence of any in some instances epidemics have occurred that have lasted other control measures. Once the situation is reached in which for several decades. each new lepromatous case fails to produce, on average, one new secondary lepromatous case, the capacity for endemic The lepromatous proportion of cases (as estimated from case persistence is broken, and disease incidence will gradually registries) also differs from region to region. In Europe this decline and ultimately reach zero (WHO 1985b). This process proportion is 20 percent, whereas in Sub-Saharan Africa it is has important implications for leprosy intervention programs 5 percent. In northern Asia it is between 5 and 20 percent in endemic areas. (WHO 1985b). These differences may be the reflection of ge- netic, environmental, and cultural differences in the host Physiological Factors populations. Diagnostic and reporting practices may also help explain the variation in the lepromatous proportion. The Claims have been made that puberty, menopause, pregnancy, immunoepidemiologic studies conducted in Sri Lanka show lactation, stress due to infections, and malnutrition favor the that 5 percent of the total household contacts studied had 264 Myo Thet Htoon, Jeanne Bertolli, and Lies D. Kosasih antibodies to phenolic glycolipid antigen, which is produced The productivity loss due to deformity from leprosy in India by M. leprae. In the hyperendemic community of Micronesia, was evaluated in a survey by Max and Shepherd (1989) of 550 approximately 10 percent of the population have been found leprosy patients randomly sampled from a rural and an urban to have this antibody (WHO 1985b). area in the state of Tamil Nadu, India. Their analysis showed In endemic areas, the prevalence of leprosy is maintained that elimination of deformity would raise the probability of at a relatively constant level. During the past century, how- gainful employment from 42.2 to 77.6 percent. The authors' ever, increases in the prevalence of leprosy have occurred in extrapolation to all of India's estimated 645,000 leprosy pa- some areas of the world according to a distinct temporal tients with deformity suggested that elimination of deformity pattern. would raise productivity by$130 million peryear. This amount In Nauru, Ponape, and Truk, Hawaii, IrianJaya, and eastern is one-eleventh of India's entire official development assis- Nigeria, leprosy epidemics have been reported with prevalence tance for all purposes from all sources in 1985 ($1,470 million). proportions of 10 to 30 percent. These epidemics have ap- In addition to the cost of loss ofproductivity, there are social peared in communities in which leprosy has been introduced costs associated with the loss of healthy life brought about by only recently and where environmental factors favor the rapid leprosy. These include (but are not limited to) the burden on spread of infection. The epidemics are characterized by a rapid family members of living with and caring for a disabled relative. increase in the incidence of paucibacillary disease and a very The cost of these social consequences should be considered as low initial incidence of multibacillary leprosy, little clustering part of the indirect costs of leprosy. Though they are not among household contacts, and fairly equal distribution of insignificant, these social costs are not readily calculable. They cases by age. Incidence reaches a peak and is followed by a fairly will vary from community to community and will be associated rapid spontaneous decline (perhaps attributable to the infec- with the social and cultural values each society attaches to an tion of all susceptible individuals). During this decline the individual's life. proportion of multibacillary forms of the disease increases, a The direct cost of an illness refers to the costs of medical shift is seen toward a higher incidence in children, and there care (paid by patients or society) to control or treat the illness. is increased disease clustering among household contacts. The direct costs of leprosy include the costs of drugs, drug Gradually the conventional pattern of endemic leprosy delivery, supervision of drug delivery, laboratory facilities for emerges (WHO 1985b). diagnosis and for monitoring response to treatment, and recon- Irgens and Skjaerven (1985) report that epidemiologic sur- structive surgery and rehabilitation. The effectiveness of a veillance in Norway, the United States, Nigeria, Japan, Ven- control program depends greatly on adequacy of case detec- ezuela, India, and China, covering periods from 1851 to 1981, tion, which in turn depends on the level of knowledge of has indicated a consistentdecline in incidence rates of leprosy. healthworkers and members of the community. Therefore, the At the same time, the age at onset, the ratio of male cases to costs of case detection or screening efforts as well as training female cases, and the proportion of multibacillary cases have of health workers and education of the community must be been increasing. The increasing age at onset may be attribut- included along with the direct costs. able to postponement of infection to a later age or an increas- Direct costs of leprosy have not been assessed systematically. ing fraction of patients with long incubation periods or both. Lechat and others (1978) tested the relative cost-effectiveness The increasing importance of long incubation periods is con- of current or potential control methods using a computer sistent with the shift toward multibacillary cases, in which the simulation model. The model has potentially serious limita- incubation period is longer than that in paucibacillary cases. tions, however (Lechat 1981), and so will not be discussed This mechanism was also reported during the decline of further here. Wardekar (1968) found that the development of tuberculosis. Irgens and Skjaerven (1985) propose the general deformity results in increased use of medical care. Deformity principle that an increasing fraction of new patients with long thus adds to direct costs as well as indirect costs. In addition to incubation periods, resulting in an increasing age at onset, the cost of medical care borne by the leprosy patient, wages should be expected in any disease in rapid decline which also lost while seeking care must also be considered as part of the has a long and varying incubation period. This theory offers a direct cost. A more detailed discussion of costs and effective- basis for assessment of secular trends. ness of various control programs is presented in appendixes 12B-12G. The total cost of a leprosy control program is Economic Costs estimated in appendix 12H. The indirect costs of an illness are all costs other than those Prevention Strategy for health care. As mentioned earlier, leprosy affects people in the prime of life, with peak incidence between ten and twenty Though leprosy has plagued humankind for centuries, gaps in years of age and again between thirty and fifty. As many as 16 knowledge about the disease still exist, especially with regard percent (Myint, Htoon, and Shwe 1992) of those with the to the natural history ofdisease and transmission. Even though disease may have serious (currently grade 2) disability with great advances have been made in the microbiology, immuno- concomitant loss of productivity. Thus, the indirect costs from logy, epidemiology, and treatment of leprosy during the past loss of productivity might be expected to be significant. two decades, from 1966 to 1985 there was a 90 percent increase Leprosy 265 in the number of registered cases. But more recently, from 1985 to 20 percent in Myanmar (Lwin and others 1985; see table to 1990, the number of registered cases has declined by more 12A-3). In a recent case-control study conducted in Tamil than 30 percent. This decline has been attributed to the Nadu, India, researchers found that BCG vaccine appears to implementation of multidrug therapy and the consequent re- increase the risk of indeterminate leprosy while offering 61 lease of large numbers of patients from treatment. Multidrug percent protection against borderline disease (Muliyil, Nelson, therapy is also thought to increase the proportion of patients and Diamond 1991). who present at an early stage of the disease, and thus it has a The wide range of effectiveness has been attributed to role in preventing and reducing the number and degree of different vaccine strains, regional and racial differences, and deformity among new cases (Noordeen, Bravo, and Daumerie the prevalence of environmental mycobacteria. Despite the 1991) The risk factors and the primary prevention measures large number of studies of the subject, however, the effective- employed in leprosy control are presented in table 12A-2. ness of BCG vaccine for leprosy control is still not well under- Leprosy is a disease with high infectivity but low pathoge- stood. The cost per dose of BCG is approximately $0.05 (UNICEF nicity and virulence. At present no specific primary preventive 1991). measure has been devised, and until this occurs, the main A combination ofkilled M. leprae and BCG vaccine was used strategy for the prevention of leprosy must be early diagnosis for the treatment of borderline lepromatous and lepromatous and adequate and regular treatment. patients in Venezuela (Convit and others 1982) with promis- ing results. Recently Talwar and others (1990) have found that Behavior lepromatous patients given a combination of multidrug ther- apy and a candidate antileprosy vaccine based on Mycobacte- Many studies on social factors influencing leprosy have shown num w show distinctly better clinical improvement. Such that it is a disease affecting the lower socioeconomic classes observations that vaccines may be effective even when given and that poor nutrition, sanitation, and personal hygiene, as after infection has occurred may revolutionize control efforts well as overcrowding, are some of the factors which interact to in many endemic communities. influence persistence of the disease (Noordeen 1985). Im- Even if BCG is only 30 percent effective in preventing lep- provement of economic conditions is beyond the scope of rosy, as was reported from southem India and Myanmar efforts to prevent leprosy. Still, any improvement of nutrition, (Noordeen 1983; Lwin and others 1985), this is good news for sanitation, and personal hygiene through more broadly based leprosy control, because the 30 percent reduction would be a health and education efforts should reduce the incidence of spillover benefit from tuberculosis control programs already in leprosy as well as that of a number of other diseases. place. This spillover benefit is another reason to support BCG vaccination in childhood immunization programs. Further Environment studies must be conducted to determine the optimal age for vaccine administration for protection against both tuberculo- Lack of an adequate supply of clean water can lead to poor sis and leprosy. Because these two diseases have different personal hygiene, which in tum may promote the transmission incubation and exposure periods, a single dose of vaccine, of leprosy, especially the skin contact mode of transmission. which is sufficient to prevent tuberculosis, may not be suffi- Though no supportive evidence has been presented for this cient to prevent leprosy. (See appendix 12E for discussion of theory, it is conceivable that the improvement in domestic the cost issues involved in immunization.) water supplies in communities in which leprosy is highly With use of recombinant DNA techniques, it may be possible endemic may disrupt the chain of transmiss.on. in the near future to develop a multivaccine which is effective The high incidence rate among household contacts of lep- against both leprosy and tuberculosis. It will be necessary, rosy cases may be attributable to overcrowding (Noordeen however, to leam more about the antigens relevant to protec- 1985). If droplet spread of infection does, in fact, occur in tive immunity. Katoch and others (1990) review recent prog- leprosy transmission, it is feasible that it might be facilitated ress in applying advances in molecular biology of M. leprae to in close quarters. Dealing with overcrowding is best done the problem of vaccine development. Several trials of vaccines through a multisectoral approach involving population con- against leprosy are currently planned or under way. trol, economic development, education, and health care ser- vices. Leprosy control would be only one of the many benefits Health Education of such a strategy. Education regarding leprosy should be directed toward two Immunization different groups: first, leprosy patients and their immediate family, and second, the community. The nature of the educa- Several follow-up studies of the efficacy of BCG (bacille Cal- tional material should also be different for these two groups mette-Guerin) vaccine for the prevention of leprosy have been because the message or aim of the education program in each conducted. These studies have consistently found a protective case is different. Leprosy patients should be thoroughly edu- effect of BCG against leprosy, but the vaccine effectiveness cated about the nature and natural history of the disease. ranges from 80 percent in Uganda (Stanley and others 1981) Certain age-old beliefs should be delicately handled, as they 266 Myo Thet Htoon, Jeanne Bertolli, and Lies D. Kosasih may not be easily dispelled. The importance of regular compli- usually have advanced disease and are already disabled. Such ance with prescribed therapy should also be stressed. Education advanced cases have also acted for some years as a source of of leprosy patients should be an ongoing process, and although infection in the community. Active case finding usually in- it is time consuming, it should not be neglected. Failures volves screening of schoolchildren, contact surveillance, and in treatment with dapsone in the past have been attributed mass population screening activities (Lechat 1985). Pre- to neglect of patient education, although it is hard to pin- employment screening of workers and military recruits may point the number of relapses attributable to lack of health also be done but is usually not of much significance in leprosy education. control. Education to prevent injury in patients with nerve damage At present there is no specific screening test for leprosy may reduce the loss of productivity which is so costly in leprosy. which can discriminate an active from an inactive case with The patient's role in preventing permanent injury to hands certainty (the current status of serological techniques in the and feet can be significant if he or she is taught to avoid epidemiology of leprosy is reviewed in Bharadwaj and Katoch activities that may potentially cause injury, to inspect anes- 1990). Screening is based on clinical examination. This re- thetic tissue regularly, and to practice early and correct wound quires special training of health workers and is an expensive, care. Similarly, patients should be taught to prevent the op- time-consuming activity. The advantage of a well-conducted thalmologic complications of leprosy by avoiding eye dryness screening program is that cases are usually detected in their and protecting the eyes from injury (Watson 1988). early stages. Because screening is costly, wyio has laid down Educating the immediate family and the community is also guidelines for countries in which prevalence of leprosy is low necessary, especially in highly endemic areas. This will pro- (less than 1 per 1,000 population) and resources and personnel mote the early diagnosis of cases by making the community are lacking. In these countries, WHO recommends focusing only more aware and accepting of the disease. The goal of the on surveillance of household contacts of lepromatous cases for program would be to encourage patients to come forward and a minimum of ten years after the index case is bacteriologically obtain treatment rather than being ashamed of and concealing negative and on surveillance of household contacts of non- theirdisease.Thisisimportantbecausethroughearlydiagnosis lepromatous cases for five years from the time of diagnosis a significant reduction in permanent disability can be achieved of the index case. If these tasks prove impossible, it is re- (Lechat 1985). Another goal of the health education effort is commended that contacts be examined at least once (WHO to dispel social stigmas through promotion of understanding of 1977). the disease. This will encourage a social environment in which Mass population screening programs are cost-efficient only the leprosy patient is treated in the same way as a person with in highly endemic communities and require careful planning any other disease. This social support is necessary if the leprosy and teamwork so that at least 95 percent of the population is patient is to lead the socially and economically productive life covered. Without such high coverage the program may be to which he or she is entitled according to World Health unsuccessful, because those who are unwilling to come forward Assembly resolution WHA30.43. to be treated will continue to act as reservoirs of infection (WHO Education of the community should be conducted as part of 1980; see appendix 1 2E for further discussion of the cost issues a general health education program. Education of persons with involved in screening). leprosy and their families should be a regular part of treatment. Because most of the control programs in developing coun- Although including education in treatment programs will add tries employ paramedical health workers for leprosy diagnosis to their cost, the costs of neglecting education must also be at the village level, diagnostic criteria are needed which are considered. In the past, leprosy control programs which ne- simple and easily taught. The World Health Organization's glected education had limited effectiveness because of social four cardinal signs are useful for this purpose, although they discrimination, poor drug compliance, and lack of openness in may not be highly sensitive or specific. These four criteria are dealing with the disease (Lechat 1985). (a) a hypopigmented or hyperpigmented patch or macule, (b) an enlarged hard or tender nerve, (c) anesthesia in the area of Treatment the skin lesion, and (d) a positive skin smear. At least two of the first three criteria must be present. If modern chemotherapy is to be of any help to leprosy patients, With the introduction of more effective multidrug therapy, early diagnosis is crucial. It is important that cases be diagnosed the cost of treatment per case has increased. Outlay of funds before deformity has occurred so that they may be cured for drugs in specific control areas will also depend on the without leaving behind residual signs of this stigmatizing dis- specificity of the health workers' diagnosis. Training of health ease. Early case-finding activity must be promoted by leprosy workers thus improves the cost-effectiveness of treatment. control programs. Case finding may be passive or active. Pas- In the future, if diagnostic tools such as the polymerase chain sive case finding happens when patients come on their own to reaction (PCR) technique are perfected, earlier and more rapid obtain treatment. It is greatly improved by increasing the diagnosis of leprosy may be possible. At present, however, PCR accessibility of health care and also by promoting health edu- is not easy to adapt to field conditions for several reasons, cation. A drawback of relying on passive case finding is that including the required use of radioisotopes and the expense of persons who are motivated to seek treatment on their own setting up the technique (Bloom 1990). Gillis and Williams Leprosy 267 (1991) review possible uses of the PCR technique for studying tached to leprosy, primary health care workers are often not leprosy. highly motivated to work with leprosy patients, and case finding and management have suffered. This experience shows Good Practice and Actual Practice that training of primary health care workers should include education about leprosy and encouragement of the workers to Treatmnent of leprosy is designed to render infectious patients take a more active role in case finding and management. noninfectious. Because there is no known animal reservoir for Because primary health care workers are not specifically the disease, the chain of transmission could be interrupted if trained to diagnose leprosy patients, they may be inaccurate infectious cases in the community were made noninfectious diagnosticians. In particular, if their diagnoses are not highly through adequate and regular treatment. The chances of specific, drugs and other resources will be wasted on patients achieving control are greatly increased when at least 90 per- who are false positives. In addition, patients who are falsely cent of the estimated multibacillary (infectious) cases are diagnosed will undergo unnecessary psychological stress. registered and treated regularly (WHO 1980). Historically, fail- Therefore, it is important that primary health care workers be ure of many control programs in hyperendemic countries has trained to improve their diagnostic skills and that the training been attributed to poor coverage and inadequate and irregular be repeated at regular intervals to maintain a high level of treatment of multibacillary cases. The number of new unreg- diagnostic accuracy. istered multibacillary cases is thus a useful indicator of contin- With the integration of leprosy control activities into pri- ued transmission (WHO 1980). mary health care programs, other activities such as WHO's As mentioned previously, a decline in the prevalence of Expanded Programme on Immunization compete with leprosy leprosy has followed the introduction of MDT. Although MnDT is control for health care workers' time. In countries where thought to have led to a reduction in the number and degree leprosy is hyperendemic and resources are limited, the role of of deformities among new cases, it does not completely prevent leprosy control may be reduced to drug delivery activities the reactions which lead to deformity (Noordeen, Bravo, and alone. Active case finding may have to be neglected, which Daumerie 1991; Cellona and others 1990). The increasing will in turn result in a greater number of new cases with acceptance of MDT among national health services and leprosy disability. The cycle of disability, social stigma, economic patients themselves is due to the fixed and relatively short losses, poor treatment compliance, and further development duration of treatment; the low levels of toxicity and side of disability would then be perpetuated as in the days of effects; low relapse rates following completion of treatment; dapsone mnonotherapy. Another danger is that poor compli- and the reduction in frequency and severity of erythema ance may lead to resistance to the newer drugs, as it did for nodosum leprosum reactions (Noordeen, Bravo, and Daumerie dapsone monotherapy (WHO 1982). 1991). Effective coverage of patients with MDT differs widely from country to country (Declercq and Gelin 1991), although Case Management on average it had reached 56 percent by October 1990 (Noordeen, Bravo, and Daumerie 1991). Treatment of leprosy is simple in the sense that there are very The mode of delivery of treatment depends on the available few alternative choices of drugs, hut successful treatment is resources, caseload, communication, manpower, and the type actually difficult. Often, case management has been simply of drug schedule used. Two types ofdelivery used in hyperende- equated with issuing patients the necessary drugs. But the mic countries are domiciliary treatment and stationary clinic psychological and social aspects of the disease must not be treatment. Domiciliary treatment results in high coverage but overlooked by health care workers, though this kind of com- is very costly. especially when supervised, because treatment prehensive care can be difficult when caseloads are high. with rifampicin and clofazimine must be given at least once a Health workers are often not trained to tackle the complexity month. With domiciliary treatment, the delivery cost may of the problems leprosy presents. Failure to treat the leprosy outstrip available resources. In the latter case, the service area patient as an individual in a comprehensive manner has often of the treatment program may have to be reduced, with resulted in high treatment failure rates in countries in which rotation to a new area after completion of a minimum of two leprosy is a significant problem. This high failure rate has also years treatment, or the program may have to be operated been attributed to the long duration of therapy. In the days of through a stationary clinic open at fixed times on fixed days of monotherapy with dapsone, paucibacillary patients were usu- the week. ally on a daily dapsone dose for five to seven years, and Treatment of locally endemic disease was one of the goals multibacillary patients were frequently on a daily dose for their set for primary health care by the World Health Organization lifetime. Lapses in compliance were common with such a long at the 1978 conference at Alma Ata. As a result of this course of treatment, especially when unsupervised. This prob- recommendation, vertical leprosy control programs were dis- lem was further compounded by the irregular distribution of mantled in favor of integration of leprosy control into primary drugs by health care workers. health care programs. The expectation was that with more In the past, despite uneven patient compliance, treatment personnel working in leprosy control, case finding would im- was often continued indefinitely because of the possibility of prove. In practice, however, because of the social stigma at- relapse. This resulted in great demand on leprosy clinics and 268 MNO Thet Htoon, Jeanne Bertolli, and Lies D. Kosasih control workers and precipitated a decline in the quality of months. A study by Neik and others (1988) found that 53 treatment (WHO 1982). percent of those with multibacillary leprosy who were treated Many control programs continue to have high dropout rates. with the multidrug regimen recommended by WHO were ren- A number of investigations have indicated that even the dered smear negative within two years and 94 percent within patients who collect the drugs regularly from leprosy clinics do four and one-half years. not necessarily ingest them. Researchers of earlier studies in Unless new developments in vaccine production occur Malawi (Ellard and others 1974) and Myanmar (Hagan and within a few years, there is little reason to expect that there Smith 1979) used urinary analysis to monitor the regularity of will be a change in case management in the next decade. Any dapsone self-administration. They found that not more than uncontrolled trial of a vaccine, of chemotherapy, or of a 50 percent of the outpatients had taken the prescribed dose in chemoprophylaxis program in a population that is beginning the three days prior to attendance. to undergo a "natural" shift in risk factors is bound to appear a This irregular treatment is suspected of having been the success, although slowly. Even a controlled trial that alters main reason for the development of dapsone resistance. In environmental or host factors may not show rapid changes in 1981 it was estimated that the prevalence ofdapsone resistance subsequent incidence rates for the population because of the for all of Malaysia was as high as 10 percent of all treated additional effects accompanying the natural decline described multibacillary patients (WHO 1988b). The secondary dapsone above. There is a minimum incubation period for leprosy, but resistance in India was estimated at 23 per 1,000 multibacillary the maximum interval from primary infection to disease onset cases in 1978 and in Jiengsin Province, China, it was 51 per may be as long as a lifetime. Thus cases may continue to appear 1,000, compared with 35 to 40 per 1,000 multibacillary cases for many years after the effective cessation of transmission, in Shanghai (WHO 1982). Pearson and his colleagues (1977) unless it becomes possible to eradicate the residual bacilli in reported an even more disturbing situation in Ethiopia. The all infected persons (Dharmendra 1986). resistance proportion was found to be 190 per 1,000. Along The case management strategy for leprosy can be divided with reports of secondary dapsone resistance from all over the into four categories. world, primary dapsone resistance has been reported in Ethio- * Treatment of leprosy pia, India, the Philippines, and Malaysia (WHO 1982). * Treatment of leprosy Fortunately, the effectiveness of the multidrug alternative * Treatment of complications of leprosy to dapsone monotherapy has been quite encouraging. Boerrig- * Reconstructive surgery ter, Ponnighaus, and Fine (1988) found a relapse rate of 4.17 * Psychological therapy and rehabilitation per 1,000 person-years among paucibacillary cases treated with Each category presents a unique treatment problem and the multidrug regimen recommended by WHO, as compared Eackledory presentsta of trease procem ase with 12^.9 per 1,0 person-years.. obere by .easn Chris- must be tackled at different stages of the disease process. Case with 12.9 per 1,000 person-years observed by nesudasan, Chris- management interventions for leprosy are summarized in table tian, and Bradley (1984) among cases treated with dapsone 12A-4. alone. The World Health Organization (Noordeen, Bravo,avo and Daumeriel1991) has recently reported arelapse rate ofQ0.10 The primary and secondary level of management facilities percndDaueriyear 1991)hasrepaucibacntlyreportedrelpsa erateo 0 should be general health care facilities that are already estab- percent per year for paucibacillary patients and 0.06 percent lished in the community. This is important not only from the per year for multibacillary patients. economic and feasibility aspect but also from a psychological In a three-year assessment of multidrug therapy in India by standpoint. Separate facilities for the treatment of leprosy Ganapati, Revankar, and Pai (1987), it was observed that out add to the isolation and stigmatization of patients with this of an initial 253 persons with multibacillary leprosy who were ase. treated, 67 percent were smear negative after twenty-four doses disease. (two-year standard WHO regimen) and 75 percent were smear negative after thirty-six doses. In the same study, 18 patients Treatment of Leprosy who were smear positive at the end of the two-year multidrug treatment were observed without further treatment; it was The treatment of leprosy with multidrug regimens serves four found that 17 of them went on to become smear negative purposes. First, it interrupts the chain of transmission in the within two years. This study showed that multidrug therapy community by rendering infectious cases noninfectious. has a distinct advantage over dapsone monotherapy and is an Though multibacillary cases may have infected others prior to effective regimen for both individual treatment and public treatment, one can assume that once they are treated, their health intervention. role in the transmission of the disease has been diminished. In a follow-up study of 129 persons with paucibacillary Second, it cures patients of the disease. Third, it prevents the leprosy who were treated for one year with the multidrug emergence of drug-resistant strains in the future. Last, it halts regimen recommended by WHO, Ramanan and others (1987) the disease process, and if the disease is treated in its early stage, found that 83.7 percent had clinically active leprosy at the end it will prevent further development of deformity. of the year of treatment. Further studies are necessary to Treatment regimens for leprosy are considered separately for determine whether it is advisable to continue treatment of paucibacillary and multibacillary cases. Paucibacillary leprosy those with paucibacillary leprosy beyond the recommended six includes indeterminate and tuberculoid cases in the Madrid Leprosy 269 classification (WHO 1980) and i, Tr, and BT leprosy in the Ridley in which second-line drugs, such as ethionamide or protionam- and Jopling classification (Ridley 1977), whether diagnosed ide, must be introduced. clinically or histopathologically. The bacteriological index The role of the secondary level of management in the must be less than two at any site (according to the Ridley scale treatment of leprosy is limited and should be mainly concerned [WHO 1982]). The World Health Organization has recom- with treatment and diagnosis of the side effects of the multiple mended a short course of therapy for paucibacillary cases with drug regimens (such as sulfone allergy to dapsone and liver 600 milligrams of rifampicin once a month for six months to toxicity from rifampicin) which are being prescribed in the be given under supervision along with 100 milligrams of dap- catchment area of that facility. The tertiary level of manage- sone (1-2 milligrams per kilogram of body weight) daily for six ment should be concemed with research and development of months. The cost of drugs for the six-month course of treat- new drugs and treatment procedures and monitoring of drug ment for a paucibacillary case is $2.15. It is estimated that the resistance and side effects. cost of delivery of the drugs is $4.50 and the cost of supervision of drug delivery an additional $1.35, bringing the total cost to Treatment of Complications $8.00 (see appendix 12C for details). It is recommended that if treatment is interrupted, the Two types of reactions occur in leprosy patients. Type I reac- regimen should be started again where it was left off and the tion, which is seen commonly in cases of the tuberculoid full course completed. If relapse occurs after termination of spectrum, is the most serious type of the two because the treatment (release from control or discharge), the same treat- patient may suffer from severe nerve damage within days and ment is to be restarted (WHO 1980). needs immediate referral, prompt care, and in some cases even Multibacillary leprosy includes both lepromatous and bor- hospitalization. Type 11 reaction, which is also known as ery- derline leprosy in the Madrid classification (wHO 1980) and LL, thema nodosa leprosum (ENL), is seen in the lepromatous BL, and BB leprosy in the Ridley and Jopling classification spectrum of cases. Although it does not give rise to nerve (Ridley 1977). The regimen recommended by WHO for a multi- damage immediately, ENL is an important clinical problem bacillary case is 600 milligrams of rifampicin and 300 milli- because patients may suffer from repeated episodes, which may grams of clofazimine in a single dose once a month to be given cause a loss of confidence in the treatment regimen. under supervision along with 100 milligrams of dapsone and Reactions are only occasionally medical emergencies, al- 50 milligrams of clofazimine daily to be self-administered. This though type I reactions require urgent action because of the treatment is to be followed for at least two years and is to be potential for nerve damage and deformities. The most effective continued wherever possible until the patient achieves skin treatment at present forboth types ofreactions isprednisolone, smear negativity (WHO 1988). The average duration of treat- which should be given in a very high dose, as much as 20 to 80 ment for multibacillary cases should be about three and one- milligrams per day, to be slowly reduced, depending on the half years (S. K. Noordeen, Chief Medical Officer, Leprosy, clinical response, to a daily maintenance dose of 5 to 10 Division of Communicable Diseases, WHO, personal communi- milligrams. The duration of treatment varies from patient to cation August 30, 1989). The cost of drugs for a course of patient. The daily prednisolone requirement may be discon- treatment of three and one-half years for a multibacillary case tinued within two to three months or continued for more than is $81.66. The cost of delivery is estimated at $31.50, and two years. Other general supportive therapy such as analgesics supervision of delivery is an additional $9.45, bringing the and sedatives, along with adequate rest and nutrition, is also total cost of treatment to $122.61 per case. Considering important (WHO 1980). that approximately 20 percent of cases are multibacillary Thalidomide has proven effective for ENL cases, with few and 80 percent are paucibacillary, the average cost per case toxic effects. But its known teratogenic effect prevents wide- is $30.92, a weighted average (see appendix 12C for further spread use in the field from being a feasible option even though discussion). the drug is relatively inexpensive (WHO 1980). Clofazimine in The case management strategy outlined above should be a 200 to 300 milligram daily dose has an anti-inflammatory implemented at all the four levels of management, namely effect, but it takes four to six weeks to exert its effect and is household, primary, secondary, and tertiary. Apart from surgi- toxic if used for long periods (Jacobson 1985). Its use in the cal care, both diagnostic and medical care should be provided treatment of reactions is limited to severe cases, but if used at these four levels but at different levels of sophistication. together with prednisolone it may lower the daily requirement At the household level, simple diagnostic procedures should of the latter. be performed by primary health care workers and the necessary Ulcers, another complication in leprosy, are a consequence drugs should be distributed if resources are available. At the of anesthesia occurring in the extremities due to peripheral primary management level, both diagnostic and medical treat- nerve damage. Once the peripheral nerves have been dam- ment should be made available, especially when lack of re- aged, nothing can be done to reverse the damage. Care of the sources prevents delivery of services at the household level. anesthetic extremity becomes extremely important for the Even in instances in which household care could be provided, prevention of ulcers. The precipitating cause of ulcers are care at the primary level is still necessary for supervision and burns, mechanical pressure, and accidental injury to the hands monitoring of therapeutic response and also in circumstances and feet. The most effective way to treat ulcers is to immobilize 270 Myo Thet Htoon, Jeanne Bertolli, and Lies D. Kosasih or rest the affected part of the extremity. Antibiotics, along * Vocational training, which may be given in an institu- with surgical dressing, will also shorten the course of the ulcer. tional setting or on a part-time basis. Such training should Treatment of mild nerve damage and ulcers may be under- not disrupt the patient's life. taken at the household level with frequent visits to the health * Prevention of disabilities by simple methods that can be care facility for necessary drugs and supervision. If a reaction applied in the field, such as simple forms of physiotherapy, is severe, with signs of nerve compression, the patient should provision of special footwear and crutches, and the teaching be referred to a secondary facility. Ulcers complicated by of patients to avoid injury and, when it occurs, to attend to osteomyelitis will require surgical intervention, which may be it early (Watson 1988). obtained by referral to a secondary-care facility. Tertiary care may be needed in some few cases in which reactions cannot be . . . controlled or if osteomyelitis and osteoarthritis have become Priorities for Operations Research so advanced that amputation is necessary. As mentioned in the section on prevention, more research is necessary to understand the variability in the efficacy of BCG Reconstructive Surgery vaccine for the prevention of leprosy and to develop a more effective vaccine. Preliminary research has indicated that cer- Because reconstructive surgery is expensive and is not import- tain vaccines against leprosy used in combination with MDT as ant for control of leprosy, it has been neglected by many immunotherapeutic agents appear to be effective in the treat- control programs. But reconstructive surgery is very important ment of leprosy patients. More studies are needed to evaluate for the social and psychological well-being of the patient. these findings. In addition, the sensitivity analyses presented Correction of various deformities will also increase the eco- in appendix 12D indicate that further efforts are necessary to nomic productivity of leprosy patients. A further consequence determine more precisely the reduction in permanent disable- is the promotion of trust in the health care system, and along ment brought about by treatment of complications. Additional with it, drug compliance. research that should be done includes study of the mechanism Reconstructive surgery must be obtained in highly special- of transmission, incubation period, sensitivity and specificity ized tertiary-care facilities with a staff of reconstructive sur- of diagnosis by various methods, and the morbidity risks asso- geons. The direct benefits in dollars are hard to estimate, ciated with early lesions of different types. Development of especially when self-esteem and happiness are the greatest drugs that act faster than the current ones should also be benefits of these procedures. Reconstructive surgery is required investigated and the effectiveness of intervention programs in 2 to 5 percent of all cases (Myanmar, Ministry of Health determined. 1985). Priorities for Resource Allocation Psychological Therapy and Rehabilitation The application of basic concepts and principles of economic It is well known that many leprosy patients suffer from psycho- analysis and management science suggests that early diagnosis logical and social problems. Many control programs focus their and adequate and appropriate treatment should be given the efforts on drug distribution and treatment of complications and highest priority in leprosy control. Current projections indi- pay little attention to the psychological and social problems of cate that a 60 to 80 percent reduction in the prevalence of the patients. This lack of attention may be contributing to high leprosy may be achieved within five to seven years with wide- dropout rates. Primary health care workers are normally not spread use of MDT (Noordeen, Bravo, and Daumerie 1991). trained to provide psychological or rehabilitation services, but Even if such a reduction in prevalence can be realized, disabil- they should at least be trained to identify problem cases and ity in cured patients will still be a problem for many years. In refer them to secondary referral facilities. addition, a continuing slow trickle of new cases (infected many Rehabilitation in leprosy involves the combined and coor- years earlier) can be expected because of leprosy's long incu- dinated use of medical, social, educational, and vocational bation period. The second priority for resource allocation measuresfortrainingorretrainingthe individualtothehighest should thus be research, with emphasis on development of a possible level of functional ability. The surest and least expen- vaccine effective in preventing transmission and development sive rehabilitation is to prevent physical disability, which is of drugs with quicker action. Finally, resources should be done through patient education about self-care and through allocated to prevention and treatment of disability in patients early diagnosis and treatment by clinicians. Rehabilitation with nerve damage. must begin as soon as the disease is diagnosed (WHO 1980). Rehabilitation activities include: Conclusion * Health education of the public so as to reduce prejudice It is our conclusion that the benefits of leprosy control pro- against the disease. grams have been underestimated in the past and that the effect Leprosy 271 of the disease is significant in many developing countries. setters to consider the significant intangible benefits of a Studies of the currently recommended multidrug therapy leprosy control program as well as the cost per YHLG presented program have found that it is effective in controlling lep- here. rosy. Furthermore, the cost-effectiveness analysis presented in this chapter indicates that the cost per year of healthy life gained (YHLG) from implementation of this regimen is quite Appendix 12A. Tables low. Ranking leprosy control on a list of health care priori- ties will, of course, depend on its comparison with interven- The tables below summarize incidence and risks of leprosy and tion programs for other diseases. Still, we urge the priority its prevention and management. Table 12A-2. Risk Factor and Points of Prevention for Leprosy Table 12A-1. Registered Leprosy Cases, by WHo Riskfactor/point Relevance Region of prevention Mechanism to leprosy Prevalence Incidence Beha WHO region (per 10,000) (per 10,000) Bea v gor Personal hygiene Transmission Low Africa 9.2 0.71 Environmental Americas 4.2 0.42 Domestic water Transmission High Southeast Asia 20.5 3.72 Europe 0.1 0.00 Public health Eastern Mediterranean 2.6 0.15 Immunization with BcG Western Pacific 1.0 0.09 vaccine Resistance to infections Medium Health education Transmission Medium Total 7.1 1.09 Treatrnent Source: Noordeen, Bravo, and Daumerie 1991. Early diagnosis Transmission High Adequate and regular treatment Transmission High Source; Authors. Table 12A-3. BCG Vaccine Efficacy in Control of Leprosy Age at vaccination Follow-up Vaccine efficacy Area Subjects First year (years) Vaccine used (years) (percent) Source Uganda 16,150 1960 0-10 (3laxo 8 80 Stanley and others 1981 New Guinea 5,000 1962 All Japan 14 48 Bagshawe and others 1989 South India 210,337 1968 All Paris and 5-10 23 Noordeen 1986 Copenhagen Burma 28,220 1964 0-14 Glaxo 11-14 20 Lwinandothers 1985 Malawi 80,622 1979 0-15 Glaxo Less than 50 Fine and others 1986 5 years Table 12A-4. Case Management Interventions for Leprosy Level of Managernent Strategy Household Primary Secondary Tertiary Secondary prevention or treatment Diagnosis, medical Diagnosis, medical Diagnosis, medical Diagnosis, medical Treatment of complicationsa Diagnosis, medical Diagnosis, medical Diagnosis, medical, surgical Diagnosis, medical, surgical Reconstructive surgery n.a. n.a. n.a. Diagnosis, medical, surgical Psychotherapy and rehabilitation n.a. Diagnosis Diagnosis, medical Diagnosis, medical n.a. Not applicable. a. In many developing countries, the leprologists and plastic surgeonLs needed for treatment of serious complications are found only in tertiary facilities. Source Authors. 272 Myo Ther Hroon, Jeanne Berrolli, and Lies D. Kosasih Appendix 12B. Cost-Effectiveness Analysis and Kosasih) who had experience treating leprosy patients in developing countries. A cost-effectiveness analysis will be used to evaluate the costs of a leprosy control program. Such an analysis differs from a * Average age at onset: 29 years (taken from urban data in cost-benefit analysis in that input (cost of the control program) Yangon, Myanmar). is considered in monetary terms and output (benefits of the * Average life expectancy (at age of onset): 41.52 years (data control program) is considered in nonmonetary terms, whereas for Myanmar males, WHO 1985a). in a cost-benefit analysis both input and output are considered * Average case fatality: 0.001 (Walsh 1988). in monetary terms. The results of a cost-effectiveness analysis might be expressed as cost per death averted or per year of * Average age at death of those who die of the disease: 39 years healthy life gained rather than as a ratio of cost of the control (educated guess). program to savings resulting from preventing loss of productiv- * Proportion of disablement prior to death: 0.8 (educated ity. This approach is useful when the goal is to compare the guess); the proportionofdisablement isproportionate reduc- costs of alternative strategies to achieve certain health out- tion in functional ability. comes or to compare the expected effect of each dollar spent * Proportion permanently disabled: 0.16 (Myint, Htoon, and across different disease control programs, the latter being the Shwe 1990). goal here. A Supercalc microcomputer spreadsheet model developed * Proportion of disablement of permanently disabled: 0.5. by Ralph R. Frerichs of the Department of Epidemiology, * Temporarydisabilitydays:30days;temporarydisabilityfor School of Public Health, University of California at Los leprosy represents days during which the patient has skin Angeles, will be used to evaluate the cost-effectiveness of a lesions and suffers the accompanying psychosocial conse- leprosy control program. The model is based on earlier work quences but has no physical disability (educated guess). done in Ghana by the Ghana Health Assessment Project Proportion of disablement of temporarily disabled: 0.05; it is Team (1981). The model output is direct cost per year of assumed that during the period of temporary disability, the healthy life gained. No attempt is made n the model to patient does not have full productivity because of the psy- quantify indirect costs (or savings from preventing indirect chosocial consequences of the disease, but that his or her costs) in monetary terms. The cost per YHLG was calculated for productivity is reduced only slightly, that is, by 5 percent, two different components of a control program, that is, drug because there is no actual physical disability during this therapy and treatment of complications, first individually and then in combination. The parameters of the model are: period (educated guess). * Incidence: 3 per 1,000 per year in Myanmar (Burma, * Average age at onset (AO) Ministry of Health 1985). For each case of the disease, the * Average life expectancy at onset (L) model calculates: * Average age at death for those who die of the disease (AD) Days lost because of premature death (xl): * Case-fatality proportion (CF) * Proportion of disablement prior to death of those who die of the disease (DPD) Days lost because of disability before death (x2): * Proportion ofthose whodo not die ofthedisease but who are permanently disabled (CD) (2) x2= CF . (AD - AO) DPD .365 * Average level of disablement of the permanently dis- abled (CDP) Days lost because of chronic disability among those who do * Temporary days of disability (T) not die of the disease (x3): * Level of temporary disablement (TPD) (3) 3= CD- L CDPD 365 * Incidence (I) These parameters are entered first for the situation in which Days lost because of temporary disability among those who no intervention is undertaken. Then the proportionate do not die of the disease and are not permanently disabled changes in these parameters after intervention are entered into (x4): the model. From these changes, the model calculates years of healthy life gained from the intervention. It is then a case of (4) x4 = (1 - CF - CD) . T TPD putting in the cost of the intervention and dividing it by the years of healthy life gained. Total days of healthy life lost (DHLL) because of the disease Parameters were estimated as follows, assuming no interven- is, then, the sum of these four parameters: tion. The estimates come from data for Myanmar or represent educated guesses arrived at by consensus of two of us (Htoon DHLL = xl + x2 + x3 + x4 Leprosv 273 Total days of healthy life lost per 1,000 population is calcu- bacillary-to-paucibacillary ratio reach 50:50. The ratio will be lated by multiplying total days of healthy life lost by inci- even higher if the new case-finding activity is poor that dis- dence per 1,000 population. charged cases are not being replaced by an equivalent number Then, considering an intervention, the days of healthy life of new cases. The cost per year of healthy life gained from gained (DHLC,) per 1,000 population is calculated by the treatment is calculated to be $12.71 per YHLG. following formula: Details of the Calculations DHLG,/1,000 = DHLL/1,000 -i PI - (DHLL - I)HL()], Cost per case was first calculated separately for treatment of where Pi is the proportionate change in incidence. multibacillary and paucibacillary cases. The WHO treatment regimens and the calculation of costs per case (with use of the The model has been used to estimate the cost per YHLG for Essential Drugs Price List) were as follows: drug therapy and treatment of complications. Some specific limitations of the analysis presented below should be empha- * Paucibacillary cases: 600 milligrams of rifampicin once a sized. The state of the leprosy literature precludes accurate month for six months. Rifampicin comes in 300-milligram estimation of the various parameters used as inputs in the tablets, so twelve tablets of 300 milligrams are needed. model. Therefore, the numbers used represent "best estimates." No sensitivity analyses were done on these "before interven- 12 - $9.87/100 = $1.18. tion" parameters. Sensitivity analyses were done, however, for the effects of the three components ofthe intervention. These 100 milligrams of dapsone daily for six months. Approxi- analyses involve varying the value of uncertain parameters mately 190 doses are needed. within an expected range to see how sensitive the output (in this case cost per YHLG) is to the changes in a parameter. 190 .$5.10/1000 = $0.97. Sensitivity analysis is useful in that it provides some informa- tion about the degree to which the estimate of cost per YHLC, The cost per case of delivering the drugs is estimated in might he inaccurate as a result of our uncertainty about a the following way (from Myanmar, Ministry of Health parameter. 1985: the cost of delivery, including salaries and traveling allowances, is estimated at 4,664,000 kyats for 1985. At the exchange rate of 7 kyats to the U.S. dollar, this is Appendix 12C. Cost-Effectiveness of Multidrug equal to 4,664,000 divided by 7, or $666,286. The total Therapy number of cases treated in 1985 in Myanmar was 221,125. Thus the cost of delivety per case is $666,286 divided by The first analysis considers the costs and benefits of drug 221,125, or $3 per case per year. This figure includes the therapy alone. The drug regimens are those recommended by cost of supervision, but not of vehicles, because the latter wHoforrmultibacillary and paucibacillarycases (see the section are to be considered as part of the capital cost. As ende- on case management). Prices are taken from UNICEF's Essential micity decreases, the delivery cost is likely to go up. This Drugs Price List for July through December 1991. As shown $3 per case per year is calculated under the assumption of below, the cost of drugs to treat a paucibacillary case is $2.15. high endemicity and represents the cost for an average of It is estimated that the cost of delivery of the drugs and four visits by health care workers. Six visits are needed for supervision is an additional $5.85 (see below for calculation of paucibacillary cases, however, so the cost of delivery for deliverycost),bringingthetotalcostoftreatingacaseto$8.00. a paucibacillary case will be $3 - 6/4 = $4.50. The addi- The cost of drugs to treat a multibacillary case is $81.66. It is tional cost for supervision of drug delivery is estimated to estimated that the cost of delivery and supervision is an addi- be 30 percent of the delivery cost, or $ 1.35. tional $40.95, bringing the total cost of treatment to $122.61 Thus, the total cost of drug therapy for a paucibacillary per case. Considering that approximately 20 percent of cases case is equal to are multibacillary and 80 percent are paucibacillary, the aver- age cost per case is $30.92 (a weighted average). $1.18 + $0.97 + $4.50 + $1.35 = $8.00. This multibacillary-to-paucibacillary ratio of 20:80 is based on the 1988 Annual Report of the Leprosy Control Program, * Multibacillary cases: 600 milligrams of rifampicin once a Ministry of Health, Burma. The ratio was estimated for a month for three and one-half years (Dr. S. K. Noordeen, leprosy endemic area in which multidrug treatment is not in Chief Medical Officer, Leprosy, Division of Communicable use. Because paucibacillary cases are to be discharged after six Diseases, WHO, personal communication August 30, 1989). months of treatment, their number will be quickly reduced Rifampicin comes in 300-milligram tablets, so 84 tablets compared with the number of multibacillary cases, which (2 - 42) are needed. require an average of three and one-half years of treatment. Only after some years of multidrug treatment will the multi- 84 . $9.87/100 = $8.29. 274 Myo Thet Htoon, Jeanne BertoUi, and Lies D. Kosasih 300 milligrams of clofazimine once a month for three and be $31.50 ($3 .42/4). Thirty percent of the drug delivery one-half years. Clofazimine comes in 100-milligram cap- cost ($9.45) will be added to cover the cost of supervision sules, so 126 capsules (3 42) are needed. of drug delivery. So the total cost of drug therapy for a multibacillary case is: 126 $96.74/1000 = $12.19. $8.29 + $12.19 + $6.52 + $54.66 + $31.50 100 milligrams ofdapsone daily for three and one-half years, + $9.45 = $122.61. so 1,278 tablets (3.5 365) are needed. Note: This may be an overestimate of cost for some multi- 1,278 $5.10/1,000 = $6.52. bacillary cases because treatment may be needed for only two years. 50 milligrams of clofazimine daily for three and one-half years (excpt the foty-two das on whic aA30 * Given that approximately 80 percent of cases of leprosy years (except the forty-two (lays on which a 300- are paucibacillary cases, and 20 percent are multibacillary milligram clofazimine dose is taken), so 1,130 capsules cases, the average cost of treating a leprosy case can be (3.5 D 1365 - 421) are needed. A price is listed only for estimated by taking a weighted average: 100-milligram capsules, so this price will be halved for the calculation. (0.80 . $8.00) + (0.20 $122.61) = $30.92. 1,130 . ($96.74/2)/1000 = $54.66. In the first analysis, the following initial assumptions were The cost per case of delivering the drugs for a multibacillary made (based on best estimates): case is estimated as above for paucibacillary cases. If it costs * That treatment of multibacillary and paucibacillary pa- $3 per case for an average of four visits to deliver drugs in tients would reduce by 33 percent the proportion of patients Myanmar, and if forty-two visits are necessary for a multi- who do not die but are permanently disabled and would bacillary case, the delivery cost for a multibacillary case will reduce by 60 percent the level of disablement of these cases. Table 12C- 1. Cost-Effectiveness of Treatment Prograrns for Leprosy With interventions Parameter Baseline Drug therapy Complications Average case fatality 0.001 No change No change Annual incidence (new cases per l,,000 population) 3.00 No change No change Average age at disease onset (years) 29 No change No change Average life expectancy at disease onset (years) 41.52 No change No change Average age at death from leprosy (years) 39 No change No change Proportion disablement of fatalities 0.8 No change No change Proportion of survivors permanently disabled 0.16 0.67 No change Proportion disablement of permantly disabled 0.5 0.4 0.8 Temporarily disabled Days of temporary disability 30 0.67 0.8 Proportion disablement during period of temporary disability 0.05 No change 0.8 DHLL/DHLG per case From premature death 11.50 0.00 0.00 From disability before death 2.92 0.00 0.00 From permanent disability among survivors 1,212.38 887.47 242.48 From temporary disability among survivors 1.26 0.36 0.45 Total 1,228 888 243 Total per 1,000 population per year 3.684 2,663 729 Costs of intervention Per case (dollars) n.a. 30.92 225.00 Per YHLG (dollars) n.a. 12.71 338.06 n.a. Not applicable. a. Expected proportionate increase or decrease as result of multiple drug therapy and treatment of complications. b. For baseline, days of healthy life lost. For intervention, days of healthy life gained from reduction of listed parameter. c. Represents 98.72 percent of total DHLL per case. Source: Authors. Leprosy 275 * That treatment would also reduce the number of days of per YHLO is compared with the cost per YHLG in the situation in temporary disability by 33 percent. which all the maximum values are entered, there is little * Thattreatmentwouldnotaffectanyoftheotherparam- difference between the two outputs ($25.15 per YHLG as op- eters of the model. (Eventually treatment of patients would posed to $10.34 per YHLG). Finally, in recognition of the reduce the incidence by decreasing the numbers of patients possibility that costs may have been underestimated in the who are acting as sources of infection. This idea will be model, the cost per case was varied to determine the effect on explored in the sensitivity analysis.) the cost per YHLG. When the cost per case was doubled, the cost per YHLG for chemotherapy was $25.42, still a reasonable figure. In table 12C- I we summarize the cost-effectiveness of treat- Even when the cost per case was tripled, the cost per YHLG for ment according to the model. drug therapy was only $38.13. Results Appendix 12D. Second Analysis: Treatment The average cost oftreating a leprosy patient (regardless oftype Of Complications of leprosy) is $12.71 per YHLG. It must be mentioned that the cost of laboratory facilities needed for confirmation ofdiagno- In the second analysis we consider the costs and benefits of sis and for determination of bacteriologic status of patients at treating complications of leprosy, including treatment of reac- the end of the course of treatment is not included in this tions and ulcers, reconstructive surgery, and all forms of reha- calculation. Neither is the benefit of reduced incidence, which bilitation. Costs of treating the complications were estimated would come in time if the treatment program is maintained to he $225.00 per case by consensus of the two members of our (although this will be discussed below). group who had had experience treating leprosy patients in a developing country. It was further estimated that 10 percent Sensitivity Analysis of cases would have complications requiring treatment. The model calculates that it costs $338.06 per YHLG to treat the A sensitivity analysis was done to determine how much the complications of leprosy. This makes the cost of the combina- cost per YHLG would be affected by varying the effects of drug tion of drug therapy and treatment of complications $350.77 therapy on the proportion of patients who are permanently per YHLG ($12.71 per YHLG + $338.06 per YHLG), assuming there disabled, the proportion of disablement of the permanently is no interaction between drug therapy and treatment of com- disabled, the temporary disability days, and the incidence. The plications, which is a reasonable assumption. The cost of first three parameters were varied to allow for inaccuracy in psychotherapy is not included in this analysis. Further details "educated guesses" regarding the effect of multidrug therapy on are provided below. these parameters. The sensitivity analysis shows whether it matters that the estimates might be inaccurate. Variation in Details of the Calculations incidence is also explored because incidence will be reduced if a drug therapy program is continued for several years, and the * The cost of treating reactions with prednisolone was cost per YHLC will subsequently be decreased (depending on the estimated to be $3 per case on average. The estimate is discount rate). complicated by individual variation in clinical response. The sensitivity analysis involved the changes shown in table The duration of treatment may be as short as two to three 12C-2 (values were varied for one parameter at a time). The months or longer than two years. Prednisolone is such an model proved to be relatively insensitive to these changes. As inexpensive drug, however, this wide variation in duration might be expected, when the minimum values are entered for of treatment is not expected to be important. all the parameters (except incidence) and the resulting cost * The cost of treating ulcers was estimated at $20 per case on average. * The cost of reconstructive surgery was estimated at $200 Table 12C-2. Effects of Drug Therapy on Cost per case, and the cost of rehabilitation was estimated at $2 per YHLG per case. Reduction Cost per YHLG 0* Therefore, the total cost per case of treating complica- Parareter (percent) (dolslrs) tions of leprosy was estimated to be: Proportion premanentlydisabled 0.10-0.50 14.54-11.63 $3 + $20 + $200 + $2 = $225. Disablement of permanently disabled 0.30-0.80 17.52-10.75 Temporary disability days 0.10-0.50 12.71-12.71 In the second analysis, the followingassumptions were made Incidence 0.30-0.70 12.71-12.71 (input and output of the model are shown in table 12C-1): Note: Values wer-e varied for one parameter at a time. That treatment of complications reduces the disable- Source: Authors. ment of permanently disabled cases by 20 percent. 276 Myo Thet Htoon, Jeanne Bertolli, and Lies D. Kosasih * That it reduces the days of temporary disability by 20 Appendix 12E. Cost Issues in Screening and percent. Immunization * That it reduces the disablement during temporary dis- ability by 20 percent. The cost per YHLG of a screening program was not considered * That it does not affect any other parameters in the in this analysis because the model used was not designed for model. this application. Additional parameters such as coverage of the population and sensitivity and specificity of diagnosis are needed Results if the model is to reflect the effect of a screening program. The cost aspect of the analysis is complicated by the fact that screening The cost per YHLG for treatment of complications was calcu- without treatment has no effect on disease control. lated to be $338.06. The cost per YHLG for drug therapy plus In lieu of a cost per YHLLG measure of the cost-effectiveness treatment of complications was thus $350.77 per YHLG ($12.71 of a screening program, some of the cost issues involved with per YHLG + $338.06 per YHLG), assuming no interaction be- screening will be discussed briefly. It was estimated (by the tween the two components of the control program. two members of our group who had experience treating leprosy patients) that the cost per person of screening by Sensitivity Analysis clinical exam would be approximately $0.10 (based on costs in Myanmar). The total cost of screening would then be A sensitivity analysis was done to determine how much the calculated by multiplying the cost per person by the number cost per YHLG would be affected by varying the effects of of persons screened. Therefore, screening 1,000 people treatment of complications on the proportion of disablement would cost $100. The number of cases that would be de- of patients who are permanently disabled, the temporary dis- tected by screening (but that would have gone undetected ability days, and the proportion of disablement of those who without screening) would have to be considered. It was are temporarily disabled. These three parameters were varied further estimated that a screening and treatment program to allow for inaccuracy of educated guesses of these values. would decrease age at onset by 10 percent, that it would Again, the sensitivity analysis shows whether it matters that decrease the proportion of cases who do not die but are the estimates might be inaccurate. The sensitivity analysis permanently disabled by an additional 20 percent (over involved the changes shown in table 12D-1. treatment alone), and that it would decrease the proportion The model was sensitive to changes in proportion of disable- of disablement of those who are permanently disabled by an ment of patients permanently disabled, indicating that knowl- additional 20 percent. The screening and treatment pro- edge of the effect of treating complications is necessary to have gram would eventually decrease incidence because the cases confidence in the estimate of cost per YHLG given by the model. would be caught and treated earlier, thereby shortening the Changes in the temporary disability days and the disablement period during which persons with the disease are transmit- of the temporarily disabled have little effect on the output ting it. Because some of the benefits of screening and treat- because initial values of these parameters are so low that even ment are future benefits, the cost should be discounted. The a 50 percent change in the values is a small amount. When the cost of screening programs is relatively high because many minimum values are entered for all the parameters, including people who do not have symptoms of the disease must be incidence, and the resulting cost per YHI-G compared with the screened so that the few who do have it are found. Thus, a cost per YHLG in the situation in which all the maximum values, screening program must be very effective to justify the cost. including incidence, are entered, the same values are obtained The cost per YHLG of a vaccination program with BcG vac- ($676.05 per YHLG as opposed to $135.27 per YHLG). cine was not considered because the efficacy of BCG vaccine Again, because costs cited in this analysis may have been for leprosy prevention is so variable from area to area, and the underestimated, it may be useful to consider the output if the appropriate vaccination schedule is still a matter of debate. average costs of treating complications are doubled ($676.12 Also, this cost would be shared by leprosy and tuberculosis per YHLG) or even tripled ($1,014.18 per YHLG). control programs. The cost per dose of scc( vaccine is small (approximately $0.10 in developing countries, Walsh 1988), however, so prevention of even a small proportion of the ex- pected number of leprosy cases might justify the use ofthe vaccine Table 12D- 1. Effects of Treatment of Compicatuons for leprosy control, especially when the efficiency of immun iz- on Cost per YHLG ing for both tuberculosis and leprosy is considered. Of course, delivery costs would have to be included in such an analysis. Reduction Cost per YHLG Parameter (percentage) (dollars) Disablement of permanently disabled 0.10-0.50 674.86-135.38 Appendix 12F. Effect of Accuracy of Diagnosis Temporary disability days 0.10-0.50 338.20-337.64 Disablementoftemporarilydisabled 0.10-0.50 338.20-337.64 In the analysis of the cost-effectiveness of drug therapy Source: Authors. (appendix 12C), perfect accuracy of diagnosis was assumed. In Leprosy 277 this analysis, another spreadsheet program will be used to sensitivity and specificity equal to 50 percent, cost per identi- calculate cost per identified true case of the disease at varying fied true case of the disease increased 100-fold to $3,092.00. levels of sensitivity and specificity of diagnosis. Inputs for this spreadsheet include the following: Discussion 10 pre100 ence g1 hswoe a he These results emphasize the need for training of health workers 10 per 1,000 (WHO 1985a) for accurate diagnosis. Unfortunately, current levels of sensi- * Sensitivity and specificity of diagnosis: to be varied. tivity and specificity of diagnosis of leprosy are unknown. This * Cost per YHLG of multidrug therapy: $30.92 as calculated knowledge is important for estimating costs more precisely. in previous model Outputs (table 1 2F- 1) include the following: Appendix 12G. General Comments about * Prevalence of disease as diagnosed by health worker Cost-Effectiveness Modeling * Predictive value of health worker diagnosis * Cost per identified true case of disease It is important to note that cost per YHLG calculated by the model does not include the cost of education of the community Results about leprosy or of training health workers. As discussed ear- lier, education of the community is essential to the success of The specificity drives the cost. Even at the relatively high a control program. The costs of educating people about the levels of 90 percent sensitivity and specificity, cost per identi- disease are very difficult to estimate, however, and because fied true case of disease increases steeply to $371.04, as com- they would probably be incorporated into other health educa- pared with $30.92 for perfect accuracy of diagnosis. With tion efforts, they would fall under the budget of multisectoral Table 12F- 1. Effect of Sensitivity and Specificity of Diagnosis on Cost per Case of Treated Leprosy Data entered by analyst Data derived by computer Cost per case Prevalence of True prevalence among Sensitivity of Specificity of diagnosed by disease diagnosed by Predictive value of Cost per identified persons visitng health health worker's health worker's health worker health worker health worker's crue case of disease worker (per 1,000) diagnosis diagnosis (dollars) (per 1,000) diagnosis (dollars) 10.00 0.99 0.99 30.92 9.91 0.500 61.84 10.00 0.90 0.90 30.92 99.01 0.083 371.04 10.00 0.90 0.80 30.92 198.01 0.043 711.16 10.00 0.90 0.70 30.92 297.01 0.029 1,051.28 10.00 0.90 0.60 30.92 396.01 0.022 1,391.40 10.00 0.90 0.50 30.92 495.01 0.018 1,731.52 10.00 0.80 0.90 30.92 99.01 0.075 413.56 10.00 0.80 0.80 30.92 198.01 0.039 796.19 10.00 0.80 0.70 30.92 297.01 0.026 1,178.83 10.00 0.80 0.60 30.92 396.01 0.020 1,561.46 10.00 0.80 0.50 30.92 495.01 0.016 1,944.10 10.00 0.70 0.90 30.92 99.01 0.066 468.22 10.00 0.70 0.80 30.92 198.01 0.034 905.51 10.00 0.70 0.70 30.92 297.01 0.023 1,342.81 10.00 0.70 0.60 30.92 396.01 0.017 1,780.11 10.00 0.70 0.50 30.92 495.01 0.014 2,217.41 10.00 0.60 0.90 30.92 99.01 0.057 541.10 10.00 0.60 0.80 30.92 198.01 0.029 1,051.28 10.00 0.60 0.70 30.92 297.01 0.020 1,561.46 10.00 0.60 0.60 30.92 396.01 0.015 2,071.64 10.00 0.60 0.50 30.92 495.01 0.012 2,581.02 10.00 0.50 0.90 30.92 99.01 0.048 643.14 10.00 0.50 0.80 30.92 198.01 0.025 1,255.35 10.00 0.50 0.70 30.92 297.01 0.017 1,867.57 10.00 0.50 0.60 30.92 396.01 0.012 2,479.78 10.00 0.50 0.50 30.92 495.01 0.010 3,092.00 Source: Authors. 278 Myo Thet Htoon, Jeanne Bertolli, and Lies D. Kosasih health education programs rather than under that of a leprosy The annualized cost, a(r,n) c, is then: control program. As illustrated in appendix 12F, money spent training health care workers is money saved through reducing a(r,n) c = 0.1172 ($100,000) the number of false-positive cases diagnosed. We did not = $11,720.00. attempt to estimate the cost of this training. It is also quite important to recognize that in the cost In a population of 100,000 in which the prevalence of analysis presented in this chapter we have not considered the leprosy is 10 per 1,000, one thousand patients would be value of intangible benefits of a leprosy control program. under treatment. So the annualized capital cost per pa- Failure to consider such intangible benefits as prevention of tient is: psychological trauma when weighing the cost-effectiveness of a control program leads to serious underestimation of the $11,720 $11.72 per case. benefits of such a program. This point was emphasized by 1,000 Creese and Henderson (1980 p. 494): "Health programs char- acteristically have effects which, though important, are ex- Cost of Case Detection tremely difficult to measure and to value. Benefits such as reduced anxiety, pain, or discomfort are typical examples: The cost of screening for case detection will be based on the these are desirable 'outputs' of the health system, but they are same hypothetical population of 100,000 assumed above. Fur- not readily comparable with other outputs such as increased thermore, it will be assumed that no screening has been done productivity." in the population in the previous five years. The screening team would consist of five workers, and sixty days per year for five years would be devoted to screening. It is estimated that Appendix 12H. Total Cost of a Leprosy Control such a team could screen 200 persons per day. Typical salaries Program for each such workers in a developing country would be $5 per day, with an allowance of $2 per day for travel (based on costs In the previous appendixes, the cost-effectiveness of various in Myanmar). New cases would be detected by the screening aspects of a leprosy control program was discussed. In those program workers at the rate of 3 cases per 1,000 screened. If analyses, we were considering a one-year program to treat new the community was fairly isolated, after five years of the cases (assuming existing cases were already under treatment). screening program, fewer cases would be detected because the We did not include capital costs and the cost of case detection, screening and early treatment would lower incidence (the primarily because the model we used was not designed to incubation period is five to seven years). We estimate that include such costs and redesigning the model would have been approximately half the cases would be discovered when they a complicated undertaking beyond the scope of this project. In presented for treatment on their own initiative, rather than the following analysis, we will discuss the capital costs and the through the screening effort. cost of case detection in the context of a five-year leprosy The total cost of the screening program per year would be control program. We will then add these costs to those calcu- as follows: lated previously for chemotherapy and complications to give an estimated total cost for a leprosy control program. Total cost = (salaries + travel allowance) x no. workers = 1(60 . $5.00) + (60. $2.00)]. 5 = $2,100.00. Capitol Costs The total number of persons screened would be equal to 60 The calculation of annualized capital cost will be based on a days x 200/day, or 12,000 persons. The estimated number of hypothetical population of 100,000 in a hyperendemic area new cases detected would be equal to the screened population with leprosy prevalence of 10 per 1,000 population. The cap- multiplied by the number of cases detected through screening, ital cost will include cost of vehicles, buildings, office equip- which we estimate to be 3 per 1,000, so 36 cases (12,000 . 3 ment, and laboratory equipment. The total capital cost (c) was per 1,000) would be detected by screening. Therefore, the estimated to be $100,000. The lifetime of equipment and screening cost per new case detected would be $2,100.00 per vehicles was estimated at 10 years. The social rate of discount 36 new cases detected, or $58.34 per new case detected. An (r) was taken as 3 percent. The annualized discounting factor additional 36 cases would present on their own initiative for was calculated as follows: treatment. Before the screening program was started, there were 1,000 registered cases in the total population of 100,000 a(r,n) = [r (1I + r)" ] (prevalence, 10 per 1,000). Together, then, there are 1,000 [(I + r)n - 1 ] registered cases plus 36 cases detected by screening plus 36 = .03 (I + .03)10 cases detected when they presented on their own initiative, or - 10 a total of 1,072 cases in the population of 100,000. The total (I +.03) - I cost of screeningper case is, then, $2,100.00 per 1,072, or $1.96 = 0.1172. per case. Leprosy 279 Total Cost per Case of a Leprosy Control Program Bloom. B. R. 1990. "An Ordinarv Mortal's Guide to the Molecular Biology of Mycobacteria." InternationalJoumnal of Leprosy and Other Mycobacterial Dis- eases 58:365-75. To calculate the total cost per case of a leprosy control pro- Bloom gram, the costs of basic chemotherapy as well as treatment of for B. R., and T. Godal. 1983. "Selective Primary Health Care: Strategies for Control of Disease in the Developing World. 5. Leprosy." Reviews of complications must be added to the capital and screening costs. Infectious Diseases 5(4):765-80. Boerrigter. G.,J. M. Ponnighaus. and P. E. Fine. 1988. "Preliminary Appraisal COST OF BASIC CHEMOTHERAPY. As discussed in appendix of a WHO-Recommended Multiple Drug Regimen in Paucibacillary Leprosy 12C, the cost of treatment of uncomplicated leprosy is esti- Patients in Malawi." lnternational]oumal of Leprosy and Other Mycobacterial mated at $30.92 per case, if it is assumed that 80 percent of Diseases 56:408-17. cases are paucibacillary and 20 percent are multibacillary. Browne, S. G. 1985. "The History of Leprosy." In R. C. Hastings, ed., Leprosy. Edinburgh: Churchill Livingstone. Cellona, R. V., T. T. Fajardo, Jr., D. 1. Kim, Y. M. Hah, T. Ramasoota, S. COST OF TREATMENT OF COMPLICATIONS. As discussed in ap- Sampattavanrch, M. P. Carrillo, R. M. Abalos. E. C. dela Cruz, and T. Ito. pendix D, the cost of treatment of complications for each of 1990. "Joint Chemotherapy Trials in Lepromatous Leprosy Conducted in those who need such treatment is estimated at $225.00. Only Thailand, the Philippines, and Korea." IntemationalJournal of Leprosy and 10 percent of all patients are expected to need treatment for Other Mycobacterial Diseases 58:1- 1. complications. Therefore, the cost of treating complications Convit, Jacinto, N. Aranzazu, M. Ulrich. M. E. Pinardi, 0. Reyes, and J. averaged over all cases, calculated by dividing the total cost for Alvarado. 198'. 'Immunotherapy with a Mixttire of Mycobactenum leprae and BCG in Different Forms of Leprosy and in Mitsuda-Negative Contacts." treating complications (1,000 0.1 $225.00) by the total lnternanonal journal of Leprosy and Other Mycobacterial Diseases 50:415-24. number of registered cases (1,000), is $22.50 per case. Courtwright, P., and G. Johnson. 1988. Blindness Prevennton in Leprosy. Lon- don: Intemartional Center for Eye Health. TOTAL COST PER CASE. The sum of the estimated costs of Creese, A. L., and R. H. Henderson. 1980. "Cost-Benefit Analysis and capital, screening, chemotherapy, and treatment of complica- Immunization Programs in Developing Countries." Bulletin of the World tions is used to estimate the total cost per case in the leprosy Health Organizanon 58:491-97. control program. Declercq E. and C. Gelin. 1991. "Global Evaluation of the Introduction of Multidrug Therapy." Leprosy Epidernrological Bulletin 6. WHO Collaborating Centre for the Epidemiology of Leprosy, Catholic University of Louvain, Annualized capital cost $11.72 per case Brussels, Belgium. Screening 1.96 per case Dharmendra. 1986. "Epidemiology of Leprosy in Relation to Control." Indian Chemotherapy 30.92 per case Joumal of Leprosy 58(l):l-16. Complications 22.50 per case Dominguez, V. M., P. G. Garbajosa. M. M. Gyi, C. T. Tamondong, T. Total $67.10 per case Stindaresan, L. M. Bechelli, K. Lwin, H. Sansarricq, J. Walter, and F. M. Noussitou. 1980. "Epidemiological Information on Leprosy in the Singu Area of Upper Burma." Bulletin of the World Health Organization 58: Discussion 81-89. Ellard, G. A., P. T. Gammon, H. S. Helmy, and R. J. W. Rees. 1974. "Urine If the screening and treatment are successful, in five to seven Tests ro Monitor the Self-Administration of Dapsosne by Leprosy Patients." years (roughly the length of the incubation period of leprosy), Amnencan Journal of Tropical Medicine and Hygiene 23(3):464-70. the number of cases detected will begin to decline as the rate Fine, P. E., J. M. Ponninghaus, N. Maine, J. A. Clarkson, and L. Bliss. 1986. of transmission is reduced. As this happens, the cost per case "Protective Efficacy of BCG against Leprosy in Northemn Malawi." Lancet detected will increase. 1:499-502. (new series). Ganapati. R., C. R. Revankar, and R. R. Pai. 1987. "Three-Year Assessment of Multi-drug Therapy in Multibacillary Leprosy Cases." Indian Journal of Leprosy 59(1):44-49. Notes Ghana Health Assessment Project Team. 1981. "A Quantitative Method of Assessing the Health Impact of Different Diseases in Less Developed The authors gratefully acknowledge Drs. S. K. Noordeen, Emmanuel Max. Countries." 1l7ternanoialJournal of Epidemiology 10:73-80. W. Felton Ross, Richard Morrow, and Paul E. NI. Fine for their comments, Gillis, T. P., and D. L. Williams. 1991. "Polymerase Chain Reaction and which guided the development of this chapter. Leprosy." International Journal of Leprosy and Other Mycobacterial Diseases 59:311- 16. Hagan. K. J.. and S. E. Smith. 1979. "The Reliability of Self-Administration References of Dapsone by Leprosy Patients in Burma." Leprosy Review' 50(3):201-1 1. [rgens, L. M., and R. Skjaerven. 1985. "Secular Trends in Age at Onset, Sex Bagshawe, A., G. C. Scott, D. A. Russell, S. C. Wigley, A. Merianos. and G. Ratio, and Type Index in Leprosy Observed during Declining Incidence Berry. 1989. "BCG Vaccination in Leprosy: Final Results of the Trial in Rates." AmencanJournal ofEpidemiology 122:695-705. Karimui, Papua New Guinea, 1963-79." Bulletin of the World Health Orga- Jacobson, R. R. 1985. "Treatment." In R. C. Hastings, ed., Leprosy. Edinburgh: nization 67:389-99. Churchill Livingstone. Bharadwaj, V. P., and K. Katoch. 1990. "An Overview of the Current Status Jesudasan, K., M. Christian, and D. Bradley. 1984. "Relapse Rate among of Serological Techniques in the Epidemiology of Leprosy." Tropical Medi- Non-lepromatous Patients Released from Control." Intemrational Journal of cine and Parasitology 41:359-60. Leprosy and Other Mycobacterial Diseases 52:304- 1 0. 280 M-vo Thet Htoon, Jeanne Bertolli, andi Lies D. Kosasih Katoch, V. M., C. T. Shivannavar, K. Katoch, G. V. Kanaujia, and V. P. Pearson. J. M., G. S. Haile, and R. J. Rees. 1977. "Primary Dapsone Resistant Bharadwaj. 1990. "Towards Clinical and Epidemiological Application of Leprosy." Leprosy Review48(2):129-32. Advances in Molecular Biology of Mycobactenumleprae "TropicalMedicine Ramanan, R., P. R. Manghani, A. Ghorpode, and S. K. Bhagolinal. 1987. and Parasitology 41:299-300. "Follow-up Study of Paucibacillary Leprosy on Multidrug Regimen." Indian Lechat, M. F. 1981. "TheTorments and Blessings of the Leprosy Epidemiomet- Journal of Leprosy 59(1):50-53. ric Mlodel." Leprosy Review 52(supplement I): 187-96. Ridley, D. S. 1974. "Histological Classification and the Immunological Spec- 1985. "Control Programs in Leprosy." In R. C. Hastings, ed., Leprosy. trum of Leprosy." Bulletin of the World Health Organization 51:451465. Edinburgh: Churchill Livingstone. . 1977. Skin Biopsy in Leprosy. Basel: Ciba Geigy. Lechat, M. F., C. Vellut, C. B. Misson, and J. Y. Masson. 1978. "Application Stanley, S. J., C. Howland, M. M. Stone, and 1. Sutherland. 1981. "BCG of an Econ(omic Model to the Study of Leprosy Control Costs." International Vaccination of Children against Leprosy in Uganda: Final Results."Journal Journal of Leprosy and Other Mvcobacterial Diseases 46:14-24. of Hygiene 87:233-48. Lwin, Kvaw, T. Sundaresan, NI. M. Gyi, L. MI. Bechelli, C. Tamondong, P. G Talwar, G. P., S. A. Zaheer, R. Mukherjee, R. Walia, R. S. Mlisra, A. K. Garbalosa, H. Sansarricq, and S. K. Noordeen. 1985. "BCG Vaccination of Sharma, H. K. Kar. A. Mukherjee, S. K. Parida, and N. R. Suresh. 1990. Children against Leprosv. Fourteeni-year Findings of the Trial in Bumsa." "Immunotherapeutic Effects of a Vaccine Based on a Saprophytic Cultiva- Bulletin of the World Health OrgaiiZation 63:1069-78. ble Mycobacterium, Mycobacterium wt in Multibacillary Leprosy Patients." Max, Emmanuel, and D. S. Shepherd. 1989. "Productivity Loss due to Defor- Vaccise 8:121-29. mity from Leprosy in India." International Journal of Leprosy and Other UNICEF (United Nations Children's Fund). 1991. Essential Drugs Pnce List. Mycobactenal Diseases 57:476-b2. July-December, 1991. Copenhagen: Procurement and Assembly Centre. Mittal, B. N. 1991. "The National Leprosy Control Programme in India." Walsh, G. P., E. E. Storrs, W. Meyers, and C. H. Binford.1977. "Naturally World Health Statstics Quarterly 44(1) :23-29. Acquired Leprosy Like Disease in the Nine-Banded Armadillo." Journal of Muliyil, Jayaprakash, K. E. Nelson, and E. L. Diamond. 1991. "Effect of BCG the Renculoendothelial Societ-y 22:363467. on the Risk of Leprosy in an Endemic Area: A Case Control Study." Walsh, J. A. 1988. Establishing Health Care Prorites in the Developig World. Internationaljournal of Leprosy and Other Mycobacterial Diseases 59:229-36. Bosron: .Adams Pu blishing Group. Myanmar, Ministry of Health. 1985. Annual Report of the Leprosy Control Wardekar, R. \ 1968. "Sulphone Treatment and Deformity in Leprosy." Program . Yangon. Leprosy in India 40:161-71. Myint, Tin. NI. T. Htoon, and T. Shwe. 1992. "Estimation of Leprosy Preva- Watson, JIM 1988. PreventngDisa*iiitNsinLeprosN Patents. London: Leprosy lence in Bago and Kawa Townships Using Two-Stage Probability Propor- Mission Intemational. tionate to Size Sampling Techniqule." International Journal of Epidemiology 21:778-83. WHO (World Health Organization). 1977. Fifth Report of the World Health Neik, S. S., N. D. Bhange, K. V. Sawant, anid R. Ganapati. 1988. "A Organization Expert Committee on Leprosy. Technical Report 607. Geneva. Bacteriological Assessment of Multibacillary Cases in Leprosy Colonies . 1980. A Guide to Leprosy Control. Geneva. after 4-1/' Years of Multidrug Therapy." lIsdian Journal of Leprosy 60(3): . 1982. Chemotherapy of Leprosy Control Programmes. Technical Report 393-99. 675. Geneva. Noordeen.S K. 1986. -B(O' Vaccination in Leprosy." DevelopmentsinBiological . 1985a. wHo Demographic Yearbook. Geneva. StandardiZation. 58(AA !:287-92. . 1985b. Epidemiology of Leprosy in Relation to Control. Technical Report . 1985. "The Epidemiology of Leprosy." In R. C. Hastings, ed., Leprosy. 716. Geneva. Edinhurghi: Churchill Livingstone. . 1988a. Sixth Report of the World Health Organization Expert Committee - 1992. Weekly Epidemiological Record 67:15 3-60. on Leprosy. Technical Report 768. Geneva. Noordeen, S. K., L. L. Bravo, arid D. Daumerie. 1991. "Global Review of WHO (World Health Organization), Regional Office of the Westem Pacific. Nlultidrug Therapv (MDT) in Leprosy." World Health Statistics Quarterly 1988b. "Final Report of the Regional Working Group on Drug Policy and 44(1):2-15. Operational Research in Leprosy Programs." icP/Beu/005. Manila. Typescript. 13 Malaria Jose A. Najera, Bernhard H. Liese, and Jeffrey Hammer Malaria is a collective name for different diseases that may malarias (benign, if from P. vivax and P. ovale; malignant, if result from infection by any parasites of the genus Plasmodium. from P. falciparum) and seventy-two hours for the quartan Four species of malaria parasites naturally infect humans: Plas- (from P. malariae). This development takes place in the pe- modium falciparum, P. vivax, P. malariae, and P. ovale. The ripheral blood for P. vivax, P. malariae, and P. ovale. But with characteristics of the disease vary with the intensity of the P. falciparum, only red blood cells infected with very young infection, the host's level of immunity, the adequacy of and parasites (called ring forms) are found in the peripheral blood; opportunity for treatment, and the parasite's susceptibility to it. those infected with developing or dividing parasites are seques- Transmission between humans occurs through the bites of tered in the capillaries of such intemal organs as the brain and certain species of mosquitoes of the genus Anopheles. In this cause the severe manifestations typical of P. falciparum. cycle, the parasite matures and reproduces sexually in the Some parasites do not follow the cycle of asexual reproduc- anopheline mosquito (the vector), which is therefore, strictly tion just described. Instead, they differentiate into male and speaking, the parasite's definitive host, and human beings are female gametocytes, which are eventually taken up by an its intermediate host. Anopheles mosquito. In the mosquito they can mature, achieve fertilization, and multiply in the stomach wall, producing A Natural History: Parasite and Vector about 1,000 sporozoites, which burst into the mosquito's body cavity and finally invade the salivary glands. This sexual cycle The life cycle of the parasite follows a general pattern. The takes between nine and thirty days or more, depending on the infecting parasite, an actively motile form called a sporozoite, temperature and the parasite species. is inoculated into the blood with the saliva of the biting Not all species of anophelines are vectors of malaria, and mosquito. After about half an hour, the sporozoites invade liver those that are vary greatly in their ability to transmit the tissue cells, where they develop and multiply. Small parasite disease. General or specific refractoriness may be due to many forms called merozoites, capable of invading the red blood causes, for example, because the plasmodia is unable to develop cells, burst into the blood-as many as 20,000 per successful or to invade the salivary glands, or because the mosquito sporozoite. The time needed to multiply in the liver, the cannot live long enough to complete the parasite's extrinsic pre-erythrocytic stage, varies with the parasite species: six to cycle, or because the mosquito has so little contact with seven days for P. faLciparum, fourteen to sixteen days for P. humans (for example, is so unlikely to bite humans) that it is malariae, and seven to eight days for P. vivax, although some unlikely to bite a human after becoming infected. Of the P. vivax parasites remain dormant in the liver for months, even roughly 400 species of Anopheles, only about 60 are vectors of a few years, in a form called hypnozoite. Once the parasites malaria under natural conditions; some 30 of these are of prime invade the red blood cells they initiate the cycle of develop- importance. ment and multiplication that causes clinical manifestations of As for all mosquitoes, the habitat of the immature Anopheles the disease. Disease symptoms are caused only by parasites in is water. Eggs are all laid on or on the edge of water and hatch the blood. The late development of hypnozoites, therefore, in two to three days to produce larvae (wrigglers). Larvae gives the disease a long incubation period, or a pattem of cure develop through five aquatic stages-four larval and one alternating with repeated relapses, because common antima- pupal-to produce adult flying mosquitoes. Only the female larial drugs that may clear the blood of parasites are not mosquito bites; it does so because it needs blood for its eggs to effective against parasites in the liver. Merozoites invade red mature. The male feeds on vegetable juices. Mating occurs blood cells, where they grow and multiply to produce eight to only once, soon after the adult female emerges. The female twenty-four merozoites (depending on the parasite species), stores the spermatozoa in a deposit called a spermatheca. The which rapidly invade new red blood cells. This development aquatic stages commonly last seven to twenty days, depending is accomplished in forty-eight hours for the so-called tertian on the temperature. The adult female may live from a few days 281 282 Jose A. Ndjera, Bernhard H. Liese, and JeffTey Hammer to well over a month, going through several cycles of blood true relapses may occur, because latent hypnozoites will mature feeds and egg laying (some 100 to 200 per batch), every two to in the liver and invade the blood after other parasites have four days. Survival and egg development depend mainly on been completely eliminated from it. Without reinfection, P. temperature and relative humidity. In extreme climates mos- tvivax may persist for three to four years. quitoes may go into hibernation, which allows some of them In the absence of other complicating factors, acute severity to survive the winter in temperate climates. and mortality occur almost exclusively in P. falciparum infec- Depending on the species, larval habitats vary enormously, tions. This parasite causes the surface of infected red blood cells reflecting mosquitoes' evolutionary adaptability. The habitats to become adhesive and to be sequestered in the capillaries of range from permanent to transient bodies of water; from fresh internal organs, leading to the pathological changes responsi- to brackish water; from standing water to flowing canals and ble for cerebral malaria and the serious renal, hepatic, and open streams; from water in open sun to that in deep shade; gastrointestinaldysfunctions. Otherseverecomplications such from shallow pools to deep wells; from clean drinking water to as shock, pulmonary edema, severe anemia and hemoglobin- water highly polluted with organic matter; from large open uria (or blackwater fever), are the results of more complex marshes to the tiny pools of water that collect between the mechanisms. leaves of bromeliads in plant axils, trees, rocks, crab holes, P. falciparum malaria can lead rapidly to death, so it is cattle footprints, or discarded artificial containers. But the important to recognize signs of severity early and refer the characteristics of breeding places are rather narrowly defined patient immediately for medical care. These signs include for each species, so larval habitats can be modified for control shock, anemia, convulsions, jaundice, hyperpyrexia, renal fail- of mosquito species. ure, impaired consciousness, spontaneous bleeding, macro- The seasonal availability of breeding places and the great scopic hemoglobinuria, and pulmonary edema or respiratory influence of weather conditions on mosquito activity and distress. Health services that suspect severe malaria should survival are largely responsible for the marked seasonality in treat it as a medical emergency, providing immediate treat- mosquito population densities and malaria transmission in ment and, whenever possible, laboratory monitoring of such most areas outside of permanently humid tropical areas. signs of severity as hypoglycemia, parasite density, and an Specific behavioral characteristics of mosquitoes may also imbalance of fluids and electrolytes. affect their vectorial ability. Mosquitoes' preferences for feed- The risk of severe malaria is almost exclusively limited to ing on humans or animals and their frequency of feeding are those who are not immune. In highly endemic areas this risk important determinants of the probability of their transmitting affects children older than three to six months, who have lost malaria. Human dwellings and domestic animal shelters-par- the immunity transferred from their mother, up to the age of ticularly those with thatched roofs, dark corners, and many about five years, when surviving children have developed their cracks in the walls-are good resting places in which mosqui- own immunity. African health authorities report that in the toes can digest the blood they have consumed while their eggs last few years cerebral malaria is being seen increasingly often mature. Such buildings favor mosquito survival but are also in older children and young adults. It has been suggested that vulnerable to insecticidal spraying. this may be the result of urbanization and personal protection, which reduce the risk of infection and delay the development Malaria as a Disease of immunity. Severity in adults is seen in areas of low endemic- ity, where people may reach adulthood without immunity. The chief symptom of malaria is fever, periodic bouts of which Equally at risk are immigrants and travelers from nonendemic tend to alternate with days of less or no fever. The classical areas-particularly laborers, who are often concentrated in paroxysm of fever lasts eight to twelve hours, typically in three camps, where nonimmunes and the infected live side by side stages: cold shivering rigor, buming dry skin, and drenching in overcrowded conditions where the risk of transmission is sweat that lowers the temperature. This pattern is more typical high. Also at risk are pregnant women, possibly because natu- of P. vivax (tertian periodicity) and P. mialariae (quartan) than ral immunity is depressed during pregnancy. of P. falciparum, which typically involves prostrating fever, Most deaths from malaria occur in young children living in with brief and incomplete remissions, more often irregular highly endemic areas of tropical Africa and the westem Pacific than clearly periodic. Untreated, the acute attack is shorter islands. The most common causes of death are cerebral malaria than that of P. vivax; in fatal cases, death often occurs in two and severe anemia. Malaria may also contribute seriously to to three weeks and sometimes as soon as two to three days after the severity of other childhood diseases. the onset of symptoms. Repeated infections give rise to an In pregnancy, P. fakLiparum malaria in the nonimmune may imnmune response in the host which eventually controls the lead to death, abortion, premature delivery, or low birth infection and the disease. Untreated or incompletely treated weight. In the semi-immune inhabitants of highly endemic infections will produce several recrudescences, after long areas, malaria represents a serious risk in a first and second symptomless periods, from parasites surviving in the blood. By pregnancy. Pregnant women are more easily infected (and are this mechanism alone, P. fakiparum may persist for one or two susceptible to anemia, hypoglycemia, and other complica- years, whereas P. malariae has been reported to recrudesce up tions) because the placenta is a preferential site for parasite to fifty-two years after last exposure to infection. With P. vivax, development. Malaria is an important cause of low birth Malaria 283 weight and high neonatal mortality in first- and second-born 1992) indicatesthattheworldpopulation(about5,300million children in endemic areas. It has been suggested that the build people) can be classified according to people's experience with up of a total uterine immune response may account for the malaria and their place of residence as follows: disappearance of these effects on subsequent pregnancies a Areas in which malaria never existed or disappeared (McGregor 1982). without specific antimalarial interventions: 1,431 million The Public Health Significance of Malaria people, or 27 percent of the world's population. * Areas in which endemic malaria disappeared after a Roughly 110 million clinical cases of malaria develop annu- specific campaign to control it was implemented and the ally. Some 270 million people are infected, carrying malaria area has remained malaria-free: 1,696 million people, or 32 parasites, although not necessarily developing symptoms. In- percent of the world's population. digenous malaria still exists in some 100 countries or areas. * Areas in which endemic malaria was reduced or even Accurate estimates are impossible because the accuracy of eliminated after control measures were implemented, but reporting varies considerably. Reporting from tropical Af- the disease was reinstated and the situation is unstable or rica-where more than 80 percent of the clinical cases and 90 deteriorating: 1,700 million people, or 32 percent of the percent of the parasite carriers may be found-is especially world's population. This category includes zones (which irregular and fragmentary. Reported cases are believed to rep- include about I percent of the world's population) in which resent about 2 to 20 percent of the actual cases. the most severe resurgence of malaria has recently devel- oped as a result of significant ecological or social changes, Geographical Distribution such as sociopolitical unrest and agricultural or other exploi- tation of jungle areas. Every year, in the World Health Statistcs Quarterly, the World * Areas in which endemic malaria remains basically un- Health Organization (wHo) publishes an overview of the world changed and no national antimalaria program was ever malaria situation (map 13-1). The overview for 1990 (WHo implemented, because of the enormous difficulties of Map 13- 1. Malaria Incidence VI~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~I' 0 0~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~_ 55 other countries EstImated number of cdncal cases 3 Masn areas where malaria transm,ss,on occurs Cases reported Source World Health Statistcs Annual. 1990 Note: The designations employed and the presentation of material on this map do not imply the expression of any opinion whatsoever on the part of the World Health Organization conceming the legal status of any country, territory, city, or area or of its authorities, or conceming the delimitation of its fron- tiers or boundaries. Source: Reproduced by permission of the World Health Organization, Geneva from W/eekly Epidemiological Record 67 (22-23): 162-167/169-174 (1992). 284 Jose A. 1Ndjera, Bernhard H. Liese, and Jeffrey Hammer achieving control: 500 million people, or 9 percent of the a national malaria control program until it had developed a world's population. Malaria is most endemic in these areas, health infrastructure that has been considered a precursor of which contain 85 percent or more of the malaria cases in primary health care; the pattern of malaria incidence is similar the world. These areas are mainly in tropical Africa; in some to the pattem in countries that eradicated malaria (map 13-1 ). of them-including forested and medium-altitude areas- If India and China are not taken into account, the incidence pilot projects were reportedly successful in interrupting ma- of malaria in the world did not show a clear trend until the late laria transmission, but in low savanna areas, particularly in 1970s, when it started a slow but steady deterioration. Data the Sahel, no pilot projects ever reported full success. from India indicate that, after recovery from the 1976 epi- demic, improvement slowed down and the situation seems to Trends be stagnating. The general pattem described here masks great local differ- The evolution of the malaria problem is traditionally described ences, not only in the intensity of the problem, but also in the by the number of registered cases reported to wIXHo by member pattem of its evolution over time. The geographical distribu- states. Figure 13-1, which excludes information on Africa tion of malaria is far from uniform; it can be seen that malaria because of inadequate, irregular reporting from that continent, clearly thrives in certain areas, and it may be said that it shows the effect of the massive resurgence of malaria transmis- occupies definable socioecological niches. The limits of ma- sion in India in 1976 and its subsequent control. Changes in laria foci are much more diffused than those for contagious China are shown separately, because the Chinese started offic- diseases such as smallpox, however, so it is more difficult to ially reporting to WHo only in 1977. China did not implement target their control. Figure 13-1. Number of Malaria Cases Reported, 1964-89 Number of Malaria cases reported (millions) 10 World (excluding African Region and China) 8 6 4 2 - _ - , - . * - - - India ,. World (excluding African Region, China and India) 0 China ' 1964 1966 1968 1970 1972 1974 1976 1978 1980 1982 1984 1986 1988 Source: WHO 1991. Malaria 285 Morbidity in Tropical Africa Smaller and more localized malaria epidemics have occurred when colonization efforts or agricultural or other economic In the last decade African countries south of the Sahara have development projects in endemic areas have attracted nonim- reported between 2 and 20 million cases a vear to WHO; but mune populations from nonmalarious areas. This happened in extrapolating from fever and parasite surveys, it is estimated the late 1950s, for example, when the lowlands of Kigezi that 100 million clinical malaria cases may occur every year, (Uganda) began to be colonized by people from the overpop- and 275 million persons may carry the malaria parasite. The ulated highlands, resulting in a tragic malaria epidemic with levels of endemic malaria are among the highest in the world. extremely high mortality. This led to the establishment in Extensive forest or savanna areas up to about 1,000 meters 1959 of one of the few successful malaria eradication pilot high, with rainfall of more than 2,000 millimeters a year, are projects in Africa (de Zulueta and others 1961). In the last few classified as holoendemic. Areas between 1,000 and 1,500 years malaria endemicity has reportedly spread in the high- meters high or lowland areas with 1,000 to 2,000 millimeters lands of Amani in Tanzania. This increased transmission has per year of rainfall are characterized as hyperendemic. As been attributed to the active colonization of those areas and altitude increases above 1,500 meters or rainfall decreases the subsequent intensification of agriculture and the attendant below 1,000 millimeters per year, malaria becomes less en- terracing and leveling of the land in and around human settle- demic, concentrating in progressively smaller valleys or where ments, which increased potential anopheline breeding places favorable microclimatic conditions and mosquito-breeding (Matola, White, and Magayuka 1987). places exist or are created by such human activities as irriga- Anotherpossiblefactorintheapparent increase in epidemic tion, dam construction, and the establishment of fish ponds. potential in the last few years in the highland areas in Africa Of course, altitude and rainfall are only rough indicators of is the so-called greenhouse effect, by which the accumulation malaria's endemicity; other factors, such as temperature, hu- ofcarbon dioxide and othergases in the atmosphere may retain midity, the distribution of rains, and the slope and permeability heat. In Madagascar the average temperature in the coastal of the soil, also play important roles. areas was 0.5 degree centigrade warmer than in the previous As endemicity decreases, the potential for epidemic out- thirty years; in the high plateau the difference was about I breaks increases because fewer people have a chance to de- degree centigrade. These figures may not be fully comparable velop immunity. Equally, in areas of marked seasonality, as in because data for the high plateau may be influenced by local the dry savanna of the Sahel, the transmission season, even if ecological changes, such as the growth of Antananarivo. Still, it occurs every year, takes on the characteristics of seasonal an increase of even 0.5 degree centigrade could increase the epidemics, at least as it affects younger age groups. Severe, potential transmission period in marginal areas, which might large-scale epidemics occur in areas that have been free from change a normally nonmalarious area into one subject to transmission for several years in a row, and they exhibit a seasonal epidemics (de Zulueta 1988). secular periodicity determined by the semicyclic occurrence of prolonged heavy rains or other climatological determinants. Morbidity in Other Malarious Areas Such epidemics have been historically reported in high- altitude areas following abnormally warm and rainy summers. Outside tropical Africa, most malarious countries have similar Such was the dramatic epidemic in the highlands of Ethiopia reporting systems that permit some degree of comparison. As in 1958, which caused more than 3 million cases and claimed for the intensity of the problem, about 80 percent of the 5.01 an estimated 150,000 deaths. They also occur in dry areas after million cases reported to WHO in 1990 (not including tropical abnormally heavy and prolonged rains, as in the epidemic of Africa) are concentrated in eleven countries. They are, in 1975 in Gezira and Central Sudan and the 1988 epidemic in decreasing order of total number of reported cases, India, Khartoum and Northern and Eastern Sudan. In 1988-90 a Brazil, Afghanistan, Sri Lanka, Thailand, Indonesia, Viet number of epidemics, or serious exacerbations ot endemicity, Nam, Cambodia, China, Solomon Islands, and Papua New occurred in several highland areas of Africa, particularly in Guinea. These countries represent 65 percent of the popula- Botswana, Madagascar, Rwanda, Swaziland, and Zambia. tion living in the world's malarious areas, excluding tropical In the high plateau of central Madagascar increasingly ex- Africa. India and Brazil, with only 26 percent of the popula- tensive and severe epidemics occurred between 1986 and tion, report 46 percent of the cases. With only 34 percent of 1988, reaching dramnatic proportions in the first four months the population, the first seven countries report 70 percent of of 1988, when tens of thousands of people died. One of the the cases. And within these countries, malaria is focused in maincausesofthisseriesofepidemicswasthat the DDTspraying certain areas. In India, for example, six states (Orissa, Uttar campaign of the 1960s and early 1970s seems to have elimi- Pradesh, Punjab, Madhya Pradesh, Gujarat, and Assam) have nated the main vectors of malaria A. funestu.s and A. gainbia, 66 percent of the cases. In Brazil, 97 percent of the cases are in sensu stnicto so that malaria transmission was interrupted for Amazonia, which has only 15 percent of the country's popula- about twenty years. But after spraying was discontinued and tion; two states (Rond6nia and Para) report 70 percent of the because of other opportunistic circumstances, both species cases, and four municipalities in Rond6nia and four in Para progressively spread in the high plateau. report more than 60 percent of the cases in those states. 286 Jose A. Nijera, Bernhard H. Liese, and Jeffrey Hammer These countries and areas show great variability not only in Drug Resistance and Proportion of P. falciparum the intensity of the problem but also in its evolution in time. Cases reported annually to WHO since the mid-1960s show The proportion of P. falciparum in endemic areas outside distinct patterns: tropical Africa, where P. falciparum remains the predominant * Malaria declined and the situation has remained favor- species, was 38 percent in 1990 (compared with 15 percent in able in Algeria, China, Costa Rica, Cuba, Egypt, Korea, the early 1970s). In the last few years there has been an peninsular Malaysia, Morocco, Panama, Paraguy,' and Tu- increased selection and progressive dispersal of P. falciparum nisa. M parasites resistant to antimalarial drugs, because these drugs are n Malaria increased markedly in certain areas in Afghani- used increasingly as prophylactics and for self-medication, * Malaria increased markedly in certain areas in Afghani- usually in insufficient doses (see map 13-2). The problem of stan,Bangladesh Belize, BhutanBolvia, BrazilCambodia drug resistance has been particularly alarming in Africa; in Colombia, French Guiana, Guatemala, Guyana, Madagas- recent years it has spread across the continent and is now car, Mexico, Myanmar, Nepal, Papua New Guinea, Peru, developing rapidly in West African countries. Its continual the Philippines, Saudi Arabia, Solomon Islands, Thailand, intensification hampers efforts to provide adequate treatment Vanuatu, and Viet Nam. in rural areas. It is difficult to assess how much to attribute this * The incidence of malaria has oscillated-at relatively phenomenon to the migration of resistant parasites and how short cycles and with a quasihorizontal general trend, at least much to local selection, because both mobility and drug con- since the early 1960s-in Argentina, El Salvador, Hondu- sumption have increased considerably. For some time the ras, Indonesia (the outerislands, reporting since 1970), Iran, widespread use of chloroquine was advocated as the most Malaysia (Sabah), Nicaragua, Surinam, and the Republic of effective way to reduce deaths from malaria in Africa; Yemen. chloroquine, it was said, should be treated as a commodity and * In the last twenty years, one or two relatively short but not as a drug. In many places in Africa people use chloroquine significant resurgences, followed by renewed control, have more often than aspirin for minor fevers and aches. Chloro- occurred in the Dominican Republic, Ecuador, Haiti, India, quine has, no doubt, helped reduce deaths from malaria, but Indonesia (Java and Bali), Iraq, Libya, Malaysia (Sarawak), maintaining this gain will require targeting antimalarial drugs Mauritius, Oman, Pakistan, Somalia, Sri Lanka, Syria, Tur- to those actually suffering from malaria, particularly in areas key, Venezuela, and Yemen. where resistant parasites require the use of more toxic and less .. . . ~affordable drugs. The only purpose of this preliminary classification is to stimulate analysis of the patterns of change. Groupings are based on total numbers of reported cases, and the fact that two Mortality countries appear in the same group does not indicate similari- ties in other epidemiological characteristics. Most deaths from malaria occur in tropical Africa. As in all Except for India and China (which are significant producers highly endemic areas, deaths occur most among the young. of cases because of their great size) and Sri Lanka (which may Maternal immunity transmitted to infants may reduce mortal- be changing from a pattern of periodic resurgence because of ity in the first three to six months of life, but this effect may be sociopolitical unrest), most large producers of cases are in the masked in areas of marked seasonality. Past studies indicated second group. These countries are characterized by recent mortality rates between ten and thirty per thousand in infants efforts to increase the exploitation of natural resources and between about seven and eleven per thousand in children (through agricultural colonization of forest or jungle areas) or one to four years old. In 1962 the WHO Regional Office for by civil war or sociopolitical conflict (including illegal drug Africa estimated that every year between 200,000 and 500,000 trade) and large movements of refugees or other mass migra- African children die from malaria (Pampana 1969). In 1969, tions. Eight of the eleven main producers of cases have been Bruce-Chwatt put that figure at about I million, a figure on that list since at least 1986. extensively quoted ever since. Molineaux (1985), reviewing All the countries in the first group, where malaria has the effect on infant mortality of some malaria control projects, declined, have shared a degree of social stability and socioeco- especially in Kisumu (Kenya) and Garki (Nigeria), concluded nomic development, including health services accessible to that malaria was responsible for about 20 to 30 percent of the public. The countries in the third group have suffered infant deaths. Greenwood and others (1987), studying deaths periodic bouts of malaria, followed by remobilized control from malaria in the Gambia, concluded that the mortality rate efforts, after which the situation improved but could not be from malaria was 6.3 per 1,000 for infants and 10.7 per 1,000 maintained, so malaria recurred. This pattern of "fire fighting" for children one to four years old, representing 10 percent of may progressively improve matters in the more developed areas the deaths of children less than a year old and 25 percent of as people become less tolerant of epidemics and health services deaths for children one to four. become more responsive. In marginal areas, the response is There are signs that in some parts of Africa general infant nearly always late and possibly ineffective, because it often and malaria-specific mortality may be declining, often inde- comes when the epidemic is naturally declining. pendently of specific interventions, reflecting social develop- Malaria 287 Map 13-2. Areas where Chloroquine-Resistant Plasmodium Falciparum has been Reported r 9,,g'It,efRs~~~~~~~ Wol Healt Org aniaon 199 ' t4'\?, fS "'i2' ' -.L * Reported after 1988 N~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~S( 1 1 Note: The designationLs employed and the presentation of material on this map do not imply the expression of any opinion whatsoever on the part of the World Health Organization conceming the legal status of any country, territory, city, ot area or of its authorities, or conceming the delimitation of its fron- tiers or boundaries. Souxrce: Reproduced by permission of the World Health Organization, Geneva from Weekly Epidemiological Record 66 (22): 162 ( 1991). ment and general education. Studies in the Congo and Burkina mortality, being relatively low, did not change significantly in Faso in the late 1970s indicated that malaria-specific mortality the year of intervention. This study (Spencer and others 1987) might be lower than expected in areas where, some decades confirmed in a small rural area the general observation that ago, malaria was a significant cause of infant mortality. The differences in child mortality can be explained largely by authors (Vaise and others 1981) attributed their findings to differences in maternal education, which no doubt influences the widespread, albeit indiscriminate, use of antimalarial the amount of drug use but, more important, improves hygiene drugs. Often these drugs were used in doses inadequate to and general living standards. Differences in infant mortality eliminate parasites but effective enough to produce a clinical between districts in Kenya, as reported in the 1979 census, cure and prevent death, even if collectively such use could be ranged from 38 to 153 per 1,000. When spraying in Kisumu contributing to the selection of drug-resistant parasites. The ended, in 1976, the area did not return to the previous infant wide availability of antimalarial and other active drugs has also mortality rates. Infant mortality rates for the district in which been identified as possibly contributing to the general decline this area is located declined from 220 in 1959 to 181 in 1969 in infant mortality observed in the Kisumu area of Kenya. and to 147 in 1979 (Spencer and others 1987). There, between 1972 and 1976, infant mortality reportedly Deaths from malaria outside tropical Africa occur mainly declined from 157 per 1,000 to 93 per 1,000 during an effective among nonimmunes who become infected by P . falciparum and malaria control program (spraying fenitrothion inside houses). get sick where appropriate diagnosis and treatment are un- Between 1981 and 1983, a decline in postneonatal mortality available. This happens especially to newcomers to endemic (from 73 to 67 per 1,000) and a marked drop in the mortality areas, such as agricultural workers, laborers, gold and gem of children of one to four years (from 25 to 18 per I ,000) were miners, and prospectors in recently colonized or other frontier recorded after implementation of a program of community- areas of economic development. Most affected are young based antimalatia treatment. But most of that decline was adults, although whole families ofsettlers may be affected-for attributed to a measles epidemic in 1981-82; malaria-specific example, in the tropical jungles of South America (especially 288 Jose A. Ndjera, Bernhard H. Liese, and Jeffrey Hammer the basins of the Amazon and the Orinoco), in the outer There have been few community-based methods used to eval- islands of Indonesia, Sabah, Kalimantan, and, on a smaller uate the economic effect of malaria. The study by Conly scale, throughout the tropics. In the Brazilian Amazon, an (1975) highlights the problems encountered in attempting to estimated 6,000 to 10,000 people a year die from malaria (J. quantify the effect on economic development resulting from Fiusa Lima, personal communication, 1990). the difficulty of collecting and processing data on a sufficiently large population and the complexity of the interactions of the Possible Patterns of Morbidity and Mortality: 2000, 2015 parameters measured. Malaria transmission is focal and depends on the dynamics COSTSOFMORTALITY AND MORBIDITY. The costs that malaria between humans and the vector, parasite, and environment. imposes are borne through increased mortality and through More important, it depends on the effectiveness of control high morbidity rates. The effect of mortality will vary with the efforts, socioeconomic development, and political stability. It age distribution of deaths, which in turn vary by ecological is thus quite risky to generalize about future patterns of mor- zones. In Africa, where most deaths are among infants and bidity and mortality. Given the increasing resistance of the young children, the effect of and the perception of mortality parasite to antimalarial drugs, however, treatment of malaria will be different from other areas, where the deaths are among in the future will be more difficult and less effective, thereby the main breadwinners or primary caretakers of families. Mor- increasing the risk of both morbidity and mortality. Although tality and morbidity among adults are high in areas of low to new drugs are being investigated, and work is progressing on moderate endemicity. See Over and others 1989 on the con- various potential malaria vaccines, alternative first- or second- sequence of adult deaths. line drugs or a vaccine are, even under optimistic circum- stances, several years away. MALARIA AND PRODUCTIVITY. Public health activities have Furthermore, population instability in areas in which there been justified as improving productivity ever since debates is the potential for malaria transmission is usually associated about the "laziness disease" at the beginning of this century with an increasing burden of illness, including mortality due (Garcia 1981). Malaria is a classic example of a debilitating tomalaria.Thisinstabilitymightresultfrompoliticalconflicts, disease that impairs productivity. As the most prevalent dis- economic development or relocation schemes, migration be- ease in the poorest rural areas, malaria produces recurrent cause of population pressure, or natural disasters. Examples infections with attacks of fever in the warm and rainy seasons, from Brazil, Thailand, Indonesia, and Sri Lanka may illustrate when most workers are needed to collect crops. Often, affected different facets of this problem. To the extent that develop- people also suffer from malnutrition and other infections and ment projects ignore effects on health or that political conflicts lack of medical care. In areas subject to epidemics, these also create large refugee populations, malaria is likely to increase in tend to strike at times of peak demand for agricultural work. each such local situation. Conversely, adequate socioeco- The focus of much of the research has been on attempting nomic development and political stability will facilitate effec- to measure the effects of bouts of illness on lost output of tive and sustainable malaria control. workers. This research has been reviewed in Barlow and As to patterns of morbidity and mortality in the future, it is Grobar 1986. Research on the physical effect of the disease can safe to predict that there will be more morbidity and possibly be found in Conly 1975; Malik 1966; Russell and Menon 1942; mortality as a result of malaria in several areas, but how much and van Dine 1916. Days of disability per case of malaria more and, more specifically, which regions will bear most of estimated in these studies range from five to twenty. Other the burden, is less obvious. Well-documented case studies physical measures include effects on output or land cleared. could serve as examples of potentially devastating effects, Audibert (1984) sets values for the former varying from 0 to given similar scenarios in other areas of the world (that is, 1.5 for the elasticity of output of rice with respect to malaria Brazil, Madagascar, Sudan, Indonesia, Afghanistan, Sri Lanka, prevalence, whereas Bhombore, Brooke Worth, and and so on). Nanjundiah (1952) found a reduction of 60 percent in cropped But most important in view of the presently deteriorating area in families with malaria. Conly (1975) traces a variety of worldwide malaria situation referred to earlier, a forceful effort adjustments in farm families in Paraguay, including increases to rehabilitate, activate, or develop new malaria control activ- in labor input per unit output as well as reallocations of land ities in those countries most affected is crucial. Without such and hired labor. The reallocations of land entailed replace- an effort malaria patterns in the year 2000 will be entirely ment of crops of high value whose crop season was malaria different and substantively worse than the ones predicted. prone with crops of relatively low value whose crop season was not. De Castro (1985) finds that such reallocations may in- Economic Costs of Malaria clude an increase in the workload of healthy family members. Although this may be seen as an ameliorative factor which The economic costs of malaria theoretically would include its reduces the net effect of the disease, it may simply mean that effect on the economy and economic development, on the some costs of the disease are borne by others besides those who local community, on the household, and on the individual. are ill. In Southem Rhodesia it has been estimated that the Malaria 289 loss of manpower to malaria was 5 to 10 percent of the labor of migrant workers than in the houses of village residents. force, with the heaviest incidence at the peak of agricultural Malaria has been concentrated among migrant laborers in production. most areas where there is extensive cultivation of cotton and Estimates are much lower in highly endemic areas, where sugar cane and in some areas where coffee, bananas, and rice anyone who survives childhood can generally tolerate a ma- are cultivated. Litsios (1990, p. 8) concludes that "as malaria laria infection, showing only minor symptoms at most, al- becomes a problem to be primarily found in marginal or fringe though malaria is an important contributing factor in severe areas, it becomes a problem that is identified with marginal anemia. Brohult and others (1981), in a study in Liberia, found people." This can be seen by the tendency in some countries no detectable loss of physical ability in people with malaria to associate malaria with "foreigners" or minority migrant parasites in their blood, but they did find a marked correlation groups, who are then accused of being responsible for carrying between anemia and loss of physical ability. the parasites into areas that might otherwise be malaria-free. It is a common convention in the literature to use the In endemic areas the burden of malaria is also bome dis- parameter of seven days of work lost to disability per bout of proportionately by the poor. malaria in assessing a program when the parameter is not independently estimated (see Niazi 1969; Quo 1959; San PROBLEMS IN MEASURING COST. The focus on days lost from Pedro 1967/68; and Sinton 1938); when independently esti- work or output forgone is oddly narrow. With regard to welfare, mated, the parameter varied between five and fifteen days. A two altemative measures would be preferable. The first is the further issue, raised by Wemsdorfer and Wemsdorfer (1988), compensating variation in income, or, the amount one would is the undermining of the effectiveness of investment in edu- be willing to pay to avoid having the disease altogether. The cation. In highly endemic areas, where adults normally have second is the equivalent variation, the willingness to accept, acquired sufficient immunity to make the symptoms less se- or the amount one would need to be paid to accept having the vere, schoolchildren are more severely affected. Judging the disease. In many contexts, these two measures are similar, but degree of impairment caused by illness would be hard to do and in cases in which possible mortality is involved the latter could one can only wonder at the cost. Macdonald (1950) estimated generate much larger measures of costs than the former. Will- that the leaming of 35 to 60 percent of children may be ingness to pay is bounded by ability to pay or lifetime eamings. impaired by malaria. In practical application, it would be limited by borrowing Other studies emphasize direct financial benefits from activ- constraints as well. Willingness to accept is under no such ities made possible by eradication or control. These are also limitation. Either of these measures would capture the subjec- surveyed in Wemsdorfer and Wernsdorfer (1988). Malaria has tive, even psychological effect of the disease. In any case, using hurt economic development projects as well as armies at war a measure which respects personal preferences would be more and police forces or border patrols in endemic areas. Malaria inclusive than simply including the instrumental effects of the had to be overcome for the successful construction of the disease on the productive capacity of the worker. There are a Panama Canal and most roads and railways in tropical coun- number of objections to using these altemative concepts. They tries, for the agricultural development of the Roman campagna require defining which set of preferences (before or after falling and the Venezuelan Ilanos, for the building of railways and ill) is relevant for the comparison. The most important criti- roads in tropical areas, and for the protection of armies from cism, though, is that obtaining this number requires a signifi- World War I through Viet Nam. Agricultural development cant research effort. Such calculations have been made in the and mining in tropical jungle areas, which attract workers from literature on environmental effects but are not widely used. It densely populated, often nonmalarious, areas, form one of the is possible to infer a lower bound, though, by calculating the particular problems associated with malaria. An example of total costs required to obtain treatment. The total costs bome research into the benefits of malaria control in this situation by families and individuals include payments for treatment, is given by Griffith, Ramana and Mashaal (1971), who esti- time and transport costs in seeking treatment, time costs for mate the increased profits derived from allowing workers to family members who look after the patient, and time and enter new areas for mining. Forgone profits are the measure of money costs of preventive action taken by households and the the cost of disease. Sinton (1938) documents many cases in community. These costs vary greatly with variables such as India where malarious regions prevented an expansion into access to primary health care, national drug distribution poli- new territories, resulting in substantial losses in forgone earn- cies, the presence of chloroquine resistance, the level of ma- ings. Demographic changes since then, however, have proba- laria endemicity, the behavior and bionomics of the local bly made such opportunities for expansion much rarer in the vector(s), and whether or not malaria is perceived as a serious subcontinent. health problem by the local community. There are two main sources of underestimation. First, this calculation misses costs DISTRIBUTION OF COST. Litsios (1990) highlights the uneven before treatment is sought. Second, there are those (inherently distribution of malaria risk across a population. Data from hard to measure) who have decided that the costs of seeking Adana, Turkey (Yumer 1980), for example, show that anoph- treatment are too high in relation to the costs of letting the eline bites perperson were five times more frequent in the tents disease run its course. For them, there is still a relevant cost 290 Jose A. Ndjera, Bernhard H. Liese, and Jeffrey Hammer and degree of necessary compensation. For people in remote Committee formulated a strategy for eradicating malaria (wlio areas, or those afflicted at peak agricultural seasons (when 1957). implicit wages are high, both for the person falling ill or, in the Soon after the WHA resolution and the report of the Expert case of children especially, for those needing to accompany the Committee, most countries of the Americas, Europe, North person), or (more difficult to evaluate) those who are unin- Africa, Asia, and the Pacific officially declared that their forned about treatment prDspects, these costs can be high. antiinalaria programs were eradication campaigns. In retro- In their careful study of Thailand, Kaewsonthi and Harding spect we can see that many of these programs were short on (1986) attempted, among other things, to measure the costs epidemiological knowledge and administrative capacity. borne by patients in seeking care. These were dubbed "external These deficiencies were overlooked because of the programs' costs," that is, external to the malaria control organization. humanitarian appeal, the sense of urgency, and the feeling, They amounted to $20 per positive case, or nine times the shared by many, that peer pressure could shake the chronic average wage. This estimate is for people presenting at the apathy of the health services. malaria clinic and therefore does not include those who have As anticipated, tropical Africa and some parts of Southeast handled the disease in other ways. This study includes costs Asia posed problems, because of their high endemicity, prim- entailed in seeking local treatment before travel to the clinic. itive state of development, and lack of human and economic These amounted to 15 percent of the costs per positive case resources. Successes elsewhere, although slower than ex- and are a component of the full cost to those who do not seek pected, were still remarkable. But as more and more areas formal treatment. The degree of underestimation of the cost advanced into their program's consolidation phase, the expec- to sufferers is probably quite high. Time lost before and after tation that a surveillance mechanism would maintain areas seeking treatment can be considerable and varies with the malaria-free, after spraying was interrupted, was not fulfilled. quality (primarily speed) of service provided. This varies sub- Resurgences occurred increasingly often in the consolidation stantially within Thailand, let alone across countries. and maintenance phase, particularly in Central America and Because the costs of malaria are borne disproportionately by Southeast Asia. And at the end of the decade a massive the poor, there are further issues of-aggregating individual costs epidemic broke out in Sri Lanka, where malaria had been into social costs. Whether disease averted should be weighted almost eradicated. Evidence began to accumulate that, al- by the income of the sufferer, because of social welfare consid- though it was possible to reduce and even interrupt malaria eration or the possibility of successfully seeking treatment, is transmission by spraying insecticide in large areas, it was an ethical issue to be appraised by policymakers. difficult if not impossible to establish effective surveillance In summary, Andreano and Helminiak (1988, p. 35) state without a solid health infrastructure. that "despite the many studies and the excellent work by Finally, in 1969, after reexamining the global strategy of Barlow and Conly, whichi represent methodological advances eradicating malaria, the World Health Assembly reaffirmed in the study of tropical diseases, we remain woefully ignorant that eradication was the ultinate goal but stated that, in of the social and economic effect of malaria in those countries regions where eradication was not yet feasible, control of of the world where it is prevalent." They also emphasize that malaria with the means available should be encouraged and findings in many of these studies cannot be easily generalized may be a necessary and valid step toward that goal (WHO 1969). from one area to another. Unfortunately, after fifteen years of strictly regimented anti- malaria action, health authorities-even malariologists- Malaria Control were reluctant to introduce the necessary changes in the programs while the concept of malaria control, and an accept- The idea of eradicating malaria, postulated as early as 1916, able global strategy for it, remained undefined. The Expert gained currency after World War 11. Malaria epidemics had Committee provided only sketchy guidance on how to trans- devastated parts of southern Europe, and DDT had been ex- form an ineffective malaria eradication programn into a control tremely effective in controlling not only those epidemics but program, emphasizing that "the objectives in these areas would also endemic malaria in both temperate and tropical areas, be to consolidate the gains so far achieved, to extend the including Venezuela, British Guiana, and Taiwan. The Expert programme to areas where protection would give maximum Committee on Malaria of the newly created World Health socioeconomic benefit and to protect high risk groups" (w1oo Organization, in its first five reports, adopted a cautious 1974, p. 30). Unfortunately, in most countries it was thought attitude, expressing concern about increasing reports of tech- that the only way to consolidate gains achieved was to main- nical problems and of some disappointing results in the use of tain as many routine activities as could be afforded, without DDT, particularly in Africa. But the goal of eradicating malaria making the necessary investment to evaluate their local effec- became irresistible, and the impending resistance to DDT was tiveness. seen as a reason for racing to eradicate malaria before resistance The formulation by WHO in 1978 of a strategy to develop a developed. In 1954 the Pan-American Sanitary Conference health care infrastructure included malaria control among its adopted a continental plan to eradicate malaria from the essential elements (WHoU,NICEF 1978). In line with these de- Americas. In 1955 this plan was extended to the world by the velopments, the thirty-first World Health Assembly adopted World Health Assembly (WHA). In 1956, the Sixth Expert a strategy of malaria control aimed at least at reducing mortal- Malaria 291 ity and the negative social and economic effects of the disease, second would be developed progressively, according to the preventing or controlling epidemics, and protecting malaria- intensity of the problem and resources available. free areas, with the ultimate objective of eradicating the dis- ease whenever feasible (WHO 1978). New Perspectives for Control Malaria Control Measures In 1985 the thirty-eighth World Health Assembly expressed its continuing concern about resurgent malaria and, in partic- The most common antimalarial measures are (a) chemical ular, about the apparent inadequacies of existing malaria con- control through residual intradomiciliary spraying with DDT or trol strategies. Consequently, the WHO Expert Committee on other insecticides and, in selected instances, aerial spraying or Malaria (wHio 1986) reviewed the global mnalaria situation and local fogging, (ultra low volume); and (b) the treatment of attempted to develop further the epidemiological approach to fever cases with antimalarials. These activities are sometimes malaria control, which had been proposed by the Expert Com- supported by limited environmental management measures mittee in 1979, giving particular emphasis to socioeconomic mostly in urban areas where such measures can be easier to factors. Within the epidemiological approach is the recogni- implement than in rural ones. They involve drainage or filling tion that variability among diverse malaria situations is the of water bodies. Water-level fluctuations or intermittent irri- result of a multitude of factors, which will also affect the gation are used in some large development schemes. For all effectiveness of control measures. Mapping the distribution of practical purposes biological control measures are presently of these determining factors would constitute a "stratification" of little relevance. In addition to these active intervention mea- the local malaria problem and would provide a useful frame- sures, all control programs undertake active or passive malaria work for selecting and testing appropriate sets of control inter- surveillance. Specific antimalarial measures can be classified ventions. according to their mode of action and the scope and scale of their use (table 13-1). Identification of Malaria Patterns Two substantially different approaches may be pursued in malaria control. In practice, the identification of all relevant epidemiological factors has not come easily to control program managers. They *Improving general health services to ensure adequate diagnosis, access to health care, and treat.ent for individual are otten not equipped to analyze and interpret the massive malaria cases, . well as promoting personal adconmunity quantities of epidemiological, parasitological, and entomolog- malaria cases, as well as promoting personal and community clifrainta edtob olce-n oueti protection. The ai of this approach is to eliminate deaths ical information that need to pbe collected-and to use this fro mlai an to reuc th seert an duato of information to define appropriate control actions. In particu- from malariatad . t c t lar, those in charge of control programs have generally lacked illness associated with It. the ability to see specific malaria situations in their economic * Establishing the capability for long-term control of mna- and social context; that is, to analyze the relationships between laria transmission, control and prevention of epidemics, and patterns of human occupation and exploitation of the environ- progressive reduction of malaria endemicity (particularly in ment and trends in malaria transmission. areas affected by P. falciparum). Nevertheless, accumulated experience and some specific The two approaches are in no way mutually exclusive, and studies of problem areas showed that there are identifiable ideally they should be complementary; however, they differ ecological and social situations in which malaria is not only greatly in their requirements for specialized services. Whereas more frequent and serious but also more difficult to control. In the former is a basic requirement in all malarious areas, the the Brazilian Amazon, for example, economic, social, environ- Table 13-1. Ma1aiia Control Measures Action Individual and famity protection Community protection Reduction of man-mosquito contact Bednets, repellents, protective clothing, Site selection, zooprophylaxis screening of houses Destruction of adult mosquitoes Use of domestic space spraying Residual indoor insecticides, space spraying, ultra-low volume sprays Destruction of mosquito larvae Peridomestic sanitation, intermittent Larvicide for water surfaces, intermittent drying of water containers irrigation, sluicing. biological control Source reduction Peridomestic sanitation, small-scale Environmental sanitation, water drainage management, drainage Destruction of malaria parasites Early diagnosis and treatment, Establishment of diagnosis and treatment chemoprophylaxis facilities, chemoprophylaxis for pregnant women, mass treatment Social participation Motivation Health education, community participation Source: Adapted from Bruce-Chwatt 1985. 292 Jose A. Ndjera, Bernhard H. Liese, and Jeffre' Hammer mental, and political factors have converged to produce three 1979 and Bruce-Chwatt and Archibald 1959 [Sokoto]; Foll epidemiological patterns, collectively referred to as "frontier and others 1965 and Najera and others 1973 [Kankiyal; and malaria" (Marques 1988; Sawyerand Sawyer 1987; Wilson and Wilson 1960 [Pare Taveta]). There are some areas or popula- Alicbusan-Schwab 1991). These patterns are found in the now tion groups, however, in which vector control may be feasible; famous "garimpos" (gold-mining areas), in areas of new agri- in particular it may be possible to introduce effectively insec- cultural settlement, and in the rapidly expanding periurban ticide-impregnated bed nets or curtains. areas of the region. Although found in less than one in ten municipalities, they account for more than 80 percent of all PLAINS AND VALLEYS OUTSIDE AFRICA. Characteristics: these malaria cases reported. Similarly, most malaria situations areas correspond to the classic descriptions of malaria as a rural throughout the wvorld, when viewed in their social and eco- disease, being more intense in the poorest areas and in periods nomic context, fall into a few main types. of economic depression. As in the African savanna, transmis- It has been suggested (Najera 1981, 1989) that it is possible sion may he from continuous to seasonal, depending on lati- to recognize and describe a limited number of prototypes or tude, altitude, and aridity. The risk of transmission tends to typical patterns, synthesizing, from a global perspective, those increase with the introduction or extension of irrigation, but observations in different countries, complemented with sum- it considerably decreases with good water management and the maries of control experiences in such situations. These descrip- improvement of farming techniques, houses, and animal shel- tions of epidemiological patterns (which have been referred to ters. In most of these areas malaria was brought under control as "prototyes" or "malaria paradigms") could help health plan- by the early eradication campaigns, and vector control has ners in the important task of designing and implementing continued over the last three decades. These areas show low appropriate sets of control measures, either to develop new endemicity and should continue to do so unless disturbed by programs or to adapt existing ones. civil unrest, insurgence, or war, which would not permit the functioning of health services. Malaria Patterns and Specific Ecological Conditions Control: in most instances, it will be possible to maintain this favorable situation through the continuing development The main determinants of malaria transmission (vector den- of their health and epidemiological services and by theirability sity and survival, human-vector contact and duration of para- to detect and control potential risk situations. site development in the vector) are dependent on availability of surface water and climate, which in turn have also influ- FOREST AND FOREST FRINGE AREAS. Characteristics: the exten- enced the distribution of rural populations and their agricul- sive forest areas of Africa, South America, and Southeast Asia tural activities. It is therefore possible to identify major have increased in importance as the exploitation of forest differences in the epidemiology of malaria associated to the resources has intensified. Malaria risks are associated with the main types of ecological areas, which would provide a first type of human activity which modifies the microenvironment characterization of the epidemiological pattern, unless man and the relation of humans and vectors to it. Nomadic and has sufficientlv disturbed the environment and introduced seminomadic tribal populations of forest areas, engaged in some of the patterns referred to in the next heading. gathering and hunting, are generally too dispersed and mobile to sustain intense transmission. In the fringe of the forest or THE AFRICAN SAVANNA. Characteristics: the African savanna deforested areas, sedentary populations tend to be engaged represents the highest malaria endemicity in the world. The mainly in agricuitire, but they also use the forest to collect factors responsible for high levels of continuous transmission firewood and for hunting. In Africa the main malaria vectors, include propitious climatic conditions for vector breeding and A. gambiae and A. funestus, follow man into the forest and, the presence of such highly efficient vectors as A. gambiae and although they are more easily controlled than in the savanna, A. fhiiestuts. This pattern is characterized by high frequencv of they are able to maintain the same levels of verv high ende- illness among young children and pregnant women, high child- micity. In Asia and the Americas, settled population groups. hood mortality, and high frequency of asymptomatic infec- engaged in regular agricultural activities in deforested areas, tions in older children and adults. Transmission may become have a differenc malaria experience from that of those engaged seasonal in areas with less rainfall and at higher altitudes. in forest activities. The former suffer mostly from P. tviax Recently, the malaria problem has been aggravated by the infection and tend to have much lower malaria incidence, rapid spread of drug resistance across the African continent. easily controllable with residual insecticides. In contrast, those Control: in the African savanna the most important goal of engaging in activities at the edge of or inside the forest have a malaria control is to reduce the effect of the disease by provid- high risk of acquiring P. falciparum malaria. ing effective treatment to all people suffering from malaria, Control: residual insecticides are practically ineffective which will require extension of services and health education against the highly exophylic (outdoor biters) forest vectors. to improve their use by the population. Pilot projects, aimed Protection has been traditionally dependent on the use of at the interruption of transmission in savanna areas, have been drugs, often excessive and irregular, because of the absence of only partially successful, and institutional problems have con- organized curative services. When intemational borders run strained expalnsion of vector control programs (Bruce-Chwatt across these areas, as is common in South America and South- Malaria 293 east Asia, there may be a concentration of illegal activities, those that can make a living from the existence of the tourist which make areas even less accessible to programmed control. resort, creating situations similar to those of periurban slums. Chloroquine-resistant P. falciparum originated in areas of this Control: disease control in tourist resorts is similar to that in type, both in the Colombian-Venezuelan and the Thai- urban areas. For rural coastal populations, whether they are Cambodian borders (Field 1967). Today more effective means engaged in agriculture or fishery, the basic measure should be of personal protection, such as pyrethroid-impregnated bed case management and engineering methods, such as opening nets and repellents, offer a possibility of complementing the or flushing estuaries, land reclamation for agriculture or tree partial effect of currently available antimalarial drugs and plantation, regulation of water courses, and so on. eventually reducing the dependence on chemoprophylaxis. URBAN MALARIA. Characteristics: except for some cities in HIGHLAND FRINGE AND DESERT FRINGE. Characteristics: alti- southem Asia, where A. stephen.si is fully adapted to the urban tude, drainage, and temperature are limiting factors in both environment, malaria transmission does not occur in well- mosquito breeding and in parasite development in the mos- established, densely populated urban areas. Nevertheless, quito. Therefore, these factors have an important effect on the many tropical cities are surrounded by rapidly growing slums, potential for malaria transmission along the fringes of highland which are basically a high concentration of shelters in what is areas. The highlands themselves, which tend to have less still primarily a rural environment. Such situations increase transmission of malaria, often are characterized by high popu- the risk of malaria transmission. Eventually a high contamina- lation density and pendular migration between the highlands tion of surface waters may prevent anopheline breeding before and neighboring valleys. These neighboring areas, which offer urbanization reaches the slum areas. Malaria transmission in economic opportunities on plantations or in other develop- urban areas varies considerably in space and time, but in ment projects, often have more transmission. Unusually warm certain situations it may be very high. rainy seasons may cause serious epidemics in highland areas of Control: malaria control in urban areas relies on environ- low endemicity, resulting in high mortality (for example, East mental sanitation, in order to eliminate existing mosquito- Africa and Madagascar in 1987-90). In Southeast Asia, vec- breeding sites and prevent the creation of new ones. In tors which breed in foothill streams (for example, A. minimus addition, human-vector contact can be reduced through im- and A. fluviatilis) are more efficient vectors than those in the proved house construction and personal protection. plains. Therefore, the foothills in such areas tend to be more malarious than the plains. In transitional zones adjoining Malaria Patterns Associated with Specific Occupations deserts, the lengthy dry season also limits vector proliferation or Social Conditions and malaria. Also, the populations of such areas tend to be dispersed and nomadic; epidemics may occur in years of excep- There are a number of socioeconomic activities which create tional rainfall or with the introduction of irrigation. major disturbances of the environment, attract large numbers Control: as everywhere, effective disease treatment is funda- of temporary workers, or disrupt the social structure and there- mental in both highland and desert fringe areas. In addition, fore the health care system. All these activities transform the surveillance for the monitoring of epidemic risk indicators, and basic epidemiological parameters, as determined by the physi- for the early detection of epidemics, is crucial. In different cal ecology of the area. The new malaria patterns so created areas, different responses may be feasible. These responses may may be not very extensive in area or may not persist very long include a combination of preventive vector control, strength- in the same location, but where and when they occur, they may ening of treatment facilities, and mass fever treatment. represent authentic epidemiological explosions and often leave marked sequelae of environmental degradation when SEASHORE AND COASTAL MALARIA. Characteristics: the most they are abandoned. typical situation is found where the mosquito vector breeds in brackish waters. Such mosquitoes are generally less efficient as AGRICULTURAL COLONIZATION OF JUNGLE AREAS. Characteris- vectors than those of neighboring inland areas, as is true of A. tics: areas of new colonization attract displaced people either melas and A. merus in Africa and A. aquasalis in South Amer- from cities or from densely populated areas that often have low ica. Still, in Southeast Asia and the Pacific, A. sundaicus and malaria endemicity. Many people have, therefore, no or little A. farauti are responsible for serious malaria transmission. In acquired immunity and suffer severely from malaria when some coastal areas, as in Central America and Mexico, fresh- exposed to the high-transmission risk in the jungle environ- water-breeding mosquitoes may cause intense seasonal trans- ment. The effectiveness of vector control based on in- mission by breeding profusely in estuaries closed during the dry tradomiciliarv spraying is limited in these areas because season by a sand bar. shelters are generally precarious and the vector does not always A frequent form of economic development in coastal areas feed or rest indoors. Social services in these areas are weak or is the establishment of tourist resorts, which often make im- absent. In time, the situation tends to improve as these settle- portant investments, not only in malaria control, but also in ments become more developed. This pattem is found in Brazil, pest mosquito control, for the protection of the installations. parts of India, and the outer islands of Indonesia (Binol 1983; The development of tourism often attracts more people than Marques 1988). The agricultural settlement of large new areas 294 Jose A. Ndjera, Bernhard H. Liese, and Jeffrey Hammer is usually accompanied by the rapid growth of supporting urban tions in which the civilian population suffers a lack of basic centers as well. These centers attract large numbers of poor, supplies, destruction of houses, and considerable displace- often unemployed or underemployed migrants, who settle in ments, and even temporary or permanent housing in refugee precarious conditions on the urban periphery. This explosive camps. These situations, combined with the disruption of periurban growth is also associated with high levels of malaria health services, may cause epidemic outbreaks even in areas transmission (Sawyer 1986; Sawyer and Sawyer 1987). previously well under control. These outbreaks particularly Control: traditionally, protection has been dependent on the affect the civilian population; the military and police contin- use of drugs, especially during the initial phases of settlement. gents are likely to benefit from organized control in their camps The use of drugs, however, has often been excessive and and chemoprophylaxis while in action. irregular because of the absence of health services in these Control: in these situations control depends on the size and remote areas. Residual insecticides have proven less effective organization of the refugee population and the intensity of the against highly exophylic forest vectors. In some areas tradi- problem. It may be possible to consider mass fever treatment, tional vector control activities may be possible. Furthermore, temporary chemoprophylaxis, and even spraying of shelters. measures for personal protection, such as pyrethroid-impregnated Sometimes relocation of camps and some sanitation measures bed nets and repellents, may be introduced in combination are possible. with appropriate information and health education. Patterns arud Measures for Control GOLD AND GEM MINING. Characteristics: malaria is usually serious in remote forest areas among populations of miners who The matrix in table 13-2 relates the pattems identified above migrate frequently between existing mining areas, new mining with control measures that have or have not proven effective areas, and urban and rural areas (for example, in Brazil and in malaria control programs. This table is neither comprehens- Venezuela). Occupation of these areas is often temporary, and ive nor prescriptive; its intent is to help operationalize the investment in basic infrastructure and services is rare, espe- concept of epidemiological stratification. cially in countries in which small-scale mining is illegal. P. It must be noted that the diagnosis and treatment of cases, falciparum drug resistance is frequent (for example, the Thai- including the management of drug resistance, applies equally Cambodian border, the Colombia-Venezuela border, and the to all patterns. Case management and drug treatment, which Brazilian Amazon). Because they have tended to penetrate in most cases represents the care of fever, without specific deeply into frontier areas, these gold and gem miners have diagnosis, are dependent on the structure of the general health often exposed highly vulnerable indigenous peoples to malaria care system. Diagnosis and treatment should be undertaken by and other diseases, with disastrous consequences. the general health services; in areas in which such services are Control: in these areas malaria control activities are exceed- weak or nonexistent, such as in forest fringe areas and frontier ingly difficult. Case management is clearly a priority. Recently, areas, special fever-treatment posts or malaria clinics may be attempts have been made to introduce insecticide-impregnated needed. Transmission control interventions should be used curtains and bed nets. In high-risk areas lacking any health selectively wherever they are affordable and can achieve sus- facilities, it may sometimes be appropriate to establish special- tainable results. ized malaria clinics. Vector control operations have relied overwhelmingly on spraying of residual insecticides. The effectiveness of residual MIGRANT AGRICULFURAL LABOR. Characteristics: cotton, spraying varies substantially with the biting and resting behav- sugar, and large-scale rice cultivation often require large con- ior of the mosquito vector, the type of housing, and the habits tingents of temporary labor for planting and harvesting. The of the people. A control measure which is receiving increasing workers generally live in crowded, unsanitary camps where attention is the use of insecticide-impregnated bed nets or mosquitoes abound and precarious shelters offer little protec- curtains. tion against the malaria vector. Because of heavy pesticide use Other techniques of vector control play a more restricted for agriculture, often sprayed by airplanes (especially in cotton role but, where indicated, may be highly effective. Space or farming), vector resistance to a broad spectrum of insecticides aerial spraying is seldom used and is rarely justified. Larvicides is common. are feasible only with easily identifiable breeding places. Tech- Control: disease control requires case management and the niques of source reduction, such as drainage and water man- application of residual pesticides, where possible, and in some agement, can be the measures of choice in urban and periurban cases aerial pesticide application is very effective. In addition, areas and economic development projects. They are normally personal protection measures, such as the use of insecticide- too costly, however, for widespread use. impregnated bed nets, can sometimes be applied. If irrigation practices allow, drainage, biological control, and other mea- Residual Practices from Malaria Eradication Programs sures to reduce vector breeding may be indicated. Many malaria control programs continue to depend on prac- DISPLACED POPULATIONS. Characteristics: sociopolitical dis- tices held over from the eradication era that require adjust- turbances (such as wars, unrest, famines) often create situa- ment. Indoor spraying of residual insecticides, the main Mlalaria 295 Table 13-2. Patterns Associated with Ecological and Social Conditions Specific occupationsl Major ecologicol conditions social conditions Plains and Highland Migrant Control valleys Forest and Seashore Agricultural agri- Displaced intervention African outside and forest desert and colonization Gold culture popula- savannah frica fnnge fringe coastal Urban of forest mining labor tions Management of clinical malaria Diagnosis and treatment + + + + + + + + + + Care of treatment failures + + + + + + + + + + Protection of pregnant women Chemoprophylaxis + - + - - - + Bednets and personal protection + + + + + + + + + + Vector control Residual spraying - Selective Selective Epidemic Selective Limited Selective - Limited Epidemic control control Fogging ULV - - - - + Limited - - Limited + Impregnated bednets or curtains + + + - + - + + Environmental control Drainage and source reduction - - - - + + - - + Larviciding - - - - + + - - + Biological control - - - - Limited + - - + Surveillance Epidemiological surveillance + + + + + + + + + + Monitoring epidemic risk - + + + - _ _ - + + Health education + + + + + + + Source: Authors. control activity of eradication programs, consumes a large part help in the diagnosis of the cause of fever, and the radical of present program budgets. In principle, coverage with resid- treatment of positives, when offered, may no longer be needed. ual insecticides should be complete and regular to achieve Reports to WHO indicate that 150 million slides are collected significant reduction or interruption of transmission. Today's each year, with an average positive result of 3 to 5 percent. spraying is seldom regular because most budgets do not provide Great efforts are made in some malaria programs to maintain enough insecticide to cover all cycles; they are rarely complete a network of laboratories staffed by reasonably trained micros- because people often refuse to allow continuous routine spray- copists, engaged most of the time in the diagnosis of ambula- ing. Therefore, more selective, targeted, and cost-effective use tory fevers. But those in charge of the programs do not feel of pesticides is needed. responsible for making competent microscopy available in the Many malaria control programs continue the practice of medical care establishments for the diagnosis and monitoring case detection as the main mechanism to diagnose and treat of treatment of suspected severe malaria cases; case detection malaria. This procedure, devised for the confirmation of the continues to serve mainly an epidemiological purpose and does disappearance of malaria during the consolidation phase of a not contribute to improve the quality of care delivered by the malaria eradication program, aims at the collection of a blood health services. slide from every fever case in the population by a system of The epidemiological services of the malaria programs were periodic house visits and the collaboration of all outpatient designed to confirm that malaria had been eradicated, and clinics of the health services. All fever cases are also given a were organized to achieve the direct confirmation that the single dose (presumptive) treatment to be followed by a full parasite reservoir had been eliminated. Indirect indicators of (radical) treatment, if the blood slide is positive, when the riskwereoverlooked.Asaresult,malariaprogramshaveapoor result from the laboratory becomes available, normally weeks record for early detection, let alone the prediction, of malaria or months later. epidemics. Whereas poorly organized general health services, The consequence of insisting on a thick blood film for every reporting abnormal increases of fever cases, detected a number fever case is that malaria microscopists are overwhelmed with of epidemics reported in the literature, the malaria program negative slides, the examination of which not only takes most case-detection mechanisms operating in the area detected no of their time but could also distract them and cause them to abnormality until months later, when slides had been exam- miss positives. The diagnosis is, most often, late, so it cannot ined, the results reported, and the reported cases consolidated 296 Jose A. Ndjera, Bernhard H. Liese, and 1effTev Hammer and analyzed at the center. Often these mechanisms are unable generates a large proportion of visits to health centers, how- to detect abnormal situations before they become large enough ever, this could argue for shorter distances to clinics in new to overcome the dilution effect of consolidated reporting, as areas of agricultural development (in relation to more stable practiced by centralized malaria programs. General health communities) because of the greater severity of the disease in services and local authorities are more sensitive and more these areas, even though the set of appropriate interventions likely to demand action in response to peripheral complaints, is the same for the two settings. Again, this is part of a much if encouraged to do so. larger problem. In general, the interventions involved with case management are not specific to malaria and include Interventions: Patterns, Cost-Effectiveness, and Choice protocols for the public health facilities; the regulation, taxa- tion, orsubsidization of drugs and private healthcare providers; The nature of the different scenarios has a strong effect on the and recommendations on care for these providers. Benefits choice of appropriate policies to combat malaria. Calculations accrue to the system from any of these interventions, but of cost-effectiveness need to be made in each specific circum- attributing them to malaria is misleading. stance. The value of calculating cost-effectiveness of interven- tions is to help policymakers make decisions about competing RETURNS TO SCALE. Costs per measure of outcome (deaths uses of resources. The steps in any such analysis are the follow- averted, discounted healthy life years gained) vary substan- ing: (a) identify the policy instrument which is actually under tially with the level of activity of the intervention. Certain the control of the decisionmaker; (b) determine the relation features of intervention programs are relatively fixed and between the policy instrument and the measure of outcome therefore independent of scale of operation (facilities, staff desired; (c) pursue the activity until the marginal effectiveness salaries in the short run), others are variable and proportional per unit of marginal cost falls to a level comparable to other to outputs, and still others rise more than proportionately with uses of funds. Each of these components is problematic and output. Assessing the (marginal) cost per unit of outcome each is sensitively related to the epidemiological pattern in achieved needs to be assessed in each context. At a global which decisionmakers find themselves. level, Molineaux (1988) speculates that there maybe "decreas- ing returns," because many early programs of malaria reduction INTERVENTIONS. Often, the set of available policy options is were recognized as having strong effects, whereas recent efforts clear. Still, there is sometimes confusion about what is actually have been more disappointing. controlled by the govemment. For example, chemoprophy- laxis for pregnant women and insecticide-impregnated bed PORTFOLIO OF INTERVENTIONS. For a number of alternative nets are included as control interventions. These are policies policy options in malaria control, there is good reason to which promote the use of techniques of control, such as expect diminishing retums to most single activities and, there- information, education, and communication activities (JEC) fore, to expect effective policies to entail a package of instru- for prenatal care (in conjunction with a protocol for drug ments. The cost of vector control activities will rise with prescriptions); a subsidy on the sale of bed nets; active distri- expansion as a result of decreasing densities of vectors and of bution of (free) materials; and an IEC campaign on appropriate people. Costs of case management operations will rise also with use. Strictly speaking it is these latter policies which should be decreasing frequency of cases and the eventual need for either evaluated on the basis of costs and effectiveness. public information or IEC campaigns, which are costly. Barlow Organizational, political, and social factors implicit in some and Grobar (1986) suggest that the great uncertainty sur- of the patterns define or impose limits on the appropriate rounding cost estimates argues that a combination of policies policies. For example, the policies involved with management need to be used in parasitic-disease control programs. This is of clinical malaria are specific to the structure of the general an analogy to financial management in which a "portfolio" of health care system. Costs associated with malaria are sensitive instruments should be used to reduce the risk of the entire to the organization of the health system. Mills (1987) com- program's resulting in failure. Here, we argue that for malaria, pares the costs of provision in vertical programs with those in at least, a combination of policies would be desirable even with integrated programs in Nepal and finds that the higher the accurate information, because ofdiminishing returns from any volume of cases, the more similar the costs of the two organi- one instrument. zational policies will be. In areas where caseloads are low, integrated programs can have substantial cost savings, because VECTOR CONTROL. The cost-effectiveness of residual spraying personnel can switch to other health needs as appropriate. varies substantially with the endemicity of the disease in the One decision conceming provision of treatment (for all locale and with the degree of intensity of use. In regions of low complaints, not just malaria) would be the density of location to medium endemicity, where either the elimination of the of health centers, or, more realistically, the location of new disease or a substantial reduction in prevalence in humans as centers. The difficulty in separating care for malaria from care a result of reductions in vector capacity is possible, the effec- for anyone who presents with fever is sharply defined in this tiveness of vector control may be high. It may also have case. Although probably the most important factor in the care thresholds or regions of "increasing returns" near levels at of malaria in endemic areas, such decisions cannot possibly be which eradication is possible. Generally, however, there are made with regard to malaria alone. To the extent that malaria good reasons to expect rising marginal costs associated with Malaria 297 increasing workers hired for spraying. With decreasing density protective measures or to influence their use of drugs. The of housing units and increased distance fromi facilities in urban protective measures include bed nets, perhaps impregnated or regional centers, costs per house protected will rise as more with insecticides, and modification of evening activities and person-hours will be needed for more remote areas. Similarly, clothing. For bed nets, a subsidy on their sale is also a possibil- small regions with high densities of the vector will be cheap to ity. Although bed nets themselves have been shown to reduce reach, whereas expansion to wider areas will become costly. disease,theprospectsforincreasingtheiruseremainunknown. Within a region, the cost per house protected will rise as a Some people (more educated or concerned) may be quite easy restilt of decreasing chemical effectiveness on the vector pool. to reach, but increasing usage of these measures to any great Refusal of populations to have their dwellings sprayed also extent is probably expensive. To the extent that behavior does make improvements on the margin more difficult. change, however, it can increase the effectiveness of antivec- Other techniques of vector control play a much more re- tor campaigns. stricted role. Space or aerial spraying is seldom used, except for More rational use of antimalarial drugs should help to slow urban epidemics or to interrupt short transmission seasons, and the spread of resistance, which has been encouraged by exces- is rarely justified on cost grounds. Larvicides are feasible only sive and inappropriate use. To the extent that behavior can be with easily identifiable breeding places and are thus of limited changed by public information campaigns, this effect can be use. Other source reduction techniques such as drainage and ameliorated. What this might cost and how much it might be land management techniques can have significant effects in worth require research geared to helping operations. mainly urban areas, planned human settlements, or economic development projects, but they cannot be a significant part of COST ANALYSIS FOR DECISIONMAKING. Most countries with widespread control operations. malarious regions have in place some institutions designed to address the problem. The activities which are undertaken by CLINICAL MANAGEMENT. The cost of chemotherapy depends these institutions depend partly on the local needs but also on on both the costs of treatment itself and the costs (both to the history, cultural acceptability, and political concerns. The health care system and to the individual) of getting the patient most practical use of cost information is its ability to assist the to seek treatment. For self-diagnosis (for over-the-counter managers of the local malaria public health facility to make purchases of chloroquine) or for spontaneous presentation at better incremental decisions in environments where they are a health care facility, these costs are generally quite moodest, constrained. When specific activities are proposed for a spe- although they depend on the availability ofdrugs and distances cific area, costs can be gauged relatively easily, because changes to facilities. For a target population of malaria sufferers who do in scale are not at issue. Incremental benefits can also be notseekhelpineitheroftheseforms,expansionofchemother- appraised with regard to the local epidemiology and institu- apy requires the use of information campaigns designed to tional and administrative conditions. Costing exercises in encourage people to seek timely care or active case-detection these cases can greatly improve allocation divisions by manag- methods, which are usually very expensive. ers. Good examples of this use is the work done by Kaewsonthi As to the effectiveness of drug treatment, to date, the use of and Harding (1986) in Thailand and by Mills (1987) in Nepal. chloroquine has been a remarkably effective and cheap method In these studies, comparisons between techniques of vector of dealing with the disease. People have been able to obtain the control and between vector control and therapy are made drug easily, and much treatment has taken place outside the clearer by careful costing procedures at local levels, and prac- formal health care system. One consequence of this may well tical recommendations for improvements are made. Mills, have been the speeding of the progress of chloroquine resis- for example, was able to suggest a reduction in active case- tance, that is, that marginal costs of chloroquine had been detection methods and an increase in malaria clinics (or other higher and increasing all along. The spread of chloroquine- treatment facilities), observing that either of these activities resistant malaria has changed the picture substantially. Not looked better than spraying. only do the drugs cost more, but they also require more profes- sional supervision; and some of them run into more serious Estimate of Cost-Effectiveness problems of compliance with drug regimens. At 1987 prices (for one adult course of drug treatment), with chloroquine, the From data in the papers by Barlow and Grobar (1986) and Mills costs are $0.23 per person; with sulfadiazine/pyrimethamine, (1987), we calculated the costs per year of life saved and $0.50; with mefloquine/sulfadoxim/pyrimethamine. $1.20; and cost-benefit ratios for a variety of countries (tables 13-3 and with quinine and tetracycline, $3.00. Drug treatment is likely to 13-4). The most striking feature of these numbers is their continue to be the principal antimalaria weapon. The possibility variability. Indeed, the differences between the studies are so of multiple resistance and the difficulty of extending suitable marked that it would be hard to make any generalizations about health care to more remote rural areas, however, are two rea- them at all. The costs per case prevented ranged from $2. 10 to sons why the cost of malaria control will continue to increase, $259 (in 1987 dollars), and the cost-benefit ratios, from 2.4 to if it is based on the indiscriminate use of antimalarial drugs. 146.Thehighercost-benefitfiguresmakemalariacontrolseem of utmost importance. The lower figures bring it into compe- PERSONAL PROTECTION. To some degree, the government can tition with many other government programs as well as with rely on public information messages to increase people's use of many estimates of the marginal deadweight loss from tax 298 Jose .A Njera, Bernhard H. Liese, and Jeffrey Hammer Table 13-3. Cost-Effectiveness Ratios in Malaria Control Cost per Cost per case death prevented averted Cost per discounted DALY saved ( 1987 (1987 with various case-fatality rates Source CountrItv Method dollars) dollars) 2% 1 % 0.5% Observed Barlow 1968 Sri Lanka Insecticide - 78 - - - 2.8 Cohn 1973 India Insecticide 2.10 - 3.6 7 14 - Gandahusada and Indonesia Insecticide 83-102 - 142-174 284-349 564-693 - others 1984 Hedman and others Liberia Vector control 14 - 24 48 95 - 1979 and chemotherapy Kaewsonthi and Thailand Vector control 27-74 - 46-127 92-253 183-502 - Harding 1984 and chemotherapy Mills 1987 Nepal V\ectorcontrol 1.30-172 - - - - 2.8-255 and chemotherapy Molineaux and Nigeria Vector control 259 - 443 886 1,759 - Gramicc ia 1980 and chemotherapy Orriz 1968 Paraguay Insecticides 60 - 103 205 407 - Walsh and Warren Developing Vector control - 990 - - - 34 1979 countries -ata not available Source. Barlow and Grohar 1989; Mills 1987; authors calculations. collection (the systematic Lindervaluation of costs of govern- The last four columns of table 13-3 contain calculations of ment sources). Part of the explanation of the wide range of the cost per discounted disability-adjusted life-year (DALY) variation is not very illuminating. Differences in data quality, saved for differing assumptions concerning the case-fatality the assumptions used in the analyses (for example, the estima- rate for those cases in which the study does not explicitly tion of mortality avoided), the definition of the relevant costs, present that value. The numbers are sensitive to this assump- the length of time studied, the discount rate applied, and the tion, much more so than to any other parameter in the DALY coverage and purpose of the original intervention account for calculation. Any attempt to calculate the cost per discounted much of this variation. As one example, in the Garki Project DALY saved by the program requires locally relevant estimates study (Molineaux and Gramiccia 1980), which generated the of case-fatality rates.' figure of $259 per case averted per year, the costs of the There are more important, systematic reasons, however, to extensive research and monitoring exercise which accompa- expect average costs per unit of output to vary substantially nied the intervention are included in the program costs. Sim- between studies: (a) differences in the ecological, epidemio- ilarly, some of the studies included administrative costs, logical, and social characteristics between areas; (b) wide whereas others used only the cost of materials. Some costs were variations over time within areas of the incidence and severity calculated on the basis of small pilot projects (Gandahusada of malaria; (c) variations in the organizational structure of and others 1984) and others on the basis of national efforts control programs; and (d) differences in the intensity of appli- (Barlow 1968). cation of the interventions being appraised. Table 13-4. Cost-Benefit Ratios in Malaria Control Source Country Method Cost Barlow 1968 Sri Lanka Insecticide 146 Griffit, Rampana, and Mashaal 1971 Thailand Chemoprophylaxis 6.5 Khan 1966 Pakistan Eradication program 4.9 Livandas and Athanassatos 1963 Greece Eradication program 17.3 Niazi 1969 Iraq Eradication program 6.0 Ortiz 1968 Paraguay Insecticides 3.6 San Pedro 1967 Philippines Eradication program 2.4 Democratic Republic of Sudan 1975 Sudan Control program 4.6 Source Barlow and Grobar 1936. Malaria 299 Priorities malariologists. Moreover, training for eradication was defi- nitely oriented toward the execution of the highly standardized Our perception of malaria has been changing rapidly over the program tasks and operations. The training of malariologists past decades. Malaria is not, as once was thought, evenly spread did not give them the epidemiological background needed to over the geographic areas in which it is prevalent. Instead, it adapt to changing situations, to solve problems, to manage un- is highly focal, primarily affecting the hardest areas to reach, certainty, or to adapt or change control methods and strategies. and it is intimately linked to development efforts such as To meet current needs and achieve sustainable control, it is agricultural development, road building, fiscal incentives, and essential to create the manpower needed and to reorient colonization projects. Furthermore, malaria, which once was human resources, not only to apply standard solutions to easily treatable with chloroquine, has reemerged as a new recognized problems, but also to identify and find a solution to disease called drug-resistant malaria. Drmg-resistant malaria future problems. In that way we may be able to avoid repeating has been spreading, and serious problems in treatment are the cycle described in 1927, in the Second Report of the becoming more and more common. Finally, parasite distribu- Malaria Commission of the League of Nations: "The history of tion has not remained stable, but there is a general increase special antimalarial campaigns is chiefly a record of exagger- and a general shift from the more benign tertial malaria caused ated expectations followed sooner or later by disappointment by P. vivax to the fatal tropical malaria caused by P. falciparutn. and abandonment of work." Taking into account the epidemiology of malaria at present, National training programs should be supported and coor- the prevalent trends in the past fifteen to twenty years, and the dinated to ensure, through technical collaboration between prevailing level of endemicity in Africa south of the Sahara, it countries, that all countries are able to do the following: is reasonable to believe that a considerable deterioration of the * Maintain acorpsofadequatelytraiiedprofessionalswith situation is to be expected before the end of this century, unless the necessary epidemiological expertise to understand the a more serious control effort is made. Even if vaccines, new drugs, or new insecticides are developed, in view of the time mra poblem atoat n required for their final testing in the field, they are unlikely to gramt ne situtions. have a significant effect on malaria in the 1990s. The most * Train and orient general health services staff in the critical activities that could accelerate the progress in malaria clinical management of malaria, recognition and treatment control can be summarized as follows: of severe and complicated malaria, monitoring of drug resis- tance, and collection and management of epidemiological * In countries of Asia, the Americas, and North Africa, in information. which organized malaria control activities have been carried * Develop appropriate training methods for nonprofes- Out for nearly three decades, the priority should be on sional workers and community health workers so thev can reassessment of activities. Replanning of programs must be better manage fever and promote personal protection and based on epidemiological analysis, and at the same time, an improved environment. necessary changes in the organization and administration of * Promote the development of curricula in schools of these programs must be implemented. medicine and schools of public health to include new strat- * In countries of Africa south ofthe Sahara, priority should egies of malaria control and to increase their ability to stay be given to the extension of coverage of population by the abreast of the latest information on the diagnosis and treat- health care system. At the same time, a nucleus of malaria ment of malaria. specialists should be trained and selective control programs started. This should allow realistic planning and im- The development of new or improved methods or materials plementation of malaria control activities. The im- for malaria control-in particular, antimalarial drugs and po- plementation of any control activity on a larger scale should tential vaccines-should continue to receive the highest pri- be preceded by epidemiological field study that would con- ority. This view, which is widely supported and, to an tribute to the better understanding of the local epidemiology important extent, has shaped the UNDP/World Bank/wHo Spe- of malaria. cial Programme for Research and Training in Tropical Diseases' (TDR) research priorities, was recently reconfirmed by Human Resources Development the National Academy of Sciences/Institute of Medicine re- port on malaria. At the beginning of the malaria eradication program a signif- icant effort was made to train the personnel needed for the Research program, but, as programs became staffed and because malaria was expected to disappear soon, technical people and, espe- The most damaging effect of the malaria eradication years was cially, professionals in medical and biological sciences became probably the neglect of malaria research and malariology's lack progressively scarce. It has been said that the global malaria of appeal as a career to young scientists and epidemiologists. eradicationprogramdidnoteradicatemalariabutdideradicate In the words of McGregor (1982 p. 126), "throughout the 300 Jost A. Ndjera, Bernhard H. Liese, andJeffrey Hammer world support for further research into malaria, even that Malaria control and malaria research have drifted apart over concerned with insecticides and chemotherapeutics, con- the years. While malaria control programs continued their tracted swiftly. Worse still, the apparently imminent demise of fight using an established set of tools, research institutions a once important disease removed the necessity for training moved off in search of new technological solutions, and both scientists in malariology. It took 10 more years and a war to were under increasing financial constraints. Support for ma- halt this tragic trend." The reawakening of interest in malaria laria research contracted during the eradication years research showed a marked bias toward new technological (McGregor 1982), even for research on pesticides and drugs, developments through laboratory-based research, mostly in but support has been revived. This renewed research effort, chemotherapy, immunology, genetics, and the genetic control however, has shown a marked bias toward laboratory research of vectors and the possible use of mosquito pathogens. and toward the development of new technologies in chemo- In particular, since 1976 the UN OP/World Bank/WHO Special therapy, immunology, genetics, the genetic control of vectors, Programme for Research and Training in Tropical Diseases and the potential use of mosquito pathogens. (TDR) has assumed a key role in coordinating and funding In some parts of the world, control programs and research malaria research. It has made malaria its first priority and institutions have developed a curious rivalry: programs are continues to provide technical drive toward development of almost defensively entrenched in the use of established new tools for control. methods, such as residual spraying, whereas researchers Actually, most available antimalaria interventions are far uncritically proclaim as alternatives what should be seen as from ideal, not only in effectiveness but in their suitability for complementary techniques. On the whole, researchers have incorporation into long-term policies or the everyday practices undertaken projects that are of little relevance to ongoing of peoples and communities. Moreover, many of them have control operations and the specific problems of control insti- lost much of their original effectiveness because resistant tutions. At the same time, control programs, which often strains of parasites or anophelines have developed. We must collect massive amounts of valuable information, have improve our understanding ofthe epidemiology of malaria atd lacked the capacity to select research priorities and carry out of problems such as parasite and vector resistance. We must research projects. improve the tools of epidemiological investigation so we can Given the present status of malaria and malaria control identify problems in the field, plan and evaluate potential programs, we recommend that priority be given to research in solutions, and more effectively target interventions to prob- the following areas: lems. We must understand and monitor social and economic Epidemiolog-. Research is needed to improve our epidemio- processes that may influence the epidemiology of malaria and logical tools and understanding and thereby improve our facilitate or hamper the effectiveness of potential control ability to identify problem areas and better target control measures. And we must come to understand how these pro- interventions. In particular we must understand and mon- cesses may facilitate the incorporation of malaria control in itervsoia an ecnoicu p e that may influenc developitig a health infrastructure.itor social and economic processes that nmay influietice the deeoing . healh inatr r e s w , , epidemiology of malaria and facilitate or hamper the effec- Funding for malaria control programs shrank when people tiessoptniacnrlmaue. began to recognize that malaria could not he eradicated, when tiveness of potential control measures. the "basic health services" strategy in developinlg a health Technolog-y. Research is needed to develop and field-test infrastructure did not succeed, and when no successfuil models new control technologies and new combinations of old developed for incorporating malaria control into the primary interventions in order to increase efficiency and cost-effec- health care strategy. Malaria and general public health services tiveness of control programs. exhibited a nearly universal reluctance to redefine their re- Organization and management. Research is needed on the sponsibilities toward the malaria problem. organization and management of control programs in order Research may provide new and improved technologies to develop more effective and efficient organizational struc- needed to extend the feasibility of control, may provide epide- tures and management processes. miological tools that improve efficiency of control, and may Health infrastructure. Research is needed on health systems show better ways to combine interventions in more effective to examine the potential and means for effective participa- and efficient strategies. But it is also necessarv that any new tion of the general health services in malaria control, in tool be validated in the field and field tested to determine its particular in epidemiological surveillance, diagnosis and applicability in disease control. It is important that researchers treatment, and community mobilization. Many countries try to find better ways to integrate new and old tools for have embarked on a process of decentralization of health control. It is important to use health systems research to find services that could impart negatively on the effectiveness the best ways to incorporate new methods of control into the of vertical malaria control programs. These countries, health infrastructure and deliver them to individuals and com- in particular, will need to move cautiously and study care- munities. And as part of the strategy of primary health care, fully the alternatives for increasing the capacity of the we must test ways to control malaria through research and general health services to assume new responsibilities in development. disease control. Malaria 301 Notes Gandahusada, S., G. A. Fleming, Sukamto, T. Damar, Suwarto, N. Sustriavu, Y. H. Bang. S. Arwati, and H. Arif. 1984. "Malaria Control with Residual Fenitrothion in Central Java, Indonesia: An Operations-Scale Trial Using 1. For purposes of presentation, the other parameters in the DALY calculation Both Full and Selective Coverage Treatments." Bulletn of the World Health were assumed to be twenty-nine discounted years gained per death averted, Organization 62:783-94. eight days of illness, and a 10 percent qualitv-of-life adjulstment for nonfatal Garcia J. C. 1981. "The Lazmess Disease." Histor- and Philosophy of the Life cases. Except at very low case-fatalitv rates, the calculations are quite insensi- Gcianc. 19. tive to large ranges in the assumed values of these three parameters. Scences 3:3-59. Greenwood, B. M., A. K. Bradley, A. M. Greenwood, P. Byass, K. Jammeh, K. Marsh, S. Tulloch, F. S. J. Oldfield, and R. Hayes. 1987. "Mortality and Morbidity from Malaria among Children in a Rural Area of The Gambia. References West Africa." Transactions of the Royal Society of Tropical Medicine and H'wgiene 81:478-86. Andreano, Ralph. and Thomas Helminiak. 1988. "Economics. Health, and Griffith, D. H. S.. D. V. Ramana, and H. Mashaal. 1971. "Contribution of Tropical Disease: A Review." In A. M. Herrin and P. L. Rosenfield. eds.. Health to Development." Interational Joumnal of Health Services 1(3):253-70. Economics, Health, and Tropical Diseases. Manila: University of the Philip- Hedman, P., J. Brohult, J. Forslund, V. Sirleaf, and E. Bengtsson. 1979. "A pines. School of Economics. Pocket of Controlled Malaria in a Holoendemic Region of West Africa." Audibert, M. A. 1984. Agn'icultural Non-Wage Production and Health Status. A Annals of Tropical Medicine and Parasitology 73:317-25. Case-Study in a Tropical Milieu. Aix-en-Provence: Universite d'Aix-en-Pro- Kaewsontht, S.. and A. G. Harding. 1986. "Cost and Performance of Malaria vence, Centre d'Economie de la Sante. Surveillance: The Patients' Perspectives." SoutheastAsianJournalofTropical Barlow, Robin. 1968. The Economic Effects of Malana Eradication. Ann Arbor: Medicine and Public Health 17:406-12. University of Michigan, Bureau of Public Health Economics. League of Nations, Malaria Comnsission. 1927. Pnnciples and Methods of Barlow, Robin, and L. 1M. Grobar. 1986. "Cost and Benefits of Controlling Antinsalanal in Europe. 2d General Report of the Malaria Commission. Parasitic Diseases." rHN Technical Note 85-17. Population, Health, and (CH/MAL/73, III Health 1927,111 5. Geneva. NUtrition Depar-tment. Woirld Bank. Washington. D.C. Litsios. Socrates. 1990. "Feasibility of Malaria Transmission Control: Eco- Bhombore. S. R., C. Brooke Worth, and K. S. Nanjundiah. 1952. "A Survey nomic Aspects." Paper presented to the 2d World Congress on Health oftheEconomic StatusofVillagersinaMalarious IrrigatedTract in Mysore Economics, Septemnber 11-14, Zurich. State, India, before and after DDT Residual Insecticidal Spraying." Indian Macdonald, G. 1950. "The Economic Importance of Malaria in Africa." Joumal of Malariolog 6(4):355-66. WHo/MAL/60,AFR/MAL/CoNF. 16. World Healtlh Organization. Geneva. Binol, K. 1983. "Transmigration and Health in Connection with Tropical MacGregor, 1. A. 1982. "Malaria: Introduction." BTinsh Medical Bulletin Disease in Indonesia." Southeast AsianJournal of Tropical Mediclie and Public 38:115-16. Health 14:58-63. Malik. 1. H. 1966. Economic Advantages of Anti-Malana Measures Amongst the Brohult, J., L. Jort:eldt. L. Rombo, A. BjOrkman. P. O. Pehrson. V. Sirleaf. and Rural Population. Publication 137. Lahore: Board of Economic Inquiry. E. Bengtsson. 1981. "The Working Capacity of Liberian Males: A Compar- ison) between Urban and Rural Populations in Relation to Malaria." Annalsmazon." of Tropical Medicine and Parasitology 75:487-94. Superintendencv for Public Health Campaigns, N inistrv of Health. Brasilia. Bruce-Chwatt. L. J. 1969. "Malaria Eradication at the Crossroads." Bulletin of Typescript. the Ne York cademyofMediine 4510):99-1012.Matoila,Y. G.,.G B. White, and S.A. Magayuka. 1987- "The Changed Pattern the Neu' York Academy of Medicine 45( 10):999-l01 2. ti ttmnlteattUabr of Malaria Endemicity and Transmission at Amani in the Eastem Usambara 1979. "Man against NMalaria: Conquest or Defeat. " Transactons of the Mountains, North-eastem Tanzania." Journal of Tropical Medicine and Hy- Royal Society of Tropical Medicine and Hygiene 73:605-17. giene 901 7_34 1985. Essential Malariolog-. 2d ed. London: Heinemann Medical Mils, Anne. 1987. "Economic Studv ofMalaria in Nepal: The Cost Effective- Books. ness of Malaria Control Strategies." Evaluation and Planning Center, Bruce-Chwatt, L. J.. and H. M. Archibald. 1959. "Malaria Control Pilot London School of Hygiene and Tropical Medicine. Project In Western Soikoto, Northern Nigeria: A Report on Four Years' Molineaux, Louis. 1985. "La lutte contre les maladies parasitaires: Le probleme Results." InProceedingsoftheSixthInternarionalCongressofTropicalMedicine du paludisme. notamment en Afrique." In J. Vallin and A. Lopez, eds.. La and HNgiene. 1958, Vol. 7. Lisbon: Instituto de Medicina. lutte contre la mort. Travaux et Documents 108. Paris: Presses Universitaires de Castro, Bonita. 1985. "Development of Research-Training Project in de France. Socio-economics of Malaria In Colombia." Report to TDR Programme. . 1988. "The Epidemiology of Human Malaria as an Explanation at Its World Health Organization. Geneva. Typescript. Distribution, Including Some Implications for Its Control." In W. H. Cohn, E. J. 1973. "Assessing the Costs and Benefits of Antii-malaria Pro- Wemsdorfer and 1. A. McGregor, eds., Malana: Principles and Practices of grammes: The Indian Experience." American JUurnlal of Public Health Malanology. New York: Churchill Livingstone. 63: 1086-96. Molineaux, LoLiIs., and G. Gramiccia. 1980. The Garki Project: Research on the Conly, G. N. 1975. The Impact of Malana on Economic Developmnent: A Case Epidemiology and Control of Malana in the Sudan Savanna of West Afnca. Study. Scientific Puiblication 297. Washington. D.C.: Pan-American Geneva: World Health Organization. Health Organization. N6jera,J. A. 1981. "La epidemiologia y losproblemasde la luchaantimalarica van Dine, D. L. 1916. "The Relation of Malaria to Crop Production." Scientific en las Americas." In Pan-American Health Organization: Malaria en las Monthly (November): 431-39. Amrncas., oPs Scientific Publication 405. Washington, D.C.: Pan-Ameri- Field, J. W. 1967. "Resistance to the 4-aminoquinolines in the Malaria can Health Organization. Infections of Brazil and South East Asia." wp/scG/l. Working paper pre- . 1989. "Global Malaria Situation." W'PR/MAL(l)/89.14. World Health sented to the World Health Organization Scientific Group on Chemother- Organization, Geneva. apy of Malaria, April 25-May 1, Geneva. Najera,J.A.,G. R. Shidraw,J. Storey, and P. E. A. Lietaert. 1973."Mass Drug FoIll, C. V., C. P. Pant, and P. E. Lietaert. 1965. "A Large Scale Field Trial Administration and DDT Indoor-Spraying as Antimalarial Measures in with Dichlorvos as a Residual Fumigant Insecticide in Northern Nigeria." Northem Savanna of Nigeria." WHO/MAiJ73.817. World Health Organiza- Bltletin of the World Health Organization 32:53 1-50. tion, Geneva. 302 Jose A. Ndjera, Bernhard H. Liese, and JeffreN Hammer Niazi, A. D. 1969. "Approximate Estimates of the Economic Loss Caused by Walsh, J. A., and K. S. Warren. 1979. "Selective Primary Health Care: An Malaria with Some Estimates of the Benefits of the MEP in Iraq." Bulletin of Interim Strategy for Disease Control in Developing Countries." New En- Endemic Diseases 2:28-39. glandJounial of Medicine 301:967-74. Ortim, J. R. 1968. "Estimate of the Cost of a Malaria Eradication Program." Wernsdorfer, W. H., and 1. A. McGregor (eds.). 1988. Malaria: PnncipLes and Bulletin of the Pan-American Health Organization 64:14-17. Practice of Malanologk. New York: Churchill Livingstone. Over and others. 1992. "The Consequences of Adult Ill-Health." In R. G. A. Wemsdorter, G., and W. H. Wemsdorfer. 1988. "Social and Economic As- Feachem,TordKjellstrom,C.J.L.Murray,MeadOver,M. A.Phillips.eds., pects of Malaria and Its Control." In W. H. Wemsdorfer and 1. A. The Health of Adults in the Developinig World. New York: Oxford Universitv McGregor. eds., Malana: Principles and Practice of Ma?aliology. New York: Press. Churchill Livingstone. Pampana, E. 1969. A Textbook of Malaria Eradicationi. London: Oxford Uni- WHHO (World Health Organization). 1957. Expert Cotnmittee on Malaria: Report versity Press. of the Sixth Session. Technical Report 123, Geneva. Quo, W. K. 1959. In "Malaria Information." MAL/INFORN/46. World Health . 1969."ReexaminationoftheGlobalStrategyofMalariaEradication." Organization, Geneva. Official Records of the World Health Organizarion 176:106-26. Russell, P. F.. and M. K. Menon. 1942. "A Malario-economic Survey in Rural . 1974. Expert Committee on Malana: Report of the Sixteenth Session. South India." Indian Medical Gazette 77:167-80. Technical Report 549, Geneva. San Pedro,C. 1967/68. "EconomicCostsand BeneficsofMalaria Eradication." . 1978. "Malaria Control Strategy." Report by the Director-General. Philippine Journal of Public Health 12:5-24. A31/19. Geneva. Sawyer, D. 1986. "Malaria on the Amazon Frontier: Economic and Social . 1986. Expert Comrmittee on MalaTia: Report on the Eighteenth Session. Aspects of Transmission and Control." Southeast Asian Journal of Tropical Technical Report 735, Geneva. Medicine and Public Health 17:342 -45. . 1992. "World Malaria Situation in 1990." \vHo Weekly Epidemniological Sawver, D., and Sawyer. D. 1987. "Malaria on the Amazon Frontier: Economic Records 22:161467:23:169-74. and Social Aspects of Transmission anid Control." CEDEPLAR. Federal Uni- WHO (World Healtlh Organization)/UNICEF (ULnited Nations Children's Funid). versity of Minas Gerais, Belo Horizonte. 1978. "Primary Health Care." Report of the International Conference on Sinton, J. A. 1935/36b. "What Malaria Costs India' Nationally, Socially, and Primary Health Care. Alma Ata, U.S.S.R., September 6-12, Geneva. Economically." Records of the Malaria Survey of India 5:413-89. Wilson, D. B. 1960. "Report on the Pare Taveta Malaria Scheme 1954-59," 1935/36c. "What Malaria Costs India? Nationally, Socially, anid Dar-es-Salaam. Economically." Records of the Malaria Survey of India 6:96-169. Wilson, J. F. and A. Alicbusan-Schwab. 1991. "Development Policies .1938.WhatMalariaCostsindia.GovernmentofIndiaHealthBulletin aisd Health: Farmers, Goldminers, and Slums in the Brazilian Amazon." 26. New Delhi. Working Paper 1991-18. World Bank Environment Division, Washing- Spencer, H. C., D. C. 0. Kaseje, W. H. Mosley, E. K. N. Sempebwa, A. Y. ton, D.C. Huong, and J. M. Roberts. 1987. "Impact on Mortality and Fertility of a Yumer, R. 1980. "InfluLence du statut socio-tconomique sur la morhidite Consmunity Based Malaria Control Programme in Saradidi. Kenya." Annals paludeennc: tn essai de mesure." Ph.D. diss., Universite des Sciences of Tropical Medicine and Parasirology 81 (supplement 1):36-45. Sociales, Faculte des Sciences Econtsmiques, Grenoble. Vaisse, D., R. Michel, P. Carnevale, M. F. Bosseno, J. F. Molez, P. Peelman, de ZUlueta, J. 1988. "Repoirt on a Mission to Madagascar." Travel report to M. T. Loembe. S. Nzingoula, and A. Zoulani. 1981. "Le paludisrne a World Health Organization, Geneva. Pltzsnodiumn falcparum et le ge'ne de la drepanocytose en Republque de Zulueta, J., G. W. Kafuko, J. R. Cullen, and C. K. Pedersen. 1961. "The Polpulaire dii Congo. 2. Manifestations cilniques du paludisme seIon d uut,J,G .Ktk,J .Cle,adC .Pdre.16."h Populairte dL etCongo 2e Manifestations chnique" Meduecaludme selon Results of the First Year ofa Malaria Eradication Pilot Project in Northem Ia parasi6mite et Ic genftype h6moglobinique." Madecine TropicaLe Kigezl (Uganda)." East African MedicalJournal 38(1 ):1-26. 14 Dengue (with Notes on Yellow Fever and Japanese Encephalitis) Donald S. Shepard and Scott B. Halstead Dengue, yellow fever, and the viral encephalitides (principally cent. Complication rates are extremely low for yellow fever Japanese encephalitis) are acute, sometimes lethal infections and dengue, but for JE, long-term sequelae include personality which, with regard to morbidity and mortality, constitute the disorders, reduced learning ability, gait abnormalities, and most significant arthropod-borne viral diseases of man. These severe incapacitating paralysis. These occur in 25 to 40 percent viruses are all in the same taxon and share important biological of surviving patients. similarities. Infections with wild or live-attenuated viruses In tropical Asia, DHF/DSS is almost exclusively confined to result in lifelong immunity. Each virus is mosquito-transmitted children under the age of fifteen years with a modal age of five and, therefore, typically causes diseases of place. Yellow fever to seven years. Male-to-female case and death ratios are ap- and dengue share the satne urban mosquito vector, Aedes proximately 1:1.2. Yellow fever epidemics in West Africa, aegypti; in some parts of Africa, related Stegomyia mosquitoes Ethiopia, and Sudan have involved children and young adults are more important for yellow fever transmission. Japanese with slightly more cases in males than females. Jungle yellow encephalitis (JE) is transmitted by the rice-paddy breeding fever in Latin America involves principally young, adult males Culex triwaeniorhynchus and related species, which are distrib- who work in or at the forest fringe. In China, under conditions uted across the Asian land mass from Japan through India. of high enzootic transmission, Japanese encephalitis is a dis- Arthropod-borne viral diseases are transmitted by injection ease of children, principally five years of age and under; in of infected saliva during the bite of mosquitoes which, after Southeast Asia, an area of intermittent transmission, children ingesting blood containing virus, have survived sufficiently up to fifteen years are vulnerable and, in South Asia, where long for the virus to mnultiply in tissues, including the salivary virus is transmitted episodically, persons up to fifty years of age gland. Following an incubation period in humans of a few days acquire encephalitis. to a week or more, illness begins acutely, usually with fever, The primary risk factor for acquiring any of these diseases is headache, myalgia, inappetence, and varying gastrointestinal living in areas where vector mosquitoes breed. Water storage symptoms. In its severe form, yellow fever evolves to an acute in houses and promiscuous disposal of modem industrial trash hepatitis, complicated by acute vascular permeability, heinor- permit Aedes aegypti breeding. Cutex tritaeniorhynchus breeds in rhage, and renal failure; dengue evolves to dengue hemor- wet rice paddies. Children, females, Caucasians, and Orientals rhagic fever/dengue shock syndrome (Duif/DSS), characterized are the populations most at risk for 1)1HF/DSS. There is no byacute vascular permeabiliryanid bleeding. Japanese enceph- evidence to suggest that age, sex, or the innate susceptibility alitis involves acute central nervous system disease, commonly of blacks plays a part in the case-fatality rates of yellow fever. with altered cerebration, coma, and paralysis. In all three, the Japanese encephalitis is more severe and has higher case- acute illness stage lasts for a week or more; severe findings and fatality rates in children than adults and in Caucasians and death or recovery occur promptly with yellow fever and blacks than Orientals. DHF/DSS, but prolonged incapacitation is a frequenit outcome of In the remainder of this chapter we discuss all these condi- Japanese encephalitis. tions, describinig their public health significance, economic Historically, case-fatality rates foryellow fever, denguehem- effects, opportunities for better case management and preven- orrhagic fever, and Japanese encephalitis have been as high as tion, and future priorities. We treat dengue in the greatest 80 percent, 50 percent, and 60 percent, respectively. Today, depth, presenting an empirical cost-effectiveness analysis. in Africa, case-fatality rates for yellow fever are thought to be Dengue was selected for this more detailed analysis, and some in the range of 0.5 to 6 percent; in tropical Asia, case-fatality original data were collected, because this condition poses the rates for DHF/DSS are 1 to 5 percent; and in continental and most difficult policy questions. Although there are current and South Asia, rates for Japanese encephalitis are 20 to 40 per- prospective technologies with considerable potential to con- 303 304 Donald S. Shepard and Scott B. Halstead trol the disease, their feasibility and cost in relation to the of male-to-female cases and deaths are 1.1:1. There have been competing demands remains an open question. dramatic increases in epidemic occurrences of Japanese en- cephalitis in India, Nepal, Sri Lanka, and Thailand in the Public Health Significance 1970s and 1980s. In this same period, JE has virtually disap- peared fromJapan, the Republic of Korea, and Taiwan (China) Trends in the spread of epidemics and the levels of disability as a result of widespread use of effective vaccines. Annual cases and death are the main public health concems. in China have decreased from 100,000 to 25,000. Morbidity and Mortality Levels and Trends, Circa 1985 Morbidity and Mortality in 2000 and 2015 Annually, junglevtellowfeverisresponsibleforabout200cases All three diseases are showing a tendency to increase abso- and 40 deaths in the tropics of the Western hemisphere, lutelv with increasing population, but at a rate inversely pro- principally in adult males. Intermittent rural and urban epi- portionate to prosperity. demics occur in Sub-Saharan Africa (West, Central, and East). Much of African experience with yellow fever is unreported. YELLOW FEVER. In Africa, there is increased risk of urban During the 1987 Nigerian epidemic in Oyo State, 805 cases yellow fever transmitted by Aedes aegypti. For the past forty and 416 deaths were reported, but surveys of three involved years, disease has predominantly affected rural areas, where it villages suggested that as many as 120,000 cases and 24,000 has been transmitted by other Stegomyia vectors. Attack rates deaths may have occurred (De Cock and others 1988). An- and the area of involvement in Africa will increase with pop- other outbreak in the same year in Niger State in northern ulation and inversely with gross domestic product per capita. Nigeria may have been of similar dimensions (Nasidi and others 1989). Forty percent of cases were children under ten DENGUE Unless vaccine or nationwide vector control pro- years; 30 percent were adults. The male-to-female ratio was grams are implemented, the absolute number of cases ofdengue 1.4:1. Yellow fever epidemics occur irregularly. The fifteen- will expand with population and growth of cities. Increases in year trend is toward increasing epidemic frequency, increasing gross domestic product per capita should reduce attack rates involvement of urban areas in N igeria, and extension to South through improved standards for residential dwellings. Nigeria. Attack rates of severe dengue illness have steadily increased. JAPANESE ENCEPHALITIS. Without incorporation ofJE vaccine In 1987, more than 600,000 cases of dengue hemorrhagic fever into the World Health Organization's (WHO's) Expanded Pro- with 24,000 deaths were reported from Southeast Asian coun- gramme on Immunization (EPI) in South and Southeast Asia, tries. N inety-nine percent of cases and deaths were in children attack rates will increase with population. under fifteen years. The male-to-female ratio is 1:1.2. In the Americas, dengue transmission has increased dramatically Economic Costs since 1963. There are four types of dengue, DEN 1, 2, 3, 4, and all four are now endemic in the Caribbean basin, and epidemic Costs include those of treating the sufferers directly and the dengue has occurred in all South American countries except indirect social costs. Argentina, Uruguay, and Chile. In 1981, DHF/DSS resulted in 116,000 hospitalizations in Cuba. Venezuela reported 50 YELLOW FEVER. Yellow fever entails the same kinds of costs deaths and 1,000 cases in a 1989-90 dengue outbreak, its first as those associated with dengue fever (see below): vector in many years (PAHO 1990). Brazil experienced thousands of control, diagnosis and outpatient treatment of mild cases, and cases in its first recent epidemic in 1986 (Schatzmayr and intensive care of the severely ill patients. Finally, there are others 1986; Secretaria de Estado de Sa6de e Higiene 1986) costs associated with loss of work of adults ill themselves or and had a second resurgence in 1991. Attack rates in Southeast attending children, and loss of life. Asia have increased from 15 per 100,000 to 170 per 100,000 during the period 1970-87. In Thailand and Viet Nam, attack DENGUE. Costs include those associated with vector control, rates in 1987 were 3,700 and 6,400 per 100,000, respectively. vaccination, diagnosis and outpatient treatment of mild cases Perhaps one-half of this increase is due to better case recogni- (which are ten to fifty times more common than reported tion and the fact that milder disease not reported earlier is now severe cases), and intensive care of the severely ill, including being reported (inflation of case identification). Nonetheless, intravenous fluids, blood or plasma transfusion, and poly- cases reported are from medical facilities and signify use of pharmacy, with average hospital stays of five to ten days for diagnostic and curative services. severe cases. Adults lose work to attend to children's illness. Cases ofjapanese encephalitis have been reported in China. Finally, there are costs associated with loss of life. mainland Southeast Asia, and South Asia. Annual cases and The literature contains no previous studies on the cost- deaths are estimated at 25,000 and 10,000, respectively, 70 effectiveness of dengue control. The literature on the eco- percent in children below the age of fifteen years. The ratios nomic consequences of dengue includes a study by Von Dengue (with Notes on Yellow Fever and Japanese Encephalitis) 305 Allmen and others (1979) from Puerto Rico, in which the Singapore dollars perperson per year in 1973 to 1974, or$1.36 economic cost of the island's dengue fever epidemic of 1977 is to $1.82 (Chan 1985). In 1988 prices, based on a 5 percent calculated. Included are direct costs for medical care and annual inflation, the amount is $2.69 to $3.60. In the early vector control measures and indirect costs for lost production 1980s, following a dengue epidemic, Cuba also embarked on a due to illness and absenteeism by patients and by parents caring program of environmental vector control at a total (not an- for sick children. The population was 3 million. Direct costs nual) cost of $6.00 to $10.00 per capita. ranged between $2.4 million and $4.7 million. Indirect costs ranged from $3.7 million to $10.9 million, with total costs of JAPANESE ENCEPHALITIS: The average hospital stay for per- the epidemic ranging between $6.0 million and $15.6 million. sons with JE is two weeks. Forty percent of survivors are physi- Expenditure on patient care and vector control measures is cally or mentally crippled and require one to five years considered to be in the range of 7.8 to 20.2 percent of the total rehabilitation; 10 percent of these require chronic care. expenses. Gubler and others have also studied this epidemic and Prevention estimated costs to be an order of magnitude higher, ranging between $100 million and $150 million, in medical costs, Some measures can be taken immediately, others await im- control efforts, lost work, and lost tourism since 1977 (D. J. proved technology. Gubler, personal communication 19 October 1992). Kouri and others (1989) have estimated the cost of the 1981 Lowering or Postponing Disease Incidence DHF/DSS outbreak in Cuba (with a population of 10 million) at $103 million. In this outbreak more than 1 6,000 persons were Elements of a preventive strategy are as follows: hospitalized within a little over 3 months. It is remarkable that in such a short period more than I percent of the Cuban YELLOW FEVER. Risk factors principally are overpopulation, population required intensive care in a hospital setting. In- rural to urban migration, vector prevalence, and inadequate cluded were direct costs for patient care and control of the domestic water supply or sewage disposal. There are two pre- vector of $41 million and $43 million, respectively, and indi- ventive strategies: (a) production, purchase, distribution, and rect costs, including lost production of$14 million and disabil- use of yellow fever vaccine; and (b) control or eradication of ity payments of $5 million. Aedes aegypti (in Africa, limited to urban vectors). The poten- Much lower direct costs were estimated for the 1980 epi- tial effectiveness ofeitherof these two strategies is l00percent. demic of DHF/DSS in Thailand, which included hospitalizations The current price of yellow fever vaccine (excluding costs of and deaths. Mosquito abatement costs and hospitalization administering the vaccine) is $0.20 to $2.00 per dose. costs, almost entirely for children, were $6.5 million (Matsurapas 1981; Halstead 1984). DENGUE. The strategy is the same as for yellow fever except Soper and others planned and executed with military-like vaccine development is in progress and outcome is not known. precision environmental vector control in Brazil, and the Dengue transmission can be interrupted by eliminating the efforts were replicated throughout the Americas (Soper and mosquito vector (Aedes aegypti or Aedes albopictus) which others 1943). Chan (1985) provides a thorough description, carries the virus. Two methods are possible: chemical con- including a cost analysis, of the Singapore vector control trol-killing adult mosquitoes by means of chemical insecti- program based on Soper's principles. The most important cides-and environmental control-elimination of sites for element of the program is source reduction-elimination of breeding of the mosquito which transmits dengue fever (Chan breeding sources for mosquitoes. Trained, uniformed public 1985). As Chan points out, the high fecundity of the mosqui- health officers are authorized to enter premises, inspect for, and toes means that they can quickly replace their population. destroy breeding sources. Destruction of breeding sources Chemical control must be repeated several times per year and, includes removing water-collecting refuse and sealing water even then, may be of limited effectiveness. storage containers. This environmental program is supple- Environmental control, though more difficult, appears to be mented, in times of epidemics, by chemical control-fogging far more effective (Chan 1985). Rubbish, such as old tires, premises that have or are near places that have high Aedes must be removed from the area; water storage vessels must be aegypti indexes. Public health education, primarily through covered and cleaned regularly; and the presence of the mos- pamphlets, seeks to motivate and teach the population to quito must be diligently monitored. These activities require eliminate breeding sites. During outbreaks, television, radio, initial capital costs to set up an infrastructure, educate the and newspapers provide additional publicity. Moreover, Singa- population about control measures, establish rewards and sanc- pore enacted the Destruction of Disease-Bearing Insects Act tions for implementing them, and train the necessary environ- (Act 26of 1968) to require thatpersons comply withdirectives mental control personnel. Recurrent costs are the costs of of the commissioner of health to eliminate breeding sources. operating this infrastructure. Violations are punishable by fines. Chan reported that the We collected original data on the costs of vector control in environmental (Aedes) control program cost three to four several countries. Thailand launched a large-scale effort, 306 Donald S. Shepard and Scott B. Halstead which, so far, has been unsuccessful. The cost-effectiveness of New Delhi, India. The planned live-attenuated tetravalent this strategy will depend on the extent to which control efforts dengue vaccine is likely to have a manufacturing cost of at least can be reduced following an initial success. Although good $10 to $20 per dose and require refrigeration during shipment data are not yet available, we will attempt in this chapter to and storage. It will have a very short shelf-life once rehydrated produce useful estimates. from the lyophilized product. In this respect, it will be similar One of the most important contributions to the eradication to yellow fever and measles vaccines. of the mosquito is the implementation of cleanup campaigns. Costs of the vaccine strategy include capital costs for vac- Organized primarily by the national or city agencies, vector cine development and operating costs for vaccine manufactur- control requires community support. For example, in Puerto ing and delivery. A tetravalent live-attenuated dengue vaccine Rico, cleanup campaigns are organized for an urban neighbor- will be expensive to produce, but delivery costs should not be hood with a population of about 50,000 people. Cleanup excessive because the vaccine will require only two doses and campaigns start with large public education campaigns to raise will be given primarily in cities. In all likelihood, the first dose awareness. These campaigns require the cooperation of the of dengue vaccine will be administered with measles vaccine leaders of each community, who work directly with state and and added to the existing infrastructure of the Expanded city officials. Householders agree to take responsibility for Programme on Immunization for children. cleaning each premise. Special teams are formned for public areas (parks, cemeteries) and difficult places (slums and JAPANESE ENCEPHALITIS. For Japanese encephalitis, one or junkyards). City cleanup workers provide trash bags, cleaning more live-attenuated vaccines are likely to be available by utensils, and pick-up trucks to collect garbage. A neiglhbor- 2000; genetically engineered vaccines are also likely to be hood campaign generally requires two to three weeks of prep- available. The reduction in cost of a genetically engineered aration and two to three days of trash removal activities. vaccine as compared with a live-attenuated vaccine will be In response to a dengue outbreak that peaked in June 1978, marginal, although a one-dose live-attenuated vaccine will Puerto Rico began the Anti-Dengue Program with funds from greatly decrease delivery costs. the Comprehensive Employment and Training Act. From August 1978 through September 1980, the number of workers Good Practice and Actual Practice increased from 300 to 900 under the Higienizaci6n Ambiente Fisico Inmediato program. The workers were paid the current Good practice is not always within reach financially, and minimum wage of $600 per month. Thus, the annual cost of actual practice may not always be effective. salaries for the clean-up campaign was about $4.3 million, or $1.30 per capita. YELLOW FEVER. Vaccine-induced antibody barrier is quite effective in preventing urban yellow fever in Latin America. JAPANESEENCEPHALITIS. Vaccine is the only preventive strat- In the seven or eight African countries in which it has been egy for combating Japanese encephalitis. The current regimen used, vaccination effectively controls yellow fever. Vector is three doses of killed, purified vaccine. Currently, it is given control in Africa is almost completely ineffectual. to children in Korea, China, Japan, and Taiwan (China). In South and Southeast Asia, the wholesale cost is $2.30 per dose; DENGUE AND JAPANESE ENCEPHALITIS. Except for those in in Southeast Asia, for children living in JE enzootic areas, two Cuba and Singapore, modern Aedes aegypti control programs doses are recommended. As yet, there is limited distribution in to combat dengue are in disarray. In contrast, excellent vac- South and Southeast Asia. cine programs to combat JE operate in Japan, Korea, and Taiwan (China), and a good program has been activated in Possible Changes in Preventive Technology China. No widespread use of vaccines exists in Southeast and South Asia. Some improvements in technology may be available by the year 2000, some not until 2015. How Much Should Reasonably Be Done? YELLOW FEV'ER. For yellow fever, an improved vaccine is not No countries have had the opportunity to examine health anticipated by the year 2000. Improvements in vaccine pro- investments in relation to projected costs to the economy of duction technology and increased production in developing yellow fever, dengue, or Japanese encephalitis. Yellow fever countries could reduce the price and improve efficiency at and JE, which involve adults or result in prolonged incapacita- delivery. tion, respectively, tend to make headlines and create hysteria. This has been the principal reason for government action in DENGUE. For dengue, a safe and effective genetically engi- the past. Fear, political pressure, and the technical capacity of neered vaccine is not likely by the year 2000, but it is likely by the society for vaccine production or vector control have 2015. To date, research on development of a vaccine has been dictated the actions adopted. performed in Thailand, supported principally by the Southeast It is likely that domestic production of JE vaccine in Thai- Asia Regional Office of the World Health Organization in land and India would result in purchase and use of the product, Dengue (with Notes on Yellouw Fever and Japanese Encephalitis) 307 whereas continued dependence on imported vaccine will re- Structure of the Model sult in temporization in adopting a national vaccination pol- icy. Eradication of Aedes aegypti throughout the entire Western In this section we seek to quantify the cost-effectiveness hemisphere currently offers the only preventive strategy for of dengue control over the long run in those areas of the control of dengue and yellow fever. world at risk of the disease. As mentioned above, currently two strategies are available to control this disease-improved Case Management case management and vector control. In the future, a third strategy-vaccinations-may also become available. As This section discusses opportunities there may be for imp- preventive strategies, vector control and vaccinations (if rovements in case management. and when available) would reduce the incidence of disease and thus reduce both morbidity and mortality. Case manage- Dengue ment primarily reduces mortality, with a small benefit in morbidity. Palliation is the objective of medical intervention of DHF/DSS, Combinations of strategies are also possible. Case manage- which is characterized by loss of fluids intemally through leaky ment may be combined with either chemical or environmental capillaries and occasionally, severe hemorrhaging. Intensive vector control. In addition, case management, vector control, hospital care is required and can successfully reduce the case- or both may be combined with vaccinations (if and when a fatality rates of DHF/DSS. Management of the leaky capillary vaccine is available). The costs of vector control or vaccina- syndrome is complicated. In some cases treatment with fluid tion are not affected by other strategies. The cost of case or fluid and plasma is useful, in other cases whole blood. management, however, is reduced by the presence of one or Incorrect treatment can lead to heart failure and a substantial more of the preventive programs because the costs of case risk of mortality. To improve case management and reduce management depend on the number of cases. Vector control case-fatality rates, fundamentally soundly educated physicians or vaccinations reduce the number of cases. and nurses are required, modem state-of-the-art and reliable The effectiveness of combinations were calculated accord- laboratory facilities are essential, and adequately functioning ing to the principle that each control strategy eliminates a pharmacies and a safe and resourceful blood supply system are certain proportion of the deaths still remaining after other required. Resources involved are capital resources for training strategies have been applied. That is, the effectiveness of a of personnel and rehabilitating facilities and equipment and combination of strategies is the product of the effectiveness subsequently for increased operating costs of the maintenance fractions of each. of these facilities and equipment. In addition, good managers are needed to ensure that the facilities, equipment, and per- Dengue Epidemiological Scenarios sonnel remain available at optimum preparedness. Realistic levels of tumover must be included. Studies of the cost-effectiveness of disease control must begin Theoretically, such improvements in case management can with prognoses of evolution of disease in the absence of any be costed and analyzed as if they were dedicated solely to the control measure. For dengue, these prognoses vary widely, treatment of DHF/DSS. That is, we could calculate the cost of an according to conditions for dengue transmission and previous education program solely for DHF/DSS, the costs ofstrengthening population exposure to one or more dengue viruses. Four of laboratories, pharmacies, and wards solely for this condition. different epidemiologic scenarios are possible. They are listed In practice, such a program might be undertaken to strengthen below, in order of increasing severity. case management for other infectious diseases and would entail training, rehabilitation of facilities, and the like for several * Endemic dengue fever, in which disease is relatively infectious diseases simultaneously. The cost and effectiveness silent except for dengue fever in young adults. Children are would be greater than for treatment of dengue, and economies seen in doctors' offices with mild fevers. The situation in of scale may be realized. Brazil in 1987 through 1989 and in Puerto Rico during the past decade illustrated this scenario. Yellow Fever and Japanese Encephalitis * An epidemic of dengue fever occurring in a largely susceptible population. This results in high morbidity in Intensive hospital care is also required for yellow fever. For adults, absenteeism, loss of tourism, some hospitalization, a Japanese encephalitis, palliation is necessary in addition to handful of hemorrhagic cases, and deaths. Brazil's outbreak intensive hospital care, which may be followed by prolonged in 1986 illustrated this situation. physical rehabilitation or even institutionalization. * An epidemic of DHF/DSS occurring for the first time. Such an epidemic results in high morbidity and mortality in Cost-Effectiveness of Dengue Control children and adults. This is a one-time-only occurrence and not a stable state. Examples are the Cuban epidemic of 1981, We look at different combinations of factors that may affect in which half of deaths and cases were in children, and the overall costs of dengue control. Venezuelan epidemic of 1989-90. 308 Donald S. Shepard and Scott B. Halstead * Endemic DHF/DSS. In this scenario there is continuous The country is important because the levels of development high morbidity and mortality, limited to children. The of the health delivery system vary widely among nations. For situation in Thailand is an illustration. one of the interventions in the model, case management, benefits depend critically on the level of sophistication of the In establishing a potential scenario for calculating the costs health delerytsystem In thi anal we haveicategorite of dengue, it seems most appropriate to choose the endemic health delivery system. In thes analysos, we have categorvzed steady state of DHF/DSS (the last of the four epidemiologic d" (or uel scenarios). This situation results when there is unlimited abun- ope venly developed). Developed systems are ones dance of Aedes aegypti. It is the most extreme scenario and the which meet five criteria: (a) most of the population has access one which control is designed to avoid. Because present evi- to quality primary health services; (b) the population is suffi- oene whichst con olytrolpse igned to i avod. Be pcaus pr nesentca ciently~ sensitized to acute problems such as dengue that a dence suggests that only tropical Asia and tropical America sev,erelyt ill child will receive medical care within twelve hours; are at risk of DHF/DSS, this cost-effectiveness study is targeted to s hypothetical populations in these regions. Operationaly, we .(c) personnel in primary and first-level referral facilities are hypothetical populations im these regions. Operationally, we ,. have defined these regions as all of Central and South America sufficiently trained that they can generally stabilize an acute and the Caribbean, and South and Southeast Asia. The cre illness and refer a case for definitive care when needed; (d) an part of these regions contains 2.22 t illion people (420 million acutely ill child can reach a secondary health facility within prth Aofrthes r ns 1c n billion pep Ale (4 2 [mrIllo twelve hours; and (e) referral facilities have the technical Bank 1990t). development, personnel, and equipment to perform current treatments safely and effectively. Health systems without this The calculateonat procedure o a the model IS concerned woth capacity are termed "not developed." Because of the level of the aggregate population of a country at risk. In the analysis development of the health system required to implement ef- the vaccination strategy, the fact that a vaccine confers benefit fevelpm aent for deiu-e to trategh ed only to the extent that it is used is taken into consideraicon tective case management for denlgue, this strategy has proved only uo the extent that it is used is taken into consideraion. feasible only in countries with strong health delivery systems. Thus, following a model developed to aid in the analysis ot For example. improved strategies of case management have vaccination programs (Shepard and others 1986), we multiply been implemented in Thailand during the past thirty years. the efficacy of the vaccine and the coverage to arrive at the During this period the case-fatality rate from dengue hemor- effectiveness of a vaccination program. For the purposes of the rhagic fever has fallen from 5.8 percent in 1958-65 to 0.5 model, "coverage" means the correct administration of a vac- cie Ths th wor inopoae fatr of digosi acu percent in 1986-89. In general, children are promptly referred cine. Thus, the word incorporates factors of diagnostic accu- in emergencies. Physicians with the equivalent of United racy, provider compliance, and patient compliance, which are States specialty training who have adequate laboratory back treated separately in some other studies.Sttssealytangwhhvedquelbotryac treated separathelyinsomSe oethersttudiesinthe. ealth r up are on duty in a pediatric intensive care unit twenty-four As with other cost-effectiveness studies in the Health Sector hours per day. Nursing staff are skilled in managing pediatric Priorities Review, the model applies to a hypothetical popula- emergencies and inserting intravenous lines to rehydrate chil- tion of one million persons of all ages in a country at risk of dren in emergency with minimal risk of infection. dengue. The model first estimates baseline results for costs and In other countries in southeast Asia, such as Myanmar, health effects if no control strategy is applied. It then estimates Cambodia, Lao People's Democratic Republic, and Indonesia, results assuming individual or combined strategies are applied, the overall level of development of the health delivery system All economic data for the model are expressed in constant 1988 U.S. dollars. The model uises fully allocated costs, rather does not meet the criterla we have listed above. Although a than marginal costs, for all inputs. This method is appropriate few centers provde excellent care, success wdth improved case because costs are being considered over the long runi in manv management has not been achieved on a national scale. Al- though this discussion of applications is based on countries, countries; results are being used to inform policies that are future extensions of it could consider regional policies based concerned with the creation or dismantling of whole programs, on variations within a count in which marginal considerations may not apply. The data also A country's mortality rate of children under five years of age needl to be comparable with companion cost-effectiveness Acutysmraiyrt fcide ne ieyaso g stud ies. serves as a good proxy for the level of development of its health Theumainmeasureofhealthbeneitsaredisability-adjusted system. We would expect that most countries which the life-yar measu in thea standard populati of one million. United Nations Children's Fund (UNICEF) characterizes as hav- lTie-years cDALYS bine stanard population of one millnlon ing "middle" or "low" under-five mortality rates (70 or less per This measure combines a loss in life expectancy and in qualhty 1,000 live births) would have strong health systems, whereas of life as a result of dengue. Future costs and health benefits are ' c both dscountd at arate o 3 perent pe yearmost countries w ith "high" or "very high" rates (greater than both discounted at a rate of 3 percent per y'ear. 70) probably have variable health systems. Thailand and In- donesia had under-five mortality rates of 34 and 97, respec- Feasible Applications in Each Setting tively, as of 1990 (Grant 1992), so their rates are consistent with this classification. In any meaningful application of the model, only potentially The time frame of an application is important because it feasible interventions should be included. The designation of determines the status of vaccine development. We have char- which interventions are potentially feasible depends on the acterized vaccine status as either "available" or "not available." country and time frame in which the application is set. As of 1992, dengue vaccines appear ready for final testing and Dengte (with Notes on Yell FeVer and Japanese Encephalitis) 309 development by a vaccine manufacturer. Nevertheless, no To evaluate the alternative policies, a cost-effectiveness vaccine is currently available for general use. It is assumed in model was applied with the best available data from the liter- the "vaccine available" case that current development efforts ature, case studies in Puerto Rico and Brazil, and subjective are successfully completed and that a vaccine for mass use is estimates. The cost-effectiveness model is specified in detail in available. If current efforts continue successfully, this would be the three appendixes to this chapter. about 1997. Up to four single interventions are considered in this cost- Results effectiveness model: * Case management improved (c) Using the model described above, we projected the results of * Immunization against dengue virus (i) applying each policy to a population of one million people. * Vector (the Aedes aegypti mosquito) chemically con- The results are expressed in the cost and benefits per year of trolled (v) application in figures 14-1 through 14-5. The benefits are * Environmental vector control (E) expressed either in deaths averted or disability-adjusted life- years (DALYs) saved. Baseline data are for the absence of any Because the two types of vector control would be duplicative, control policy and are considered to be those with no costs and they are considered exclusive. Otherwise the single interven- no health benefits. In table 14-2 we show the results for all the tions can be combined up to three at a time. dengue control strategies. Here all the policies (both single and The combinations of development of the health delivery combination) are listed alphabetically, regardless of their fea- system and availability of vaccine create four settings or cases: sibility in a particular setting. no vaccine in a developed and in an undeveloped health Examination of the data in table 14-2 shows that the inter- system and available vaccine in a developed and in an unde- ventions in the combined policies interact nonlinearly. For veloped system. In table 14-1 we show the various policies example,thecombinationvcavertsfewerdeathsthanthesum available in each setting. Even in the most constrained setting of v and c. The cost is also somewhat less than the sum of the (no vaccine in a developed system), more than one policy is costs. This is so because any preventive strategy, such as v, available. In the most inclusive setting (vaccine available in reduces the number of cases requiring treatment. Thus, the an undeveloped system), ten choices are possible. benefit and added costs from better treatment are both less In the analysis of the cost-effectiveness of potentially feasi- than when there were more cases. ble alternatives, we consider two criteria: dominance and The results of applying the model in each of the four settings relative cost-effectiveness. The dominance criterion means are shown in graphic form in figures 14-1 through 14-5. The that some interventions or combinations can be eliminated in main part of the analysis, summarized in the first four figures, some settings because they are inferior to another interven- presents the results in cost per disability-adjusted life-year. In tion (or mixture of interventions) on both costs and effective- these figures, each feasible policy (a single intervention or ness. The relative cost-effectiveness criterion indicates the combination of interventions) is denoted by a square or a dot. important policy tradeoffs between resources allocated to The letters above the square or dot are the label for the policy, dengue control and results. These concepts will be clarified and as described in table 14-2. All the figures begin with the displayed graphically in the context of specific numerical baseline (no control), which entails zero cost and zero health results below. benefits. Table 14 1. Interventions for Dengue Control in Developed and Undeveloped Health Systems Vaccine availability Developed health system Undeveloped health syste7n No vaccine Vector chemically controlled (v) Case management improved (c) Environmental vector control (E) Vector chemically controlled (v) Environmental vector control (E) Environmental control and case management (EC) Vector chemically controlled and case management (VC) Vaccine available Immunization (i) Case management improved (c) Vector chemically controlled (v) Immunization (I) Environmental vector control (E) Vector chemically controlled (v) Immunization and vector chemically controlled (iv) Environmental vector control (E) Immunization and environmental control (IE) Environmental control and case management (EC) Vector chemically controlled and case management (vc) Immunization and case management (IC) Immunization and vector chemically controlled (IV) Immunization and environmental control (IE) lmmunization, case management, and environmental control (ICE) Source: Authors. 310 Donald S Shepard and Scott B. Halstead Figure 14-1. DALYS Saved without Vaccine Available impairment. By comparison, the regular (undiscounted) life in Undeveloped Health System expectancy at age 6, calibrated to areas at risk of dengue, is 63.8 (per I million populati(on) years. In figure 14-1, only three policies are shown: BASE, V, and E. Cost per year (millions of dollars) These are the only feasible policies in countries without a 3.0 developed health system and without a vaccine available. The 2.8 position of policy E at the right of the graph shows that it is the 2.6 most effective of the feasible policies in this setting. The fact 2.4 E that it is also the highest on the vertical axis indicates that it 2.2 [ is also the most costly policy. The slope of the line segment 1.8 - from the baseline to the first policy (v) corresponds to the 1.6 cost-effectiveness of that policy, in average cost per DALYsaved. 1.4 - According to table 14-2, this cost-effectiveness ratio is $1,992. 1.2 - It is the ratio of the net cost of that intervention (approxi- 1.0 - mately $435,000) divided by its effectiveness (219 DALYs 0.6 8 saved), also shown in table 14-2. 0.4 - Chemical vector control is technically a more cost-effective 0.2 policy than the altemative of environmental control, which B 400 6 8 0 costs $3,129 per DALY, because its cost-effectiveness ratio is lower. That is, a given amount of money can buy more DALYs DALYs saved per year if spent on v rather than E. An ideal policy would fall in the lower right corner of this n Efficient policies graph-substantial health benefits and minimal costs. A poor policy would lie in the upper left corner-few benefits but high Note: BASE = Baseline; Efficient policies: v= Vector chemically controlled, costs. The frontier of current efficient policies, shown by the E = Environmental vector control. solid line, is obtained by connecting those currently available Source: Authors' cost-effectiveness model policies for which no other policy is closer to the lower right comer. Thus, the baseline, v, and E form the frontiers of current In calculating DALYs, each death averted was 25.5 dis- efficient policies. counted years, which becomes 25.5 DALYs. No quality adjust- Figure 14-2 is an analogous graph for the situation in which ment was required for deaths, because if a person survives an no vaccine is available in a developed health system. Improve- episode of DHF/DSS, he or she will not have any long-term ment in case management is a feasible intervention, both alone Table 14-2. Efficacy and Costs of Interventions for Dengue (per I million population) Deaths _ DALYs saved froni Net cost Average cost Average cost Intervention averted Mortalir Morbidit\' Total Percent ($000) per DALY($) per death ($) Baseline (BASE) 0.00 0 0 0 0 0 0 0 Casemanagementimproved(c) 21.74 554 3 557 92 327 587 15,042 Environmental vector control (E) 22.52 574 118 692 95 2,172 3,139 96,461 Environmenital control and case management (EC) 23.61 602 118 720 100 2,189 3,040 92,712 Immunization (1) 16.44 419 86 505 69 727 1,440 44,251 Immunization and case management (IC) 23.10 589 87 676 97 828 1,224 35,827 Immunization, case management, and environmental control (ICE) 23.68 604 123 727 100 2,959 4,071 124,968 Immunization, case management, and vector chemically controlled (ICV) 23.28 594 98 692 98 1,250 1,806 53,691 Immunization and environ- mental vector control (IE) 23.34 595 122 717 98 2,954 4,117 126,537 Immunization and vector chemically controlled (iv) 18.62 475 97 572 79 1,180 2,062 63,372 Vectorchemicallycontrolled(v) 7.11 181 37 219 30 435 1,992 61,234 Vector chemically controlled and case management (vc) 22.33 569 39 609 94 664 1,091 29,754 Source: Authors' cost-effectiveness model. Dengue (with Notes on Yellow Fever and Japanese Encephalitis) 311 Figure 14-2. DALYs Saved without Vaccine Available v. In technical terms, a combination of BASE and c dominate in Developed Health System policy v. This dominance is shown graphically by the fact that (per I million population) the line from BASE to c passes underneath the dot for policy v. In these figures, efficient strategies are shown by squares, Cost per year (millions of dollars) whereas dominated policies are shown by dots. Strategy v is 3.0 not economically efficient because this strategy was considered 2.8 only 30 percent effective. Because the dengue-carrying mos- 2.6 quito breeds quickly, populations reduced by chemical spray- 2.4 EC ing have been found to return quickly. 2.2 E* The technique of incremental cost-effectiveness analysis is 2.0 - useful to illustrate the tradeoff between cost-effective and 1.8 effective policies. In table 14-3 we present this analysis in 1.4 - tabular form. The tradeoff is judged by the number of addi- 1.2 - tional DALYs gained in relation to the additional cost incurred. 1.0 / The ratio of the additional cost to the additional gain in DALYs 0.8 - vc is the incremental cost-effectiveness ratio. For example, the 0.6 v c comparison between policies ve and c show an incremental 0.4 2 gain of fifty-one DALYs at an incremental cost of $337,000. The 0.22 0 incremental cost-effectiveness ratio is $6,568, corresponding BASE 200 400 600 800 to approximately $337,000 per fifty-one DALYs. Graphically, DALYs saved per year this ratio corresponds to the slope of the line segment from c to vc. This line segment is substantially steeper than that from BASE to c, showing that the cost to save each of these few * Efficient policies * Inefficient policies additional DALYs is quite high. In common parlance, it illustrates the decreasing marginal returns of larger invest- Note: BASE = Baseline; Efficient policies: c = Case management improved, ments in dengue control while holding the population fixed. vc = Vector chemically controlled and case management, EC = Environmental control and case management; Inefficient policies: v = Vector chemically controlled, Only efficient policies are listed in table 14-3, because they are E = Environmental vector control. the only ones to which incremental cost-effectiveness applies. Source: Authors' cost-effectiveness model and combined with other interventions. The efficient policies Figure 14-3. DALYs Saved with Vaccine (c, vc, and EC) are shown by squares. These results show that in Undeveloped Health System case management is now the most cost-effective policy. It saves (per I million population) one DALY for $587, about a third of the cost of achieving this Cost per year (millions of dollars) benefit with policies v or E, respectively. Case management is also a powerful policy, being able to save 92 percent of the 3.0 IE morbidity imposed by dengue in the cohort. It must be empha- 2.6 - sized, however, that this seemingly attractive policy is feasible 2.4 - because of and is dependent on a developed health system. 2.2 E Policymakers often face the choice between the most cost- 2.0 - effective policy and the most effective policy. In the case of 1.8 - dengue control, an efficient policy that would yield still greater 1.6 _ benefits is a combination of chemical vector control and case 1.4 iv management (vc). As shown in table 14-2, policy vc saves 609 1.2 1.0 DALYs, compared with 557 saved by policy c. At a cost of$ 1,091 0.8 per DALY, policy vc is somewhat less cost-effective than policy 0.6 V c. The next step in effectiveness is to replace chemical by 0.4 environmental vector control as an addition to case manage- 0.2 1 l ment (policy EC). The number of DALYs gained by this policy BASE 200 400 600 800 (720) is virtually the entire burden of dengue, but the cost for the cohort of one million persons ($2,172,000) would be DALYs saved per year substantial. m Efficient policies * Inefficient policies Although policy v is still feasible in the case of no vaccine in an undeveloped health system, economically it is no longer A. pata aplcto of thecae angemntstt Note: BASE= Baseline; Efficient po/icies: F = Immunizationys = Immunization efficient. A partial application of the case management strat- and vector chemically controlled, IE = Environmental vector control, egy to part of the population of one million persons could E = immunization and environmental control; Inefficient policies: V= Vector chemically controlled. achieve the same benefit in DALYs at a lower cost than policy Source: Authors' cosl-effectiveness model 312 Donald S. Shepard and Scot B. Halstead Figure 14-4. DALYs Saved with Vaccine assumed high cost of the vaccine itself ($17.50 per dose) and in Developed Health System the need for two doses. The relatively high cost of vector control arises because vector control must be practiced for the Cost per year (millions of dollars) entire population every year, whereas case management affects 3.0 IE _only sick patients. 2 8- IE A sensitivity analysis for vaccination showed immunization 2.6 would become as cost-effective as case management if the cost 2.4 of the series were to drop to $18.00. Allowing $3.07 (in future 2.2 E * Ec value) for the two vaccination contacts, as assumed in the base 2.0 case, this would leave about $7.00 per dose for the vaccine 1.8 itself. The dramatic drop in the price of hepatitis B vaccine 1.6 1.4 illustrates that such a drop is possible. Initially introduced at a 1.2 IV CV prohibitive price of $100.00 per dose, a plasma-derived hepa- 1.0 l titis B vaccine is now available for only $1.00 per dose for bulk 0.8 vc IC purchase by developing countries. A sensitivity analysis on 0.6 v case management showed that if the base cost of hospitaliza- 0.24C tion episode (TREAT) rose from $200.00 to only $438.00, it 0.2 would no longer be the most cost-effective strategy. BASE 200 400 600 800 The analysis for the case in which vaccine is available in a DALYs saved per year developed health care system was also calculated in cost per death averted, as shown in figure 14-5. Although the numbers * Efficient policies v Inefficient policies change, the demarcation between dominated and efficient policies, and the ordering among the efficient policies remains Note: BASE = Baseline; Efficient policies: c = Improved case management, the same. The advantage of policy c over policy I in cost- ic = Vaccination and case management; icv = vaccination, case management, and vector chemically controlled, EC = Environmental vector control and case effectiveness is seen more dramatically in deaths averted, management, ICE = Vaccination, case management, and environmental vector control; Inefficient policies: v = Vector chemically controlled, I = Immunization. vc = Vector chemically controlled and case management, IV = Immunization and vector chemically controlled, E = Environmental vector control, lE = Immunization and environmental control. Figure 14-5. Deaths Averted with Vaccine Source: Authors' cost-effectiveness model in Deqveloped Health System (per I million population) Figure 14-3 introduces the case in which a dengue vaccine Cost per year (millions of dollars) is available, but the health system is still not developed. As explained in the appendix 14C, a dengue vaccine is expected 32 8E, ICE to be 95 percent effective in protecting persons who receive it. 2.6 In a population, however, its effectiveness is limited by a 2.4 EC coverage of only 73 percent, the rate obtained for the third 2.2 E dose ofDPTaccording to 1990 UNICEFdata. As of 1992, a vaccine 2.0 is expected to be available in three to five years, assuming the 1.8 final development continues as planned. For this analysis, we 146 have taken the longer estimate, giving a target date of 1997. 112 ICV Immunization is the most cost-effective strategy, followed by 1.0 IV, E, and IE. Again, strategy v is not economically efficient. 0.8 The incremental cost-effectiveness analysis comparing poli- 0.6 cies E and IE shows that adding immunization onto environ- 0.4 v mental control is not particularly cost-effective. The 0.2 incremental cost per DALY gained is $30,927. BASE 4 8 12 16 20 24 In figure 14-4, we present the analysis for the case in which the full range of alternatives is available. A vaccine is available Deaths averted per year and the health system is developed. Again, case management * Efficient policies . Inefficient policies is the most cost-effective strategy. It is interesting that none of the preventive single inteTventions-immuinization, chemical Note: BASE = Baseline; Efficient policies: c = Improved case management, vector control, or environmental vector control-was eco- vc = Vector chemically controlled and case management, IC = Vaccination and nomically efficient alone. Each was dominated by a combina- case management; cv = vaccination, case management, and vector chemically nomically efficient ~~~~~~~~~~~controlled, EC = Environmental vector control and case management. tion of strategies that include case management. The ICE = Vaccination, case management, and environmental vector control; Inefficient policies: V = VeCtDr chemically controlled, I = Immunization, immunization result is because of the relatively high cost of Iv = Immunization and vector chemically controlled, E = Environmental vector control, A E = Immunization and environmental control. immunization- of $40.87 per person; the cost is a result of the Source: Authors cost-effectiveness model Dengue (with Notes on Yellow Fever and Japanese Encephalitis) 313 Table 14-3. Incremental Cost-Effectiveness of Interventions for Dengue Control DALYs Net cost Average cost Average cost per Additional Additional cost Incremental cost Intervention saved ($000) per DALY ($) death ($) DALYs gained ($000) per DALY ($) No vaccine and health system not developed Baseline (BASE) 0 0 0 0 0 0 0 Vector chemically controlled (v) 219 435 1,992 61,234 219 435 1,992 Environmental vector control (E) 692 2,172 3,139 96,461 474 1,737 3,668 No vaccine but developed health system Baseline 0 0 0 0 0 0 0 Case management improved (c) 557 327 587 15,042 557 327 587 Vector chemically controlled and case management (vc) 609 664 1.091 29,754 51 337 6,568 Environmental control and case management(EC) 720 2,189 3,040 92,712 III 1,524 13,696 With vaccine but health system not developed Baseline (BASE) 0 0 0 0 0 0 0 Immunization 0) 505 727 1,440 44,251 505 727 1,440 Immunization and vector chemically controlled (iv) 572 1,180 2,062 63,372 67 452 6,754 Environmental vector control (E) 692 2,172 3,139 96,461 120 992 8,278 Immunization and environ- mental control (IE) 717 2,954 4,117 126,537 25 781 30,927 With vaccine and developed health system Baseline (BASE) 0 0 0 0 0 0 0 Case management improved (c) 557 327 587 15,042 557 327 587 Immunization and case management (IC) 676 828 1,224 35,827 119 501 4,217 Immunization, case management, and vector chemically controlled (ICv) 692 1,250 1,806 53,691 16 422 26,363 Environmental control and case management (EC) 720 2,189 3,040 92,712 28 939 33,643 Immunization, case management, and environmentalcontrol (ICE) 727 2,959 4,071 124,968 7 770 112,933 Note Only policies that are feasible and economically efficient are listed. Source: Authors' cost-effectiveness model. because the number of deaths averted does not count the management, we have a good argument for cure. Preven- morbidity avoided by vaccinations. tion is expensive and is directed to a condition that is rela- Under both criteria, deaths averted and DALYs, case manage- tively rare when both epidemic and nonepidemic years are ment remains the most cost-effective first strategy. The cost averaged. per DALY gained, $587, is comparable to the per capita income of an average low-income developing country. Thus, case Priorities management of dengue fever, although not as cost-effective as some of the other interventions examined in this collection, On the basis of our analysis, we recommend policies in two is still reasonable and cost-effective for all but the very poorest areas. First, applying existing knowledge, we recommend mea- country. The cost per death averted, $15,042. is also an accept- sures for cost-effective control of dengue. Second, we examine able investment for a middle-income country. Among the how operational research could allow better disease control in preventive interventions, immunization, at $1,440 per DALY the future. gained, is the most cost-effective policy. Finally, the analysis adds future interventions to those Priorities for Resource Allocation under consideration. Case management remains the most cost- effective program, but the next intervention is to add im- Policies differ among the three mosquito-borne diseases. munization to case management (ic). That is, we first make For the two for which effective vaccines are available (yellow sure treatment facilities can do a good job; then we add fever and Japanese encephalitis) the major questions con- the preventive component. These results are opposite to the cern overcoming the technical and financial constraints to usual adage that prevention is cheaper than cure. With case vaccination. 314 Donald S. Shepard and Scott B. Halstead DENGUE. Policies for dengue control vary with time and the cost-effectiveness of case management would then be equal to level of development of a country's health system. In the case that of vaccinations. in which no vaccine is available and the health system is not One factor in favor of each of these control programs is that developed, chemical vector control was most cost-effective they can be implemented on localized scales. Improvements in ($1,992.00 per DALY), although not very effective in an abso- case management could be implemented at a single hospital. lute sense. Environmental vector control, through reduction Vector control, whether chemical or environmental, could be of breeding sites, is the only other alternative. As practiced in implemented at the level of a single city, and in part, at a Singapore, this policy proved highly effective but somewhat neighborhoodlevel.Thedeliveryofavaccine,onceithasbeen costly ($2.25 per person per year, even after excluding costs for developed, can be directed to receptive populations. Thus, controlling nuisance mosquitoes). It would be highly effective, although the cost-effectiveness of dengue control policies may and only slightly less favorable on cost-effectiveness than not place them in the highest priority for low-income coun- controlling mosquitoes through spraying. tries, they certainly are feasible for middle-income countries In the case in which no vaccine is available but the health and particularly for middle-class populations within middle- system is developed, the cost-effectiveness analysis suggests income countries. that case management is the most cost-effective policy ($587 Finally, the choice between preventive and curative policies per DALY). The analysis suggests that this method should be involves ethical issues. Some public health officials feel that a undertaken first. For additional control, chemical or environ- society has an obligation to prevent disease if it can reasonably mental vector control should be added. do so, even if curative policies appear somewhat more cost- In the case in which a dengue vaccine is assumed to be effective in the short run. The public's willingness to undertake available (beginning in 1997) but a country's health system is future preventive measures would be an additional benefit. not developed, immunization would be the most cost-effective alternative at $1,440 per DALY. In the model, we estimated a ELLOWFEVER.Partlybecauseofitshighcase-fatalityrateand relatively high cost for the assumed two-dose series for this partly because yellow fever has been controlled on a hemi- vaccine of $40.87 per person vaccinated. If this price dropped spheric basis both by vector control and by vaccination, mod- with increasing volume, the cost-effectiveness of this option ern societies regard epidemics of yellow fever as intolerable. In could improve substantially. For example, if the price per Africa, adequate supplies of potent yellow fever vaccine must dose of vaccine dropped from its assumed value of $17.50 to he on hand for preventive immunization programs and to $7.00, then immuniztion would become as cost-effective as combat yellow fever epidemics. Nigeria, repeatedly affected by case management. epidemics, still depends heavily on an antiquated manufactur- In the case in which a dengue vaccine is available and the ing facility established by the Rockefeller Foundation in 1939. health system is developed, all policies would be technically It is quite possible that batches of this vaccine have relatively feasible. Case management proved to be most cost-effective at poor thermostability compared with those produced by other a cost per DALY of $587, but immunization would be a valuahle manufactLrers and that they still contain avian leukosis virus. addition at an incremental cost of $4,217 per additional D)ALY Some lots may be contaminated with other organisms. Nigeria gained. Case management and vaccination would be expen- and other African countries also import vaccine from Brazil, sive for the countries with the lowest income. Many countries Senegal, and France. Even potent yellow fever vaccine is at risk of dengue in Asia and South America are middle- or extremely dependent upon an intact cold chain; on adequate upper-middle-income countries. For them, dengue control is a supplies of jet injectors, needles, and syringes; and on trained reasonable part of their healrh priorities. vaccinators. All are at present inadequate. The authors recom- An important caveat is that our analysis of vector control mend the following steps to address these problems: captures only direct patient benefits. Thus, certain secondary * Funds should be made available to purchase and stock- benefits of vector control are not captured or are incompletely pile potent yellow fever vaccine. measured. They include reduction in the nuisance and discom- p potectiyeliverysstminee fort of mosquito bites, possible reduction in transmission from * An effective delivery system is needed. Success in build- fewer infected people, and a possible reduction in the risk of ing a sustainable delivery system for EPI will also permit other vector-bome diseases, such as yellow fever. The impact routie or emergency delivery of yellow fever vaccine. on yellow fever would be only a theoretical advantage in most * Yellow fever vaccine should be incorporated into the EPI regions of the world, however. The otlher, less tangible benefits, program for those countries in Africa at risk of yellow fever. cannot be valued within the scope of this chapter. Although a thorough sensitivity analysis has not been done, JAPANESE ENCEPHALITIS. Public outcries and political pres- the cost-effectiveness of case management depends on the sures for action seem to be particularly powerful against this availability of moderately priced, high-quality referral hospital disease. Nonetheless, many affected nations have postponed care. We assumed that the base cost of treating DHF/DSS was the purchase of sufficient quantities of vaccines needed to $200 per case (for an average hospital stay of five days) and immunize all at-risk children. This is largely because of the that improved case management would raise this cost fivefold. high cost in hard currency of Japanese-manufactured killed If the base cost were about twice as high ($487 per case), the vaccines (at least $4.60 to $6.90 just for vaccine alone). India Denigue (with Notes on Yellou Fever and Japanese Encephaliis) 315 and Thailand, with Japanese assistance, are investing in do- COHORT: Number of persons in one year's birth cohort in the mestic vaccine manufacturing facilities. These will require standard population. huge colonies of laboratory mice. There is reason to doubt, Effectiveness of interventions based on past performance, that output of vaccine will he EARectiscotinterenii pe sufficient for national needs. Estimates of current and pro- YeA aer io eodremaining life dexpe ctan e jected costs and losses resulting from Japanese encephalitis the average age of death of a fatal dengue case. mighte cosrbtseand loss rationaltingvestmet Japolese ecephaitis SALVAGE: Proportional reduction in case-fatality rate of DHF/DSS might contribute to rational investment policies, whether for after improved case management. domestic manufacture or for vaccine importation. SHORTEN: Proportional reduction in duration of illness among Priorities for Operational Research hospitalized cases after improved case management. VACC.EF (vaccine efficacy): Proportional reduction in number Priorities for operational research focus on ways to control the of cases. mosquito), development of vaccine (for dengue) improvement tCOVERAGE: Proportion of birth cohort vaccinated. mof vaccine. (for dpenge) eiimeprovm VTRC EF: (vector chemical efficacy): Proportional reduction in of vaccines (for yellow fever and Japanese encephalitis), nube ofcssfo hmcaetrcnrl numb er of cases from chemical vector control. DVTRE.EF: (vector environmental efficacy): Proportional reduc- DENGUE. First, careers in vector control need to be entirely reconstituted. The leaders and experienced veterans of the tion in number of cases from environmental vector control. Latin American Aedes aegypti campaigns have disappeared Costs of case nanagement without replacements. Second, politically acceptable, cost- TREAT: Current cost per case of treating hospitalized DHF/DSS. effective methods of Aedes aegypti control or eradication are IMPRoVE: Cost as a multiple of TREAT per case of DHFPDSS of needed. Third, research on cost-effective, efficacious, safe, and improved case management, converted to future value as of thermally stable vaccines requires adequate funding. Current the average expected age at death from dengue. programs are very poorly funded. Costs of vaccines DEVELOP: Annualized expected development cost for cohort YELLOW FEVER. A second-generation genetically engineered vac- cine might overcome the present requirement for lyophilization. until successful development. VACCINE: V,accination cost per person vaccinated, converted JAP'ANESE ENCEPHALITIS. First, a potent, safe, thermallv stable to future value as of the average age at death from dengue. live-attenuated vaccine is needed. A reasonable candidate has Costs of vector control been developed in China. This requires internationially ac- VECTORC: Cost per person per year in target population ceptable phase 1, 11, and III testing and introduction into the of chemical vector control, including amortization of initial affected countries of Thailand, Myanmar, Nepal, Bangladesh, costs. India, and Sri Lanka. Second, investments, boti technical and VECTORE: Cost per person per year in target population of financial, are needed in vaccine production capacities in most environmental vector control, including amortization of of these affected countries. initial costs. The model also uses the following intermediate variables: Appendix 14A. Definitions of Variables in the DEATHS: Number of dengue deaths in the standard population Cost-Effectiveness Model in one year with a specified control program. D.AVERTED: The number of deaths averted in the standard The model involves the following parameters: population in one year. Morbidit-y and mortality D.BASELINE: The number of dengue deaths in the baseline situ- STAND to;P: The number of persons in the standard population ation of no dengue control program. STAND f,,P ,.nbroestsihtna DALY.MORB: The number of disability-adjusted life-years saved (an arbitrary size) to which the model is applied. through morbidity averted. STAND POP is one million persons of all ages. LDALY.MORT: The number of disability-adjusted life-years saved CASES: Number of dengue infections without vaccination or through mortalit averred. vector control in the hvpothetical birth cohort (all births through morted avered. withi th stndr pouato in on yer) YEAR D: The discounted life expectancy at the average age at withnm the standard population in one year). whc on. tews ol hv ido ege SHOCK.R (Shock rate): Proportion of dengue infections that whch one otherwise would have died of dengue. progress to dengue shock syndrome. FATAL: Case-fatality rate of DSS 1960-65. Appendix 14B. Relationships in the CLINICAL: Proportion of dengue infections which are clinically Cost-Effectiveness Model apparent. DUR: Average duration of clinical illness, expressed in disability- We look at these relationships in the context ofdearhs averted, adjusted life-vears. disability-adjusted life-years saved, and aggregate costs. 316 Donald S. Shepard and Scott B. Halsteacd Deaths Averted DEATHS = CASES ( X VACC-EF COVERAGE) ( - VCTRC.EF) SHOCK.R (1 - SALVAGE) FATAL. The number of deaths is expressed as the product of the three factors: the number of infections times the proportion of those ronmental vector control (policy ICE): infections which progress to the potentially fatal condition of DHF/DSS times the proportion of DHF/DSS cases which are fatal. DEATHS = CASES (1 -VACC.EF COVERAGE) C(1 -VTRC.EF) For consistency with the formulas in the computer spreadsheet SHOCK.R (1 - SALVAGE) FATAL. in which this model was written, multiplication is shown by an asterisk. The equations below show how this principle Number of deaths averted applies to each of the control strategies. For each strategy, the number of deaths averted is Single interventions D.AVERTED = D.BASELINE - DEATHS. At the baseline (policy BASE): DALYs Saved D.BASELINE = CASES SHOCK.R FATAL. The number of disability-adjusted life-years saved with each policy is the sum of the number saved through deaths averted DEATHS = CASES SHOCK R FATAL ( - SALVAGE). and through morbidity avoided or reduced. That is, for all With immunization or vaccination (policy 1):interventions, the overall number of DALYS saved is DALY.MORT + DALY.MORB. DEATHS = CASES (1 - VACC.EF COVERAGE) SHOCK.R FATAL. The prevention strategies (vector control and vaccination) avoid morbidity, whereas case management shortens the mor- DEATHS = CASES ( - VCTRC.EF) SHOCK.R FATAL. bidity of serious cases. The DALYS saved through deaths averted With environmental vector control (policy E): are DALY.MORT =D.AVERTED YE.AR.D. DEATHS = CASES ( l VCTRE.EF) SHOCK R FATAL. On the basis of experiences in Thailand (Halstead 1980b), the Two-way combinations number of cases hospitalized is assumed to be twice the number With vaccination and case management (policy ic): experiencing dengue hemorrhagic shock or dengue shock syn- drome. DEATHS = CASES (I - VACC EF COVERAGE) SHOCK.R (I - SALVAGE) - FATAL. Single interventions With vector chemically controlled and case management (pol- Because case management benefits only hospitalized cases, icy vc): only these cases experience a reduction in morbidity. The DEATHS = CASES . (1 - VCTRC.EF) S[HOCK.RR. ( SALVAGE) number of DALYs saved through shortened morbidity in case . FATAL. management (policy c) is OALY.MORB = CASESC 2 .SHOCK.R SHORTEN DUR. With environmental vector control and case management (policy EC): Infections with dengue virus, like some other infections, are not always clinically apparent. The benefit of reduced morbid- DEATHS = CASES (1 - VCTRC.EF) .SHOCK R (1 - SALVAGE) ' FATAL. ity applies, of course, only to clinically apparent cases. For chemical vector control (policy v), the morbidity avoided is With immunization and vector chemically controlled (pol- DALY MORB = CLINICAL CASES VCTRC.EF DUR. icy IV): CASESS = C 1SES VACC.EF -.COVER-AGE) VCTRC.EF) Similarly, for environmental vector control (policy E), the DEATHS =CSS( l VC.FCOVERAGE) ( l VCTRC.EF) mriiyaoddi SHOCK.R FATAL. With immunization and environmental vector control (pol- DALYMOR= CLINICAL CASES VCTRC.EF DUR. icy IE): Immunizations are assumed to lower the attack rate of dengue but not to affect the severity of a dengue infection. DEATHSHCASES RI - FATALE COEAE I CRE) Thus, for immunizations (policy I) the morbidity avoided is SHOCK.R FATAL. DALY.MORB = CLINICAL- CASES VACC.EF COVERAGE DUR. Three-way combinations With immLinization, case management improved, and vector Two-way comibinations chemically controlled (policy ICV): When chemical or environmental vector control and case Dengue (with Notes on Yellow Fever and Japanese Encephalitis) 317 management are combined, the benefits from cases avoided are With improved case management (policy c): supplemented by shorter morbidity for the hospitalized cases COSTS = CASES SHOCK.R TREAT IMPROVE. among those that still occur. The morbidity avoided from chemical vector control combined with better case manage- With vaccination (policy I): ment (policy vc) is COSTS = CASES- (I - VACC.EF- COVERAGE) SHOCK.R- TREAT DALY.MORB = CLINICAL CASES VCTRC.EF- DUR + CASES + DEVELOP + VACCINE- COVERAGE COHORT. . (1 - VCTRC.EF) 2 . SHOCK.R- SHORTEN - DUR. With vector chemically controlled (policy v): Similarly, the morbidity avoided from environmental vec- COSTS = CASES- ( 1 - VCTRC.EF) SHOCK.R- TREAT tot control combined with better case management (policy + STAND.POP VECTORC. EC) is With environmental vector control (policy E): DALY.MORB = CLINICAL CASES VCTRC.EF DUR + CASES (I - VCTRC.EF) 2 * SHOCK.R- SHORTEN DUR. COSTS = CASES* (1 - VCTRC.EF) SHOCK.R TREAT With immunization and the vector chemically controlled + STANDPOP- VECTORE. (policy Iv) the morbidity avoided is Two-wav combinations DALY.MORB = CLINICAL- CASES VCTRC.EF- DUR + CASES With vaccination and case management (policy IC): (1 - -VCTRC.EF) - VACC.EF - COVFRAGFE- DUR. COSTS = CASES - (1 - VACC.EF- COVERAGE) - SHOCK.R- TREAT With immunization and environmental vector control (policy - IMPROVE + DEVELOP + VACCINE - COVERAGE - COHORT. IE) the morbidity avoided is With vector chemically controlled and case management (pol- DALY.MORB = CLINICAL- CASES VCTRC.EF DUR + CASES icy vc): ( I - VCTrRC.EF) VACC.EF COVERAGE - DUR. COSTS = CASES (1 -VCTRC.EF) SHOCK.R- TREAT With both vaccination and case management (policy IC), the * IMPROVE + VECTORC- STAND.POP. morbidity avoided would be With environmental vector control and case management DALY.MORB = CLINICAL - CASES- VACC.EF - COVERAGE (policy EC): DUR + CASES- (1 - VACC.EF- COVERAGE) COSTS = CASES (1 -VCTRC.EF) - SHOCK.R- TREAT- IMPROVE - 2 - SHOCK.R - SHORTEN- DUR. + VECTORE STANDPOP. Three-way combinations With vaccination and environmental vector control (pol- With immunization, case management improved, and vector icy IE): chemically controlled (policy Icv): COSTS = CASES (1 -VACC.EF- COVERAGE) - (VCTRC.EF) DALY.MORB = CLINICAL- CASES (VACC.EF - COVERAGE SHOCK.R -TREAT + DEVELOP + VACCINE - COVERAGE + VCTRC.EF- VACC.EF - COVERAGE - VCTRC.EF) COHORT + STAND.POP - VECTOR. D )UR + CA~SES - (I - VACC-EF - COVERAGE) With immunization and vector chemically controlled (pol- 2 - SHOCK.R - SHORTEN - DUR. icy IV): With immunization, case management improved, and envi- COSTS = CASES VACC.EF COVERAGE) (VCTRC.EF) ronmental vector control (poliCy ICE): SHOCK.R TREAT + DEVELOP + VACCINE - COVERAGE DALY.MORB = CLINICAL - CASES - (VACC.EF - COVERAGE COHORT + STAND.POP - VECTOR. + VCTRC.EF - VACC.EF - COVERAGE - VCTRC.EF) DUR + CASES - (I - VACC EF - COVERAGE) Three-way combinations * 2 - SHOCK.R - SHORTEN - DUR. With vaccination, case management, and vector chemically control (policy Icv): Aggregate Costs COSTS = CASES - (I-VACC.EF - COVERAGE) - (I - VCTRC.EF) - SHOCK.R -TREAT -IMPROVE + DEVELOP + VACCINE The aggregate costs are expressed as the number of people in CKAGE IMPRT + DEVELORC. the standard population receiving each service times the unit cost of that service. With vaccination, case management, and environmental vec- tor control (policy ICE): Siasline (itervBentions COSTS = CASES - (i-VACC.EF - COVERAGE) - (1 - VCTRC.EF) BlSHOCK.R - TREAT - IMPROVE + DEVELOP + VACCINE COSTS = CASES - SHOCK.R - TREAT. COVERAGE - COHORT + STAND.POP - VECTORE. 318 Donald S. Shepard and Scott B. Halstead Appendix 14C. Numerical Values of Input main. These steps are estimated to require a further invest- Parameters ment of $25 million and require five more years from 1992. Thus, the target date remains 1997. Because the average age Parameter values are listed alphabetically below. For each at death was six (as described below), there is an additional parameter, we give the best estimate and the basis of that five-year delay from administration of the first dose at age estimate. If no units are shown, the parameter is a pure oneuntiladeathispotentiallyaverted.Thusdeathswillnot dimensionless number. Values are shown in the form in which be averted until ten years in the future (five fordevelopment they are entered into the model. Thus value of CLINICAL below plus five after administration). (16 percent) is shown as the decimal share 0.16. In full use in a stable (long-term) situation, the vaccine would be offered to the birth cohort in all countries at risk CASES: 52,400 dengue infections in standard population per of dengue. The population of countries at risk of dengue is year. We estimated two infections per person per lifetime. 1,210 million people, or 1,210 times the standard-size cal- Thus, CASES equals two times COHORT. Although a dengue culation of 1 million people used in this analysis. To make infection confers immunity to the type of dengue virus costs commensurate with the timing of benefits, the future which caused the infection, the person is still at risk of the value of the expenditure needs to be calculated, at the time remaining three of the four types of dengue virus. This is a the vaccine would be in full use. Furthermore, because the long-term average level. During the 1981 dengue outbreak, success of research is not certain, the expenditure needs to Cuba (with a population of 10 million persons) had 2.36 be adjusted for the expected chance of success, now esti- million infections (based on serological data), or an infec- mated at 90 percent. Because research and development is tion rate of 236,000 per million population (Guzman and a capital cost, the expenditure must also be annualized over others 1990). Epidemics result from a buildup of susceptible its expected useful life and rescaled for the birth cohort. We persons. The long-term rate for CASES is about one-fifth of assign a twenty-year useful life to the current research effort, the Cuban rate. on the grounds that an improved vaccine would be available CLINICAL: 0.16. This share is based on clinical data for DHF/DSS after that time. Several other important vaccines, such as in children from Thailand (Halstead 1980b). measles and polio vaccines, have benefited from substantial COHORT: 26,200 persons born per year in the standard popula- improvements over this period. Thus, the cost per cohort tion. The birth cohort size is the weighted average of the was calculated as: crude birth rate in the countries at risk of dengue. This was $25,000,000- (1.03)(5 5) / (1,210 . 0.90 calculated as a weighted average of the crude birth rate per annualizing factor) = 2,488 1,000 population in countries with at least 1 miillion persons, based on data (generally for 1988) in the World Development where the annualizing factor is the present value of 1 for Report (World Bank 1990). The value of this parameter twenty years at 3 percent interest. corresponds to a crude birth rate of 26.2 per 1,000 popula- tion per year. DURATION. 0.0148 year. A clinical episode of DHF/DSS is esti- COVERAGE: 0.73. This is the overall coverage of DPT-3 among mated to last nine days. This time counts the patient's one-year-old children in developing countries at risk of inability to pursue his or her usual activities before, during, dengue in 1987-88 (Grant 1990). If and when a dengue and after treatment. This duration is slightly longer than vaccine is developed, it will probably be offered to children Osani's (1983) estimate of six days for dengue fever and the through the delivery mechanism of the Expanded Pro- policy of the Brazilian social security system, which allows gramme on Immunization. a worker seven days of authorized disability for a case of DEVELOP: $2,488. According to a study by the Institute of dengue fever (Kiela, personal communication, Everardo Medicine, Vaccine Development (1986), the cost of re- Chagas Hospital, Rio dejaniero, July, 1989). As Brazil then search and development to try to produce a useful dengue had virtually no DHF, the mean duration should be lower vaccine was estimated at $25 million; the probability of than in areas in which this complication occurred widely. success was 0.75; when estimates were compiled in approx- The ill person has a fever, severe aches, and is generally imately 1985, twelve years were then thought to be required prevented from working or carrying out his or her usual to license and adopt the vaccine. That is, the projected activity. Except for the minority of cases that progress to target year was 1997. During the past decade, researchers in DHF/DSS, the victim can remain at home and is conscious but Thailand, who have received about $5 million in external feels extremely uncomfortable. In cases that progress to support from the Rockefeller Foundation and the Italian hemorrhagic fever, the patient may be in shock for part of government, and the equivalent of several million dollars of the illness. We have assigned a quality level of 0.4 to this in-kind support from the Thai government, have now pro- acute illness on a scale where 0 denotes death and 1 perfect duced a tetravalent vaccine in the laboratory and tested it health. Thus, the morbidity loss is converted to an annual successfully on 200 volunteers. (Replication of this research equivalent as 9/365 . (1 - 0.4). in an industrial country today would have cost about $100 FATAL: 0.058. This rate was the case-fatality rate of DSS cases in million.) Final development, full-scale testing in humans, Thailand in 1958-65, before good treatment became avail- and development of production methods and capacity re- able (Halstead 1980a). On thebasisof 158deaths in 116,000 Den gue (with Notes on Yellow Fever and Japanese Encephalitis) 319 hospitalized patients, Cuba's case-fatality rate in hospital- VACCEF: 0.95. The Bureau of Biologics standard for immuno- ized cases was 0.0014 during its 1981 dengue epidemic genicity (and efficacy) of live-attenuated viral vaccines in (Kouri and others 1987). Cuba has a good health system, so the United States is 95 percent. Tetravalent dengue vaccine its case-fatality rate should reflect the effect of SALVAGE. would not be released until it is at least that effective. Undoubtedly, the hospitals included some cases that were VCTRCEF: 0.30. Areas with vigorous efforts at vector control not DHF/DSS. appear to have avoided outbreaks of dengue fever, whereas IMPROVE: 5. This is the estimated ratio of costs in a referral such outbreaks appear to have occurred in areas that lacked specialty hospital to those in a typical secondary hospital. such programs. For example, Venezuela suffered a DHF/DSS SALVAGE. 0.917. This rate of salvage of hemorrhagic cases is epidemic in 1989-90 after apparently lax control programs. based on experience in Thailand following improvement in The Brazilian state of Sao Paulo, which has had a well- hospital care. It is the reduction in the former case-fatality organized dengue control program, including clean-up cam- rate of 0.058 (see FATAL, above) to the rate in 1986-89 of paigns, has had minimal dengue cases. Puerto Rico's 0.0048. ongoing spraying programs have helped to prevent large SHOCK.R: 0.0078. This is the average of the rates of DHF/DSS epidemics, although dengue still continues on the island. (corrected to include only cases meeting WHO criteria) in VCTRE.EF- 0.95. The efficacy is based on the success of the Thailand in 1962 and Cuba in the epidemic of 1981. control program in Singapore, which combined environ- Thailand's rate was 7.5 DHF/DSS per 1,000 persons, calculated mental control (elimination of breeding sites), education, from the experience at Children's Hospital (Halstead localized chemical fogging, a law prohibiting conditions for 1980b). Cuba's rate of 0.0080 is based on 20,000 DHF/DSS disease-bearing insects, and slum clearance. Prior to the compared with 2,360,000 infections during the epidemic establishment of a vector control unit, dengue epidemics (Guzman and others 1990). occurred annually. In 1966, for example, 630 cases were SHORTEN: 0.25. Good clinical management improves the reported and 24 persons died of DHF. In a small epidemic a DHFjDSS patient's rehydration, shortens the period of shock, decade later (1978), only 2 deaths were reported. If the cycle reduces bleeding, and hastens return to normal function. of five-year epidemics had continued, another epidemic STAND.POP 1,000,000 persons. The size ofthe standard popula- would have occurred in 1983, but none happened (Chan tion (total of all age groups). The population ofone million 1985). Thus the control program reduced both the severity was chosen for consistency in comparing dengue with other and frequency of dengue epidemics. Puerto Rico controlled interventions. Any other convenient size could be chosen, dengue to low levels in 1973 when large numbers of workers but the value of COHORT would have to be modified accord- were hired to clean up neighborhoods under the War on ingly. Poverty's Comprehensive Employment and Training Act. TREAT: $200. The cost of treating one case of dengue hemor- VECTORC. $0.46. This per capita cost is the average of per capita rhagic fever is based on $40 per hospital day (the average in costs of dengue control in 1988 in Brazil ($0.25) and Puerto Brazil) times five days (the average for Thailand) of hospital Rico ($0.67) based on original field studies. Although envi- care per case of DHF/DSS. ronmental control was used occasionally in these two areas, VACCINE $40.87. The vaccine is expected to require two doses; both relied primarily on chemical control during this year, the first at age one and the second five years later. This especially spraying of streets and placement of abate or schedule is expected to offer protection at least through the temefos in places where water collects. period of greatest risk, from infancy through youth, if not VECTORE: $2.25. In the Singapore program, described above longer. Because the vaccine contains four antigens to pro- (Chan 1985), the cost was approximately $3.00 per capita. tect against all four dengue types, it is relatively complex to Environmental vector control not only reduces the risk of produce. In the study on vaccine development for the de- dengue but also reduces the population of Culex mosquitoes, veloping world by the Institute of Medicine (1986), esti- whose bite is itchy and annoying. Because the dengue- mated possible dengue vaccine costs ranged from $12.00 to carrying Aedes aegypti mosquitoes are smaller, their bite is $48.00.Wenowestimateacostperdoseof$10.00to$25.00 less noticeable. Thus, the cost of effective vector control witha midpoint of$17.50for the vaccine itself. In addition, needs to be allocated between dengue and control of nui- administration of the first dose at age one was assumed to sance mosquitoes. To perform this calculation, we obtained cost $0.50, because it could likely be given during the figures from New Orleans, Louisiana, a city known to spend contact for another vaccine, such as measles. The second public funds on control of nuisance mosquitoes. dose, at age six, was assumed to require a separate contact. The expenditure ($1 million) and denominator (500,000 Because this might be done on a mass basis in schools, persons) in New Orleans give a per capita expenditure of however, the delivery cost could be modest. We estimate a $2.00 We interpret this amount as a revealed preference of cost for this contact of $2.50, which is consistent with the willingness to pay for control of nuisance mosquitoes. In per contact costs found in cost studies of the Expanded trying to extrapolate this result to Singapore, we assumed Programme on Immunization if all doses are considered that this expenditure would be slightly income elastic, as is (Shepard and others 1986). The combined two-dose cost is: health expenditure generally. Assuming an income elastic- ity of 0.3 and using the fact that Singapore's per capita (17.50 + 0.50)- (1.03)5 + (17.50 + 2.50) = 40.87. income ($7,500) is half that of New Orleans, we estimate 320 Donald S. Shepard and Scott B. Hal stead that the per capita willingness to pay for control of nuisance Seroepidemiologic Study." AmencanJournalof TropicalMedicine andH-giene mosquitoes would be $0.75, or one-quarter of the total per 42:179-84. capita spending on environmental vector control. Subtract- Halstead. S. B. 1980a. "Dengue Haemorrhagic Fever-A Public Health Prob- ing this amount leaves a per capita expenditure of $2.25 lem and a Field for Research." Bulletin of the World Health Organizarion 58:1-21. allocated to dengue control. . 1980b. "Immunological ParametersofTTogavirus Disease Syndrozmes." YEAR D 25.5 years. In Southeast Asia, where there are good data In R. W. Schlesinger, ed., The Togaviruses. New York: Academic Press. on the age distribution of dengue deaths (Halstead 1969), . 1984. "Selective Ptimary Health Care: Strategies for Control of the average age at death was about six years, and we have DiseaseintheDeveloping-World. I1. Dengue."RevrewsoflnfectiousDiseases used this age for all areas at risk of dengue. We calibrated a 6:251-64. model life table to the areas at risk of dengue. The calibra- . 1987. "Arboviruses of the Pacific and Southeast Asia." In R. Feigin tion was based on the model West life table, which best fit and J. Cherry, eds., Textbook of Pediatric Infecuous Diseases, 2d ed. Philadel- the weighted average life expectancy for areas at risk of phia: W. B. Saunders. dengue. The West table best describes an "average" mortal- Halstead, S. B 1. E. Scanlon, P. Umpaivit. and S. Udomsakdi. 1969. "Dengue deng.e. The West table best describes an average"mortal- and Chikungunya Virus Infection in Man in Thailand, 1962-1964. 4. ity pattem, and it is recommended when "no reliable infor- Epidemiologic Studies in the Bangkok Metropolitan Area." AmencanJour- mation on the age pattern of mortality is available" (Newell nol of Tropical Medicine and Hygiene 18:997-1021. 1988, p. 138). The weighted average (based on countries Institute of Medicine. 1986. Newu Vaccine Development. Vol. 11, Establishing with a population of one million or more at risk of dengue), Pnorities, Diseases of Importance in Developing Counrnes. Washington, D.C.: was 66.3 years (World Bank 1990). This average was best fit National Academy Press. by the Level 20 model table, which yields a life expectancy Kouri, G. P., M. G. Guzman. and J. R. Bravo. 1987. "Why Dengue by ~~~~~~~~~~~~~~~~~~Haemorrhagic Fever in Cuba? An Integral Analysis-" Tranisactions of the of 65.6 years. We estimated remaining discounted life ex- Roval Sociey ofeTropical Medicine ad I1ygiene 81821-23 pectancy at age 6 years, using a discount rate of 3 percent. Kouri, G. P., M. G. Guzman, J. R. Bravo, and C. Triana. 1989. "Dengue This estimation used 5-year age intervals beyond age 10, Haemorrhagic Fever/Dengue Shock Syndrome: Lessons from the Cuban with a maximum at age 102.5 years. Epidemic, 1981." Bulletin of the World Health Organization 67:375-80. Matsuraspas, W. 1981. "The Results of Evaluarions of DHF Prevention and Control Programs, 1977-1980" (in Thai). Journal of Communicable Diseases Notes 7:327-47. Nasidi, A.. T. P. Monath, K. M. De Cock, 0. Tomori, R. Cordellier, 0. D. The authors are indebted to Dr. Francisco Ranmos for leading the case studies Olaleye, T. 0. Harry, J. A. Adeniyi, A. 0. Sorungbe, A. 0. Ajose-Clocker, of Brazil and Puerto Rico. to Dr. Duane Gublet and Dean Jamison for valuable G. Van Derlan, and A. B. 0. Oyediran. 1989. "Urban Yellow Fever comments and suggestions, to Dr. Antonio Carlos Rodopiano de Oliveira. Epidemic in Western Nigeria, 1987." Transactons of the Royal Society of former director of the Superintendencia de Campanhas de Satide Puiblica Tropical Medicine and Hygiene 83:401-6. (SUCAM1) in Brazil for assistance in the Brazilian case study, to Dr. Carl Kendall Newt II, C. 1988. Methods and Models it Demography. New York: Guilford. for facilitating financial arrangements, and to Aravan Trangarn for data about Osani, C. H., P. A. Travassos, A. T. Tang, R. S. do Amaral, A. D. Passus, P. Thailand. This work was supported in part by the Rockefeller Foundation L. Tanil. 1983. "Surto de Dengue em Boa Vista, Roraima Nota Previa." through the Harvard Institute for lnternational Development and Johns Revista Instituto Medicine Tropical Sao Paulo 25:53-54. Hopkins Universitv. PAHO (Pan-American Health Organization). 1990. "Status of Dengue Out- break in Venezuela." Caracas. Schatzmayr, H. G., R. M. Rogueira, and A. P. Travassos da Rosa. 1986 "An References Ouitbreak of Dengue Virus at Rio de Janeiro." Memorias do Institisto Osu'aldo Cruz 81:245-46. Chan, K. L. 1985. Singapore's Dengue Haemorrhagtc Fever Control Programme: Secretaria de Estado de Saude e Higiene. 1986. "Informe Epidemniilogicosobre A Case Study on the Successful Control of Aedes aegypti and Aedes albopictus Dengue" 1:1-7. Unpublished report, Brazilian Ministry of Public Health. Using Mamly Etivironmnental Measures as a Partof IntegraLed Vector Control. Shepard, D. S.. L. Sanoh, and E. Coffi. 1986. "Cost-Effectiveness of the Tokyo: Southeast Asian Medical Information Center. Expanded Programme on Immunization in the Ivory Coast: A Preliminarv De Cock, K. M., T. P. Monath, A. Nisidi. P. M. Tukei, J. Enriquez, P. Lichfield. Assessnienit." Social Science and Medicine 22:369-77. R. B. Craven, A. Fabuj6, B. C. Okafor, C. Ravaoiijanahary, A Soper. F. L.,D. B. Wilson S. Lima,and W. SaAntunes. 1943.TheOrganization Sorungbe. 1988. "Epidemic Yellow Fever in Eastern Nigeria, 1986." Lancet of Permtarent Nation-Wide Anti-Aedes aegypti Measures in Brazil. New York: 2:630-33. Rockefeller Foundation. Grant, J. P. 1990. The State of the Wo'rld's Children, Y990. New York: Oxford Von Allmen, S. D., R. H. Lopez-Correa, J. P. Woodall, D. M. Morens, J. University Press. Chikiboga, and A. Casta-Velez. 1979. "Epidemic Dengue Fever in Puerto - 1992. The State of the World's Children. 1992 New York: Oxford Rics, 1977: A Cost Analysis." Amencan Journal of Tropical Medicine and University Press. Hygiene 28:1040-44. Guznlan, M. G., G. P. Kourt, J. Bravo. M. Soler, S. \Vazque:, and L. Mosier. World Bank. 1990. World Development Report. 1990. New York: Oxford 1990. "Dengue Heniorrhagic Fever in Cuba, 1981: A Retrospective Uiiiversitv Press. 15 Hepatitis B Mark Kane, John Clements, and Dale Hu Hepatitis B, one of the main diseases of mankind, is now endemic, internationalagencies and donors have not made the preventable with safe and effective vaccines-the first vac- vaccines available to developing countries who are dependent cines against cancer. More than 2 billion individuals alive on donors for the vaccines. An analysis of the reasons for this today have been infected at some time in their lives with the may shed light on the future of immunization as a viable public hepatitis B virus (HBV), and approximately 350 million are health strategy. chronically infected carriers of this virus. These carriers are at high risk of serious illness and death from cirrhosis of the liver History and Epidemiology of HBV Infection and primary liver cancer, diseases that kill more than 1 million carriers per year (Maynard, Kane, and Hadler 1989). Primary Hepatitis B virus infection leads to one of three outcomes in liver cancer caused by HBV infection is one of the top three humans. An infected individual may die of fulminant hepatitis causes of cancer death in much of Africa, Asia, and the Pacific within days or weeks of clinical onset of disease, may recover Basin (Parkin 1986). In addition, these carriers constitute a after symptomatic or asymptomatic acute infection and de- reservoir of infected individuals who perpetuate the infection velop lifelong immunity, or may become a chronic carrier, from generation to generation. harboring a persistent infection which usually lasts for life. The Most people in Africa, eastern Asia, Southeast Asia, the age of infection is the primary factor in determining the Pacific Basin, the Amazon Basin, and parts of the Middle East outcome of HBV infection. become infected with this virus during childhood, either from Approximately 25 percent of chronic carriers will die from an infected mother (perinatal transmission) or from another cirrhosis or primary hepatocellular carcinoma (PHC), also child. Infection during childhood is especially likely to lead to called primary liver cancer (Beasley and Hwang 1984). Cir- the chronic carrier state. In Europe, North America, much of rhosis is usually preceded by chronic active hepatitis, wvhich Latin America, and Australia, hepatitis B infection is an can cause years of morbidity and significant work loss. Death important sexually transmitted disease and a significant cause from cirrhosis and PHC usually occurs during the third to sixth of morbidity for health care personnel and certain other groups decade of life, during the peak years of adult productivity. defined by lifestyle and occupation (cDC 1990). Hepatitis B (HB) vaccines, if given prior to infection, can Geographical Distribution prevent disease and the carrier state from developing in almost all individuals. These vaccines have been used in more than Hepatitis researchers have divided the world into areas of 100 million persons and have proven to be among the safest, "high," "intermediate," and "low" HBV endemicity, basing this most immunogenic, and most effective vaccines yet developed. division on the prevalence of HBV markers and on the primary The vaccines are most effectively used as a routine part of the modes of HBV transmission. Areas of high endemicity include infant immunization schedule, although they can be used at those in which most of the population becomes infected with any age. the virus, usually during the perinatal period or during child- Recent dramatic decreases in vaccine cost in developing hood. Various authors have used figures of 5 percent to 10 countries (from $20 to$1-$2 perpediatric dose) have allowed percent to define the lower limit of the prevalence of HBV public health officials to consider the mass use of these vac- carriers for this category. The upper limit of the prevalence of cines in infant immunization programs (Kane, Ghendon, and the carrier state is about 20 percent. Most countries included Lambert 1990), but it is still considerably more expensive than in this category have a carrier prevalence of 10 to 15 percent, the other routine childhood vaccines. Although these vac- and 50 to 95 percent of the population have serologic evidence cines have been widely used by health care workers itn indus- of prior HBV infection. Africa, Asia east of the Indian Subcon- trial countries and as a routine infant immunogen in countries tinent, the Pacific Basin, the Amazon Basin, the Arctic Rim, with relatively more resources but in which the disease is the Asian Republics of the Commonwealth of Independent 321 322 Mark Kane, John Clemenr.s, and Dale Hu States (us), and portions of the Middte East, Asia Minor, and Transmission from child to child, often called horizontal the Caribbean are areas of high endemicity. Parts of eastern transmission, is responsible for the majority of HBV infections Europe such as Bulgaria, Romania, Albania, and Moldova and carriers. Although the relative importance of the various have a carrier prevalence of between 5 and 10 percent in the modes of transmission from child to child have not been general population. established, many hepatitis researchers believe that skin le- Areas of internediate endemicity generally have an HBV sions such as impetigo, scabies, abrasions, and infected insect carrier prevalence of 2 to 5 percent, and 30 to 50 percent of bites play an important role. These lesions provide a route for the population have serological evidence of prior HBV infec- the virus to leave the body of infectious children and a route tion. Some parts of southern and eastern Europe, the Middle for it to enter the body of susceptible children with whom they East, western Asia through the Indian Subcontinent, and parts have skin-to-skin contact, such as in wrestling or sharing the of Central and South America are included in this category. same bed. Other modes oftransmission include reuse ofunster- In these areas both child-to-child and adult-to-adult transmis- ile needles and other medical and dental equipment, tattooing sion occur. Acute viral hepatitis with jaundice is a primary and other scarification procedures, sharing ofhousehold items cause of morbidity because a substantial proportion of infec- such as toothbrushes, and sexual activity. Premastication of tion occurs in older adolescents and adults, who are much more food and insect transmission have been postulated as modes of likely to present with acute clinical disease. transmission but remain unproven. North America, western Europe, Australia, and parts of The transmission of HBV to adults is the primary mode of South America are considered to be areas of low endemicity. transmission in regions of lower endemicity where large popu- In these areas permnatal and child-to-child transmission is lations of susceptible adults are found. About one-third of relatively uncommon1, and most infectionis occur in adults infected adults develop clinical hepatitis B with jaundice, and through sexual activity, needle sharing dtiring drug abuse. 6 to 10 percent become chronic HBV carriers with a subsequent or dtiring occupational exposure to blood. Acute hepatitis B is risk of chronic active hepatitis, cirrhosis, and PHC. Sexual a significant cause of morbidity in many countries in this transmission, both heterosexual and homosexual, accounts for category. the majority of adult transmission. In some Western countries, needle sharing by drug abusers is also important. Hepatitis B is Modes of Transmission and Outcome of HBV Infections the main infectious occupational hazard to health care workers in areas of low and intermediate endemicity. Transmission to Understanding the outcome of HBV infection in children and patients by contaminated blood product and unsterilized re- adults is critical to designingeffective control strategies. Young used medical and dental instruments also occurs. In addition, children rarely develop symiptomatiC HBV infection with jaun- any of the modes of transmission discussed for child-to-child dice, but abohut 25 percent of children infected before the age transmission may occur. of seven will become carriers. The younger the child, the more likely it is that this will occur. Many carriers who acquire Hepatitis B Vaccines infection during childhood will live long enough to develop PHC after a latency period of thirty to sixty years. After the age Hepatitis B vaccines are alum adjuvented highly purified prep- of seven, children exhibit an adult pattern ofdisease outcome, arations of hepatitis B surface antigen, the glycoprotein that with about 5 to 10 percent becoming carriers. Even if the forms the outer coat of the hepatitis B virus. Hepatitis B surface duration of protection from HB vaccines were only seven years, antigen can either be purified from the plasma of HBV carriers children immunized early in life would be protected during the (plasma-derived vaccines) or produced in yeast or mammalian most critical period of HBV infection, and significant reductions cells by recombinant technology (recombinant vaccines). in HBV transmission, cirrhosis, and FHC would occur. Hepatitis B vaccines are highly immunogenic, even in new- Perinatal transmission is one of the most efficient and seri- borns, and can induce protective anti-hepatitis B surface ous modes of HlBV transmission (Stevens and others 1985). antibody in 90 to97 percent ofhealthy individuals, depending Perinatal transmission occurs from mothers who are positive primarily on the age of the recipient. for both the hepatitis B surface antigen (HBsAg) and the Hepatitis B vaccines have been successfully used in field hepatitis B "e" antigen (HBeAg). More than 90 percent of these trials in many parts of the world where the immunogenicity of women are chronic H-V carriers, although women acutely the vaccines in infants is usually measured at 95 to 99 percent. infected with the virus during pregnancy may also transmit to The protective efficacy of the vaccines against the develop- their children. Mothers who are HBeAg-positive carriers ment of disease or the carrier state is often 95 to 99 percent in have a 70 to 90 percent chance of infectinig their newborns cohorts of immunized infants. perinatally, and almost all these infected newborns become HBV carriers. Infected newborns rarely develop acute hepatitis, Plasma-Derived Vaccines although there have been several reports of fatal fulminant hepatitis. These carriers form a pool of infectious individuals In natural HBV infections, livercells produce much moreHBsAg who will infect others in the community and eventually their thani is needed to coat viral particles, and the excess HBsAg own offspring. Infants of mothers who are iaBeAg-negative forms 22-nanometer spherical and long tubular particles. carriers rarely become carriers through perinatal transmission. Plasma-derived HB vaccines are prepared by purifying HBsAg Hepaiitis B 323 particles from the plasma of HBsAg-positive donors. These yellow fever antigens. Similar data are available from labora- vaccines are inactivated to ensure that no infectious viral or tory animal studies as well as from human trials in Italy, other microorganisms are present, and then alum is added as Senegal, Myanmar, and China. an adjuvant. Plasma-derived vaccines, available since 1981, have an outstanding record of safety and efficacy and have Target Age been used in more than 70 million individuals. When HB vaccine is given to an infant of a carrier mother who Recombinant HB Vaccines is HBeAg positive, he or she has, in most cases, already been exposed to the virus, and the vaccine must provide postexpos- These vaccines are produced from HBsAg derived from yeast or ure prophylaxis. Plasma-derived HB vaccines alone have an mammalian cells in which replicating plasmids containing the efficacy of about 75 percent in preventing such infants from viral HBsAg gene are inserted. The HBsAg forms spherical becomingcarriers ifthefirst dose isgiven soon afterbirth. This particles similar to the natural 22-nanometer spherical particle may be feasible if infants are delivered in hospitals or clinics, in both chemical composition and immunogenicity. Recom- but it may be difficult to achieve at home deliveries unless they binant HBsAg for vaccines may be produced in almost unlim- are attended by midwives specially trained to administer vac- ited amounts in brewery-like fermentation vats, so there need cine or unless there is very rapid reporting of births to vacci- be no concem that lack of availability of antigen will compro- nators. A hepatitis B vaccine trial in Lombok, Indonesia, has mise future vaccine supply. Manufacturers could produce tens successfully stimulated birth registration and achieved a high of millions of doses in the next few years but will require firm level of immunization of infants within one week from birth commitments from vaccine purchasers before they make the in an area where home delivery predominates. In another trial, capital investments to produce the more than 300 million in Long An County, China, midwives have been successfully doses necessary to provide this vaccine to the world's children. trained to deliver the vaccine at the time of birth. Infants of mothers w,ho are not HBeAg-positive HBV carriers HB Vaccines and Immunization can receive the first dose of HB vaccine either near the time of birth with BcG, or with DPT-1, because many are protected by The single most important step in the global control of hepa- passive maternal antibody and because the risk of horizontal titis B will be the integration of these vaccines into the infection is low during the first few months of life. Expanded Programme on Immunization (EPI). This integration was recommended by the Technical Advisory Group on Viral Duration of Immunity Hepatitis and the Global Advisory Group of EPI in 1987. In 1991 the Global Advisory Group set targets for the introduc- The duration of protection from HB vaccines is a crucial issue tion of HB vaccine into national immunization programs which will be understood only by carefully following long-term (WHO/EPI 1992a, b, and c), and these targets were approved by HB vaccine trials. Cohorts of immunized adults and older the World Health Assembly in 1992. Targets call for all children followed for five to ten years show no evidence of countries with a prevalence of carriers of 8 percent or greater clinical hepatitis B, antigenemia, the development of elevated to integrate HB vaccines into routine infant immunization liver enzymes, or the development of the carrier state, despite programs by 1995: all other countries should have programs in declining levels of serum antibodies. It is clear that protection place by 1997. against disease outlasts detectable serum antibody levels, al- Integration ofHB vaccines into routine infant inmunization though some individuals have developed antibodies to the raises many practical questions which need to be addressed hepatitis B "core" antigen (anti-HBc), indicating that subclin- when the addition of a new antigen is considered. Hepatitis B ical infection has taken place. vaccines have characteristics which make them ideal for the It is unclear how long clinical protection against disease and integration into EPI. They are flexible enough to integrate into the development of the carrier state will persist. Some experts immunization schedules without requiring additional patient believe that long-term, even lifelong protection against signif- visits, do not interfere with the immune response to currently icant infection will occur following the immunologic "prime" used antigens, have an extremely low rate of unacceptable side provided by the initial vaccine series. Others think that loss of effects, are immunogenic from birth with no interference from protection against significant disease will occur at some point matemal antibody, and have shipping and storage character- and booster doses may be necessary. Further follow-up of istics similar to currently used antigens. immunized cohorts is indicated to answer to this question. Compatibility with Other EPI Antigens Number of Doses and Schedule Coursaget and coworkers (1986, 1990) have shown that EPI In early trials ofHB vaccines and initial licensure ofthe product antigens do not interfere with the immune response to HB for adults, vaccine schedules were used that were not necessar- vaccines, and conversely, that HB vaccines do not interfere ilv consistent with EPI schedules, and the vaccines were given with the response to BCG (bacille Calmette-Guerin), DPT (diph- at 0, 1, and 6 months or 0, 1, 2, and 12 months. For use in EPI, theria-pertussis-tetanus), or inactivated polio, measles, and the vaccines should be given during existing visits to avoid the 324 Mark Kane, John Clements, and Dale Hu expense and trouble of additional patient contacts. Fortu- frozen. At temperatures of 2 to 8 degrees centigrade, the nately, the vaccines have proven themselves to be extremely vaccines appear to be stable for many years. Some plasma- and flexible and capable of retaining their immunogenicity and yeast-derived products appear to be relatively stable at higher efficacy in virtually any EPI schedule (Hadler and others 1989). temperatures. This raises the possibility that HB vaccines could The first dose should be given with BCG near birth, if possible, be used in the field without a cold chain, something not or with the first dose of DPT if there is no immunization contact previously contemplated for other EPI antigens. Such a proposal at birth. The second dose should be given with the next dose needs careful field testing before widespread implementation, of DPT, and the third dose with the third dose of DPT or at the but it opens up the possibility that the vaccines could be carried time of the measles immunization. by those attending births in the home. Currently, EPI recommends that the vaccines be handled in Low-Dose and Intradermal Administration the same way as triple antigen (DPT), that is, kept between 2 and 8 degrees centigrade. A temperature-sensitive marker in Studies of healthy children and young adults in certain settings the vaccine vial or on the exterior of the vial would be a helpful have shown good immunogenicity of doses of HB vaccines addition to the product. The marker would indicate if the substantially lower than that for which the vaccines are li- vaccine had been frozen before use. Thus, with regard to the censed. Although there may be some savings in using one-half EPI cold chain, the vaccines are easy to integrate and need no or even one-quarter the manufacturer's recommended dose, new developments or conditions. there is concern that many vaccine recipients may not get an immunologically sufficient dose under conditions which may Storage Bulk be found in developing countries. These conditions include malnutrition, immunodeficiency, less than optimal adminis- Vaccines that are delivered in vials which contain multiple tration, missed doses, and schedules which are not maximally doses are less expensive per dose and require less storage space immunogenic. than those which are packaged with few doses per vial. Wast- Some investigators have attempted intradermal administra- age rates rise with increasing number of doses per vial, how- tion of approximately one-tenth the recommended dose of HB ever, and this becomes an important issue while HB vaccines vaccines in healthy adults and children; they have achieved remain more expensive than other EPI antigens. There is con- relatively good rates of seroconversion but substantially lower cern that the cold-chain storage space in some locations will geometric mean titers. Intradermal administration in infants be exceeded if programs adopt HB immunization plans using was less successful, with lower efficacy in infants of carrier vials containing few doses each. However, modeling done by mothers and reports ofdifficulty in administration and pain in EPI suggests that most cold chains could accommodate the the recipients. Additional concerns include reliability of per- addition of HB vaccines with little or no expenditure for addi- sonnel in administering intradermal injections, and the use of tional equipment. Currently available HB vaccines vary enor- extremely low doses of vaccines that may vary somewhat in mously in packaging volume, and the World Health potency. Intradermal administration ofHB vaccines is presently Organization (WHO) is working with manufacturers to develop not recommended by EPI and the Technical Advisory Group efficient packaging standards. Careful calculations will need to for Viral Hepatitis. be made to estimate space requirements before introduction of the vaccines. Immune Globulin Equipment If hepatitis B immune globulin (HBIG) is given to newborns of HBeAg-positive mothers in addition to HP vaccines, the effi- Hepatitis B vaccines are given by injection in the same manner cacy may be increased to 85 to 95 percent. Use of HBIG adds as other EPI antigens: no special equipment is needed. The use considerably to the cost of treatment because it is expensive of reusable equipment will require more episodes of steriliza- ($25 to $50 per child) and because it requires serologic testing tion and the replacement of equipment sooner. If disposable of mothers to determine their HBsAg status. Such testing is needles and syringes are used, three more per child will be itself expensive and requires laboratories and prenatal testing needed. These items will have to be budgeted for and supplied programs that are generally unavailable in developing coun- appropriately. tries. For these reasons it is generally accepted that it is more cost-effective to devote resources to routine infant immuniza- Future Strategies to Increase the Efficiency of HB tion and that most developing countries will elect to forgo the Immunization use of HBIG. Drop-out rates (parents not bringing their children for sched- Stability and Temperature Requirements uled immunizations) between the second and third doses ofDPT are significant, and a similar drop-out rate must be expected Both plasma-derived and recombinant types of hepatitis B with HB vaccines. Any strategy that might reduce this drop-out vaccines are adsorbed on aluminium salts. As with other such rate would be advantageous. One possibility is the develop- vaccine preparations, they should be protected from being ment of preparations which surround the antigen with a poly- Hepatiis B 325 mer which allows timed, slow release in pulses. This technol- and the Pacific Basin have begun infant immunization pro- ogy is, in theory, practical for HB vaccines and could mean that grams, which are presently at various stages. Countries with a single injection might be sufficient to immunize a child fully. more health resources have embarked on national programs, Such research underlines the importance of developing a sys- whereas those with fewer resources have begun demonstration tem which allows for the maximum number of antigens to be projects in selected areas with the intention of expanding them administered as early in life as possible. into national programs. Countries in this region which cannot Another way to increase the efficiency of immunization afford vaccines are looking for donor support to allow them to would be to use combined DPT and HB vaccines and DPT, IPV begin immunization. (inactivated polio vaccine), and HB vaccines; these are under development but will not be available for several years. Such Hepatitis B Virus Control in Africa vaccines would eliminate additional storage and delivery costs and would spare the recipient several additional injections. In In Sub-Saharan Africa, about 10 to 15 percent of the popula- Asia, an additional dose of univalent HB vaccine could be tion are HBV carriers, and about 70 to 95 percent of the delivered at birth to prevent perinatal infection. population show serological evidence of prior HBV infection. These prevalence figures are consistent in this region, and most Strategies for Control experts do not feel that serological studies need to be done in each country before HB vaccine programs are begun. It may be Recommended HB immunization strategies have differed in necessary, however, to do such studies to convince health various regions of the world because of the different epidemi- authorities that their country has an HIBV problem of this ological patterns of HBV infection. When the vaccine became magnitude. available in 1982, expert groups recommended universal in- The risk of a newborn infant acquiring HBV infection peri- fant immunization as the proper strategy for areas in which HBV natally in Africa is much lower than that of his or her Asian infection was moderately or highly endemic, and immuniza- counterpart (Mariner and others 1985). Although about 10 tion of "high-risk groups" as the recommended strategy for percent of women of childbearing age in Africa are HBsAg areas of lower endemicity. positive, only 10 to 15 percent of them are also HBeAg positive, Although high-risk individuals will undoubtedly benefit so only about I to 1.5 percent of African children are bom to from immunization, there is now considerable doubt from both HBeAg-positive mothers. For this reason perinatal transmission epidemiological and practical viewpoints that such high-risk- is much less common. In contrast, 70 to 90 percent of children group strategies will ever lead to a significant reduction of Hsv become infected before the age of seventeen. In many coun- infection on a national or intemational scale. It is likely that tries in Sub-Saharan Africa, PHC is the first or second cause of universal infant immunization is the proper strategy for long- cancer death in males. range control of HBV infection everywhere. Because perinatal transmission is uncommon, and because most African children are passively protected by matemal Hepatitis B Virus Control in Asia and the Pacific Basin antibodies for about six months, the timing of the first HB vaccine injection is not critical. About 95 to 98 percent of fully In the hyperendemic areas of Asia, about 7 to 10 percent of immunized children would be protected whether the first dose pregnant women are HtssAg-positive chronic carriers, and were given at birth or at three months of age. Several studies about 40 percent of these women are also HBeAg positive. are in progress in Africa which are designed to examine the Because the mothers of about 2.5 to 4 percent (40 percent of question of optimal timing of doses. 7-10 percent) of Asians are HBeAg-positive carriers, and be- cause about 10 to 15 percent of the population are carriers, it Hepatitis B Virus Control of Intermediate Endemicity Areas follows that 25 to 40 percent of carriers may have resulted from perinatal transmission. The majority of infection in the com- Less attention has been paid to HBV control in areas of inter- munity and the development ofthe carrier state in the majority mediate endemicity than in areas of higher or lower preva- of carriers are the result of childhood infection which is not lence. Public health officials in these areas often believe that perinatal. immunization of health care workers and matemal screening The need to treat infants of HBV-carrier mothers soon after with treatment of newborns of carrier mothers make up the birth poses a problem in areas in which there is no contact with proper strategy for control of hepatitis B. For reasons that will immunization services until several weeks or months after be discussed below, however, immunization of all newboms birth. In such areas children bom to carrier mothers who are with HB vaccine is probably the only strategy that will provide HBeAg positive may not receive protection from the vaccines. Iong-term control. Immunization programs in Asia will need to provide HB vac- cines near birth to maximize the prevention of HBV-carrier Hepatitis B Virus Control in Areas of Lou Endemicity children. The problem of HBV infection and its relation to PHC is well Since the availability of HB vaccines in 1981-82, strategies for understood by Asian health authorities, and HB control is high HBV control have stressed the immunization ofhigh-risk groups on their list of health priorities. Most countries in eastern Asia and the screening of pregnant women and treatment of the 326 Mark Kane, John Clements, and Dale Hu infants of HBV-carrier mothers. With few exceptions, the effect of carriers with chronic hepatitis. Treatment may cause an of this strategy has been the immunization of health care early seroconversion from HBeAg to anti-HBe, which may bring workers-about 85 percent of HB vaccines sold in the United about histological and biochemical improvement in chronic States and Europe has been used in this group. Although it is liver disease, but only a few patients lose their HBsAg, which certainly desirable to immunize these workers, cases of hepati- defines the carrier state. It is not known whether oncogenic tis B in health care workers represent fewer than 5 percent of progression to PHC will be affected by this treatment. The reported cases in the United States, and it is unlikely that treatment requires six months of thrice weekly interferon immunization of one small group will control HEv infection in injections, is expensive ($3,000 just for the drug), causes the community (Alter and others 1990). significant side effects in most patients, and is usually ineffec- Intravenous drug abusers, those who acquire HBV infection tive if the carrier state has been present since childhood, which through homosexual or heterosexual activity, and those who is the usual situation in developing countries. If a country finds belong to ethnic minorities in which HBV infection has a it difficult to prevent the development ofa carrier with vaccine higher-than-average prevalence are difficult to reach with for $3 per child, it will probably not be able to treat hundreds health care services and are often infected before they go to of thousands or millions of carriers for thousands of dollars any health setting where immunization could be offered (Kane each. and others 1989). For these reasons immunization of all infants and, in some areas, of adolescents for an interim period is now Cost-Effectiveness Considerations national policy in the United States (cDc 1991), Italy, and New Zealand and is being viewed by many experts worldwide Although there are numerous publications on the cost-effec- as the only strategy that will provide long-term control of HBV tiveness of HB immunization of health care workers and of infection even in areas of low endemicity. maternal screening for HBsAg in industrial countries, there are few published studies on the cost-effectiveness of universal Treatment of Chronic Sequelae of HBV Infection infant HB immunization in industrial or developing countries, and the studies which exist do not use the outcome measure of There is no effective treatment for PHc, which is essentially 100 disability-adjusted life-years (DALYs). Because immunization in percent fatal even with tertiary care in industrial countries. industrial countries is not the issue of concem in this collec- Death occurs within one to four months of presentation in tion, it will be mentioned but not discussed in detail. developing countries, so PHC itself does not carry with it pro- There are no effective treatment alternatives to immuniza- longed morbidity (Beasley 1988). MostpatientswhodieofPHC, tion, although in industrial countries patients with severe however, have underlying cirrhosis, and many patients have chronic disease or PHC may receive expensive palliative treat- had years of morbidity from their chronic liver disease. In ment or treatment designed to slow or alter progression of industrial countries, PHC patients are often treated with che- disease. No one would argue, from a medical, ethical, or motherapy, hepatic artery embolization, or surgery. Although economic point of view that we should consider not prevent- these measures can prolong life for several months at great ing this disease in favor of treating infected individuals. The expense, the cure rate is extremely low. effect of treatment on economic analyses is greatly to increase Alpha-fetoprotein is a host protein produced early and in the cost of not vaccinating in industrial countries. high quantities by PHC tumor cells. It is detectable with sensi- In industrial countries, with a relatively low incidence of tive assays, and trials are in progress to assess its usefulness as a infection and high vaccine cost, immunization of health care screening test for early detection of PHC at a stage when the workers (Mulley and others 1982; Lahaye and others 1987; tumor may be resectable. Trials in Alaska (McMahon and Koplan 1986), immunization ofother high-risk groups (Adler others 1990) and Shanghai have met with success in screening and others 1983), screening of pregnant women, and im- populations with this marker and resecting patients with early munization of newborns of carrier mothers (Arevalo and tumors. Japanese gastroenterologists use yearly ultrasound ex- Washington 1988; Kane and others 1988) have been shown aminations ofthe liver incirrhotic patients in asimilarattempt to be cost effective and even cost saving. The author of one to find curable early PHC tumors, which they treat with either study in Greece (Hatziandreu and others 1991) and the inves- surgery or ethanol injections. Alpha-fetoprotein and routine tigators of several as yet unpublished studies in the United ultrasound screening are expensive, require knowledge of who States have examined the cost-effectiveness and cost benefit in the population are carriers, and are not practical at this time of universal HB immunization in industrial countries of low and in developing countries. intermediate endemicity. (Personal communications [19921 There is also no practical treatment for chronic hepatitis or from Dr. P. Coleman, Centers for Disease Control; Dr. B. the carrier state in developing countries. In industrial coun- Bloom, University of Pennsylvania; Dr. R. Anderson, Univer- tries, liver transplantation is sometimes attempted to prolong sity of London; and Dr. G. Ginsberg, Jerusalem Ministry of the life of a cirrhotic patient or one with fulminant hepatitis, Health.) Taking into consideration the high cost of treatment but this is not a reasonable option in developing countries. and medical care in these countries, those researchers who Interferon treatment is being tried to alter the natural history attempted to do a cost-effectiveness analysis found that the Heparitis B 327 strategy of universal immunization of infants was cost effective favorably with data on other EPI vaccines from other countries and cost saving if the cost of vaccine for public sector programs (WHO 1982; Bamum and others 1980). was less than that charged in the private sector. We are aware of only one attempt at a cost-benefit analysis In 1985 the Institute of Medicine published a comprehens- of HB immunization in a developing country (Zhy Yi Xu and ive report on priorities for new vaccine development in devel- Richard Mahoney, personal communication, August 1992). oping countries (Institute of Medicine 1985). In its analysis, These researchers, who measured the cost benefit of routine HB the burden of disease from HBV infection ranked second to immunization in China, found an extremely high benefit-to- Streptococcus pneumoniae among all infectious diseases consid- cost ratio for vaccine use at a vaccine cost of approximately ered. The researchers estimated that, in 1985, 572,650 deaths $1.00 per dose. They also calculated the cost per DALY gained per year occurred from PHC, 246,239 deaths occurred from for universal HB immunization in China, which they found to cirrhosis, and 3,347 deaths occurred from acute and fulminant be $17.50 to $22.40, about one-half the cost calculated by HB infection. Unfortunately, they considered the probable cost Barnum and Greenberg (chapter 21, this collection). of HB vaccine to be $30.00 per dose and did not perform a There is an obvious need for additional cost-effectiveness sensitivity analysis on vaccine cost. Because the cost of HB and cost-benefit studies concerning the addition ofHB vaccines vaccine is now $0.65 to $2.00 per dose in developing countries, to routine immunization programs in other areas of the world, it would he useful to recalculate their outcome measures. and there is a need for researchers to convert their outcome For developing countries, the authors of two cost-effective- measures to DALYs so that results are comparable with other ness studies (Maynard, Kane, and Hadler 1989; Hu 1990) have interventions discussed in this collection. Nevertheless, the constructed decision trees to model the number of carriers and existing published and unpublished studies, whether based on deaths from liver disease that will result from doing nothing modeling or actual country data, are quite consistent in finding and from integrating HB vaccines into EPI programs. These that at present it costs only $20 to $60 to prevent, through a studies predict a cost per carrier prevented ranging from routine infant immunization program, the development of a $65.00 to $100.00 and an undiscounted cost per death pre- chronic carrier of hepatitis B in countries of high endemicity. vented ranging from $260.00 to $400.00 at a vaccine cost of This finding, plus the cost per death prevented, discounted or $1.00 to $1.40 per dose. It is assumed in the studies that the undiscounted, places HB immunization among the most cost- delivery cost of the vaccine will add an additional one-sixth to effective health interventions available. It is equally important the current EPI costs. In one analysis the undiscounted cost to realize that this intervention is possible within the existing per death prevented drops to $75.00 if vaccine cost drops to framework of national immunization programs which are ef- $0.50 per dose, a level probably achievable now in very large fectively delivering vaccines to 80 percent of the world's purchases. children. An important study of the cost-effectiveness of hepatitis B Hepatitis B is a disease with a burden of morbidity and immunization in the Gambia, the only African country with mortality comparable to measles, each having the potential to routine infant HP immunization, has recently been prepared causeupto3percentofthemortalityinapopulation.Hepatitis (Robertson and others 1992). The authors of this study used B vaccine, because it can be used at birth, is more effective actual costs from the Gambian EPI program, and they used the than measles vaccine. Deaths from chronic hepatitis, however, HBV prevalence and incidence rates as measured by the Gambia usually occur during the third to sixth decade of life and are Hepatitis Intervention Study. At an HB vaccine cost of $1 .00 heavily discounted by the DALYr method. Families in developing per dose, the marginal costs of adding the HB vaccine to the countries, however, might "discount" (in a noneconomic existing Gambian Fpi averaged $4.23 per fully immunized sense) the loss of a four-month-old child from measles and the child. The delivery costs, the cost of all inputs except vaccine, death of a forty-five-year-old father in a much different way, averaged only $0.28 per dose for a three-dose regimen. because the loss of the father could have dire social and The analysis revealed that at a vaccine cost of $1 per dose, economic consequences for the entire family. the cost per carrier prevented was $30 to $40, and the undis- counted cost per death averted was approximately $150 to Hepatitis B Vaccine and Future Immunization Policy $200. Discounting raised the cost per death averred to approx- imately $1,000 to $1,400. Because the carrier state develops Excellent HB vaccines have been available for ten years but are during early childhood in the Gatnbia, discounting is not a still not available to children in many developing countries. significant factor in the calculation of the cost per carrier There is a serious problem in our ability to integrate a devel- averted. These results compare favorably with the cost- opednewvaccine intothe international immunization system. effectiveness of other vaccines in the Gambia. The authors of The problem is not technical but relates to the willingness of the study note that the cost per carrier averted is higher than donors to increase their contribution to add vaccines to na- the cost of preventing a case of measles or pertussis but much tional immunization programs that are dependent on donor lower than the cost of preventing a case of neonatal tetanus, support. A discussion of this issue must consider the economics polio, or diphtheria (Robertson and others 1985). The cost- and politics of vaccine development and delivery, and the effectiveness of HB immunization in the Gambia also compared goals and realities faced by private sector companies, interna- 328 Mark Kane, John Clements, and Dale Hu tional and nongovernmental organizations, donors, and indus- WHO requirements and that uniform shipping and storage con- trial and developing countries. ditions were met. The Expanded Programme on Immunization is a network of Ministry of Health officials in countries that can afford the international, regional, national, and local immunization pro- vaccine are often confused by contradictory information grams that deliver vaccine to approximately 80 percent of the from manufacturers, local experts, international agencies, and world's children. Some of these vaccines are produced by nongovernmental organizations. They are variously advised laboratories in the public sector in various countries, and some to purchase plasma-derived vaccine, purchase recombi- vaccines are purchased from manufacturers by national gov- nant vaccine, import and repackage bulk vaccine, produce ernments, but approximately 50 percent of vaccines for chil- their own plasma vaccine or recombinant vaccine, and enter dren are purchased from private manufacturers by the United into regional production or procurement schemes. National Nations Children's Fund (UNICEF), which must raise money committees are often set up and given the task of sorting from donors to purchase this vaccine. The cost fully to immu- these options out, but the committees are often unable to come nize a child varies from country to country, but it averages up with acceptable recommendations even after years of about $15. Most of this cost goes for salaries, equipment, deliberation. training, and delivery of the vaccine: the cost of the vaccines In many developing countries, children do not get vaccine themselves is less than $1 per fully immunized child. unless it is provided by donors. The newer vaccines, with HB Most vaccines are produced by private manufacturers in vaccine as the prototype, will cost more per antigen than the industrial countries, who are in business to make a profit. Yet six vaccines currently supplied to EPI. Donors and international prices obtained by UNICEF are only pennies per dose. How can immunization authorities must increase the resources devoted such prices be obtained? Because manufacturers' books are not to the purchase of new vaccines and fund new vaccine devel- made public, the actual costs of production are proprietary opment if the world's children are to benefit from these tech- information, but these prices can reflect little more than the nologic advances. marginal cost of production plus the cost of the vial and packaging. Some manufacturers claim that they can only pro- Transfer of Technology vide low cost vaccine to UNICEF because they sell the same vaccine for much more in industrial countries. Indeed, in the A number of countries have expressed interest in local produc- United States and Europe, pediatric vaccines sell for between tion of HB vaccines. In theory, local production of plasma- $5 and $20 per dose. A second reason for the relatively low derived vaccines and the use of a readily available local cost is that these are older vaccines, whose costs for research, material, HBsAg-positive plasma, which is identified during development, and capitalization have already been recovered. blood donation and is otherwise discarded, could make vac- Manufacturers of hepatitis B vaccine claim that this is a cines available at an estimated cost of $0.10 to $0.40 per dose relatively new vaccine and that they are still recovering the (Mahoney 1990). The technology to produce plasma-derived very substantial costs of research, development, and marketing HB vaccine has been successfully transferred to China, which and the capital costs of the new production facilities. In has produced up to 20 million doses per year. Several countries, addition, some manufacturers pay a substantial royalty per dose including China, are exploring the possibility of transfer of to other companies who developed the clones used to make technology to produce recombinant HB vaccines. the vaccine. They claim that they cannot sell this vaccine at Critics of transfer of technology point out that there are few prices comparable to the other EPI antigens. In industrial coun- instances of successful transfer of technology of biologicals in tries HB vaccine sells for between $5.00 and $20.00 per pediat- the public sector. They also point out that current producers ric dose. In developing countries vaccine may be obtained for in industrial countries could produce large volumes of addi- between $0.65 and $2.00 per dose. It is unclear how low this tional vaccine at a marginal cost that is less than the cost of price would go if tenders for tens of million to hundreds of putting into place new transfer of technology schemes. There million doses were offered. is concem that although a few well-trained scientists and A vicious cycle exists between potential vaccine purchasers, engineers could produce a vaccine, the overall infrastructure such as UNICEF, and the manufacturers. Even at $1.00 per dose to ensure that high-quality vaccine is consistently produced ($3.00 per child), UNICEF claims, the vaccine cost per fully maynotexist.Theremaynotbeconsistentavailabilityofgood immunized child would more than double and thus it cannot water, electricity, trained technicians, and reagent kits for afford this vaccine. The manufacturers claim that without quality control. National control authorities with indepen- guaranteed substantial orders, they cannot make the invest- dent, high-quality testing facilities and trained personnel may ments to scale up production that would allow them to offer a also not be present. lower price. Pooling of orders from a number of countries and Proponents of the idea argue that although efficient produc- competitive bidding with large orders should lead to signifi- tion by a few large producers may be a good strategy in an ideal cantly lower prices, which in turn would allow more countries global economy, HB vaccines have been available for ten years, to afford HB immunization. Pooled procurement would also and most developing countries still cannot afford them. Hard ensure that participating countries received vaccine that met currency considerations make it such that countries that can- Hepatitis B 329 not afford foreign vaccines may be willing to spend local Beasley R. P., and L.-Y. Hwang. 1984. "Epidemiology of Hepatocellular currency to produce vaccine. Purchase from a few large pro- Carcinoma." In G. N. Vyas, J. L. Dienstag, and J. H. Itoofnagle, eds., Viral ducers also does nothing to reduce long-term dependence on Hepatitis and Liver Disease. Orlando, Fla-: Gnune and Stratton. cL)c (Centers for Disease Control). 1990. "Protection against Viral Hepatitis: foreign aid to supply vaccines. There has been much consoli- Recommendations of the Immunization Practices Advisory Committee dation among large pharmaceutical firms, and vaccine produc- (ACIP)." Morbidity and Mortality Weekly Report 39(RR-2):1-26. tion, not a very profitable enterprise compared with the . 1991. "Hepatitis B Virus: A Comprehensive Strategy for Eliminating production of other pharmaceuticals, could conceivably be Transmission in the United States through Universal Childhood Vaccina- dropped by many producers in the future. tion: Recommendations of the Immunization Practices Advisory Commit- Additional arguments in favor of transfer of technology tee (AcIP)." Morbidity and Mortatity Weekly Report 40(RR-13):l-25. include the economic benefits of developing a capability for Coursaget, P.. B. Yvonnet, E. H. Relyveld, J. L. Barrero, I. Diop-Mar, and ]. P. Chiron. 1986. "Simultaneous Administration of Diphtheria-Tetanus-Per- production of biologicals and recombinant biotechnology. tussis-Polio and Hepatitis B Vaccines in a Simplified Immunization Pro- The production cotild be placed in the private sector, where gram: Immune Response to Diphtheria Toxoid, Tetanus Toxoid, Pertussis, there may be more motivation for efficient operation. Excess and Hepatitis B Surface Antigen." Infection and Immunity 51(3):784-87. vaccine could be sold in the private sector or to neighboring Coursaget, P.,B. Yvonnet, E. Relyveld, A. Brizard, C. Bourdil, L. Bringer, E. countries to recoup part of the cost of production. Jeannee, S. Guindo, 1. Diop-Mar. and J. P. Chiron. 1990. "Simultaneous A significant obstacle to local vaccine production is the Injection of Hepatitis B Vaccine with BCG, Diphtheria, and Tetanus Tox- oids, Pertussis and Polio Vaccines." In P. Coursagel and M. J. Tong, eds., large initial cost of purchasing the technology, the equipment, Progress in Hepatitis B Immunization London/Paris: Colloque INSERNVJohn and the plant. International funding agencies and donors Libbey Eurotext Ltd. 194:319-24. should consider loans or grants to cover this period if an Hadler, S. C., M. A. de Monzon, D. R. Lugo, and M. Perez. 1989. "Effect of analysis of the economic aspects shows local production to be Timinig of Hepatitis B Vaccine Doses on Response to Vaccine in Yucpa reasonable, and if competent national regulatory authorities Indians." Vaccine 7:106-10. exist to ensure that a high-quality product is consistently Hatziandreu, E. J., A. Hatzakis, S. Hatziyiannis, M. A. Kane, and M. C. produced. In some countries, the level of technology and Weinstein. 1991."Cost-EffectivenessofHepatiLisBVaccinationinGreece: infrastructure seem adequate to produce an affordable, high- A Countri of Intermediate HBV Endemicity." lnternationalJournal of Tech- nology Assessment in Health Care 7(3):256-62. quality vaccine. Hu, D. 1990. "Integration of Hepatitis B Vaccine into the EPI" EPI/GAC/WP. 14. Working paper for the Global Advisory GrOup of EPI Cairo. Conclusions Instituce of Medicine. 1986. New Vaccine Development. Vol 11, Establishing Priorities. Diseases of Importance inDevelopingCountres. Washington, D.C.: Hepatitis B vaccines have proved themselves to be stable, safe, National Academy Press. immunogenic, and effective. There are no technical or scien- Kane, M. A., M. J. Alter, S. C. Hadler, and H. S. Margo[is. 1989. "Hepatitis tific impediments to their immediate use. Indeed, the dimen- B Infection in the United States: Recent Trends and Future Strategies for sions of the global problem of HBV infection make it clear that Control."AmericanJournalofMedicine 87(supplement 3A):lls-13s. Kane, M. A., Y. Ghendon, and P. H. Lambert. 1990. " 1990-Where Are We: their use is a matter of urgency. As with all vaccines, research The WHO Programme for Control of Viral Hepatitis." In F. B. Hollinger, S. and development efforts must continue to improve the vaccine NI. Lemon, and H. S. Margolis, eds., Viral Hepattis and Liver Disease: and its preparations, determine the most cost-effective strate- Proceedings of the 1990 Internantonal Symposium on Viral Hepantis and Liver gies for delivering it, and lower the costs of production. The Disease. Baltimore: Williams and Wilkins. final task before us is to develop national programs and obtain Kane, M. A., S. C. Hadler, H. S. NMargolis, and J. E. Maynard. 1988. "Routine financial resources to provide this important antigen to the Perinatal Screening for Hepatitis B Surface Antigen." jAMA 259:408-9. world's children. Koplan,J. P. 1986. 'AssessmentofNew Vaccines in Immunization Programs." lsraelJ(urnal of Medical Sciences 22:272-76. Lahaye, D., P. Strauss, C. Baleux, and W. Vav Ganse. 1987. "Cost-Benefit Anal'sis of Hepatitis B Vaccination." Lancet 2:365-69. References McMahon, B. J., S. R. Alberts, R. B. Wainwright, L. Bulkin, and A. P. Lanier. 1990. "Hepatitis B Related Sequelae: Prospective Study in 1400 Hepatitis Adler, M. W., E. M. Belsey, J. A. McCuitchan, and A. Mindel. 1983. "Should B Surface Antigen Positive Alaska Native Carriers." Archives of Internal Homosexuals he Vaccinated against Hepatitis B Virus? Cost and Benefit Medicine 150:1051-54. Assessment." British MedicalJournal 286:1621-24. Mahoney, R. T. 1990. "Cost of Plasma Derived Hepatitis B Vaccine Produc- Alter, M. J., S. C. Hadler, H. S. Margolis,W. J. Alexander. P. Y. Hu, F. N. tion." Vaccine 8:397-401. Judson, A. Mares, J. K. Miller, and A. Meyer. 1990. "The Changing Mariner, E., V. Barrois. B. Larouze, W. T. London, A. Cofer, L. Diakhate, and Epidemiology of Hepatitis B in the United States: Need for Alternative R. S. Blumberg. 1985. "Lack of Perinatal Transmission of Hepatitis B Virus Vaccine Strategies." JAM. 263:1218-22. Infection in Senegal, West Africa." Journal of Pediatrics 106(5):843-49. Arevalo, J. A., and E. Washington. 1988. "Cost Effectiveness of Prenatal Maynard, J. E., M. A. Kane, and S. C. Hadler. 1989. "Global Control of Screening and Immunization for Hepatitis B Vrus."JAMA 259:365-69. Hepatitis B through Vaccination: Role of Hepatitis B Vaccine in the Bamum, H. N., D. Tarantola and 1. F. Setiady. 1980. "Cost Effectiveness of an Expanded Programme on Immunization." Reviews of Infectious Diseases Immunization in Indonesia." Bulletin of the World Health Organization lI(Supplement3):S574-78. 58:499-503. Mulley, A. G., M. D. Silverstein, and J. L. Dienstag. 1982. "Indications for Beasley, R. P. 1988. "Hepatitis B Virus: The Major Etiology of Hepatocellular Use of Hepatitis B Vaccine, Based on Cost-Effectiveness Analysis." New Carcinoma." Cancer 61:1942-56. EnglandJournal of Medicine 307(11):644-52. 330 Mark Kane, John Cletnents, and Dale Hu Park in, D. NM., ed. 1956. Cancer Occurrence in Developing Couitries. Inrema- Stevens, C. E., P. T. Toy, Ni. J. Tong, P. E. Taylor, 0. N. Uyas, P. V. Nair, M. tional Agency for Research on Cancer (IARC) Scientific Publications 75. Gudavalli, and S. Krugman. 1985. "Perinatal Hepatitis B Transmission in Lyons. Distributed for IARC by Oxford University Press. the United States: Prevention by Passive-Active Immunization." JAMA Robertson, R. L., .J.A. Hall, P. E. Crivelli, U. Lowe, H. M. lnskip, and S. K. 253:1740&45. Snow. 1992. "Cost Effectiveness Analyses of the Expanded Programme on WHO (World Health Organization)/EPI (Expanded Programme on Immuniza- Immunization (EPI) and of the Addition of Hepatitis B Virus Vaccination tion). 1982. "Expanded Programme on Immunization Cost Effectiveness." to the EPI of The Gambia." Health Policy aid Planning (June). Weekly Epidemniologic Record 67(22):170-72. Robertson, R. L., S. O. Foster, H. F. Hull, and P. J. Williams. 1985. "Cost . 1992a. Weekly Epidemiological Record67(3). Effectiveness of Immunization in The Gambia."Jounial of Tropical Medicine . 1992b. Weekly Epidemiological Record 67(4). and Hygiene 88:343-51. . 1 992c. Weekly Epidemiological Reccord 67(5). PART THREE The Unfinished Agenda, II Reproductive Health and Malnutrition Excess Fetility Matemal and Periatal Health Protein-Energy Malnutrition Micronutrient Deincy Disorders 16 Excess Fertility Susan Cochrane and Frederick Sai The health priorities addressed in the other chapters of this ences indicate excess fertility of 30 percent. Not only do levels collection have all been conditions that cause debility or differ according to the definition, but the location of excess death. Among these conditions are maternal and perinatal fertility differs even more dramatically. Because of these very health problems. High fertilitv and close child spacing are a substantial differences we shall pay considerable attention to significant determinant of poor health of mothers and infants the issue of definition in this chapter. in the first week of life, as discussed in the chapter by Walsh Two important conclusions can be drawn from the analysis and others and in the recent review of the National Research in this chapter. First, improved spacing and the deferment of Council (NRC 1989). High fertility and close spacing also have birth until reproductive maturity is achieved are more import- consequences beyond the first week of life, at least up to age ant for improving child survival than are other high-fertility five, and have negative consequences beyond those immediate behaviors. Second, the societal economic benefits of reducing health consequences. They have negative consequences on fertility must be weighed against the costs of doingso, and these the health and on the economic and social well-being of the costs depend on the motivation of women to control their family by diluting the resources available for each child and fertility. Thus, the individual and societal measures of excess putting pressure on parents to work harder and save less. The fertility are linked. balancing of the costs and benefits of fertility to the woman In the first part of this chapter, we document the current and, to a considerable degree, to the larger household is cap- levels and trends of fertility in the various regions of the world. tured by her stated fertility preferences. High fertility may also We shall then use these levels to determine the levels of excess have negative consequences for society as a whole. Fertility fertility by different definitions from the point of view of beyond which such negative consequences occur is deemed to society as a whole. In the next part, we document the levels of be excess fertility. Such excess fertility is to be considered a excess fertility from the individual's perspective, and then we health prioriry, not only because many of the negative effects document the costs of excess fertility. Given the great discrep- are on health, but also because the delivery of family planning ancy between the measures of excess fertility, it is necessary to to prevent excess fertility is provided primarily through the provide a link between the measures that will provide useful health system and thus places claims on the system beyond policy guidance. This link will be made in the section on the those purely for health considerations. costs of fertility regulation. Although we give considerable The definition of excess fertility is, however, difficult, be- attention to the measurement of excess fertility. in the rest of cause some level of fertility is desirable from both the individ- the chapter we follow the outline laid out for the other chapters ual and the societal perspective. Excess fertility may be defined in the collection: the costs of excess fertility are examined, in several ways. From a health perspective, births to women strategies and costs of preventing excess fertility are estimated who are too young or too old, who are of too high a parity, or and case management is discussed, and finally funding and who have pregnancies too closely spaced increase their own research priorities are identified. risks of mortality and poor health and those of their offspring. Births that fall into any of these categories could be considered The Significance of Excess Fertility excess. From a societal point of view, population growth rates above 2 percent are considered by many economists to be The significance of excess fertility needs to be established to detrimental to development. Another way of defining excess determine its priority as a health issue. fertility would be to consider what women themselves or their husbands report as excess fertility. The level of excess fertility Levels and Trends in Fertility in the Developing World differs substantially among these definitions. Very crude appli- cation of the first two definitions gives an estimated excess The levels and trends in fertility in the developing world vary fertility of 14 to 25 percent of all births in the countries greatly between and within regions. In table 16-1, we report covered. The reports from individuals of their fertility prefer- regional averages of total fertility rates (TFR), the crude birth 333 334 Susan Cochrane and Frederick Sai rates (CBR), and rates of natural increase (RNIS) for the main AGGREGATE MEASURES OF EXCESS FERTILITY. At the societal regions of the world. In 1985, fertility in the developing world, level, excessive population growth may have a number of whether measured by the crude birth rate or the total fertility negative effects, in particular on economic growth and on the rate, was lowest in Latin America and the Caribbean and in environment. There is substantial debate on the effects of high Asia, where rates were almost identical. Fertility was highest rates of population growth on economic growth. (See the in Sub-Saharan Africa, where TFRS were almost twice as great section "The Consequences of High Fertility," below, for more as in Asia and Latin America. The Middle East and North detail.)Sixty-fiveof 131 developingcountries report that they Africa had rates closer to those in Africa than to the those in perceive that their population is growing too fast (United the areas of lower fertility. Nations 1988). Although each country has its own perception The regional averages hide substantial variation. In table of what rate of growth is too high, a rule of thumb that was 16A-I we provide levels and trends in fertility and the rate of developed in The World Bank's The World Development Report natural increase for selected countries of the developing (1984) was that a rate of natural increase in excess of 20 per world.' These data show that in Sub-Saharan Africa, not only 1,000, or 2 percent, was likely to be detrimental to economic are growth rates quite high, often in excess of 3 percent. but in development. many cases they have increased substantially since the early Even if it is agreed that a population growth rate above 2 1960s. The reason for this is that although death rates have percent is excessive, establishing a correspondence between fallen from very high levels, crude birth rates have fallen little the rate of population growth and the level of fertility is unitil very recently. Significant fertility declines have just difficult because population growth reflects both fertility and recently been observed in Zimbabwe, Kenya, and Botswana. mortality.4 An alternative method of estimation is to identify For Latin America and the Caribbean the pattern is different. excess fertility by the level of fertility that has negative health The rate of natural increase has fallen in almost every case from consequences. The accepted definition is that births too early 3 percent or higher to between 2 and 2.6 percent. This has (to women under eighteen), too late (to women over thirty- been accomplished by decreases in birth rates that were so four), too frequently (closer than twenty-four months apart), dramatic that they exceeded the great declines in death rates and too many times (more than four children) are likely to be documented elsewhere in this collection.! Asia shows a differ- detrimental to maternal and child health.5 The evidence on ent pattem with less uniformity. China, the Republic of Korea, these consequences will be discussed in detail below. In the and Thailand have had some of the most dramatic fertility aggregate we could say that, using the rules of thumb, fertility declines ever recorded, the TFR falling almost 60 percent during is excessive if the rate of natural increase exceeds 2 percent. if the period, whereas Nepal has recorded no decline. Growth the mother is younger than eighteen or older than thirty-four, rates have remained constant in Bangladesh, have risen in if the births for one mother are closer than twenty-four moniths, Nepal, and have decreased substantially in China, Malaysia, and if the births for one mother exceed four. the Philippines, Korea, and Sri Lanka. Although growth rates Although it is easy to determine the extent to which the have not declined dramatically yet, TFRs have fallen substan- population growth exceeds 2 percent, it is quite difficult to tially in Indonesia. The Middle East and North Africa display determine how many births represent health risks. We do not a mixed pattem as well, but on the whole fertility rates are know how many births are beyond the fourth parity, but if the higher than in Asia, there has been less decline, and growth average TFR is four, we know a substantial number of births are rates are high. With the exception of Turkey, fertility and of fifth parity or higher." We can also get very rough estimates growth rates are lower in North Africa than in the Middle East of the percentage of births to women who are too old or too for the countries in this sample. young from looking at age-specific birth rates. Although we will make rough estimates of the proportion of births that Estimating Excess Fertility represent health risks, precise estimates are possible only from individual level data sets, because many women and births are This section discusses various ways of measuring excess fertility. in more than one risk category. There are, however, only three Table 16- 1. Projected Fertility and Rate of Natural Increase 1985 2000 2015 Region TFR CHR RNI TFR CBR RN] TFR CBR RNI World 3.4 27 17 2.9 23 14 2.6 20 1 Industrial countries 1.7 13 5 1.8 12 -3 2.0 12 1 Nonmarketeconomies 2.3 17 7 2.1 15 4 2.1 14 3 Latin America and the Caribbean 3.6 29 20' 2.5 21 14 2.2 18 7 Sub-SaharanAfrica 6.4 46 31' 5.4 40 29 4.0 32 23 Middle East and North Africa 5.6a 40 30 4.3 32 24 3.1 26 19 Asia 3.3 27 18 2.6 21 13 2.2 17 9 a. Excess. Source: World Bank data. Excess Fertility 335 countries where individual level estimates have been made. In find that there would be excess fertility of 14 percent, 16 table 16-2, we summarize these estimates. None of the coun- percent, and 31 percent, respectively, by each definition.9 tries mentioned in the table has marriages involving very Looking at the individual countries one sees a pattern that young women, but some other countries would have a much is more mixed. In table 16A-I, the countries that have excess higher risk factor on this dimension. For these three countries, fertility by one definition or the other are noted. All countries the risk factors differ substantially because of definitions. The of Africa are noted as having excess fertility on measures of proportion who are at risk because of high parity reflects the both TFR and RNI. Excess fertility ranges from a 48 percent TFR level of the TFR, being highest in Kenya and lowest in the and a 105 percent RNI in Kenya to a low of about 30 percent Philippines.7 The proportion of births beyond four can be for both in Lesotho.IC In Latin America and the Caribbean calculated for selected countries for selected years from The only fourof thecountrieshaveTnRs abovefour, butall countries United Nations Demographic Yearbook (United Nations 1986). except Trinidad and Tobago have a RN[ of 2 percent or greater. These data are not generally available unless a system for Brazil, Colombia, Guyana, Haiti, and Jamaica, however, have registering vital statistics is in place or fertility surveys have rates of 2 or 2.1 percent. Thus it is the other countries that been conducted. The available survey evidence is given in have excess fertility of a substantial amount, ranging from over table 16-2. The United Nations (UN) data from registration of 30 percent in Paraguay to 20 to 25 percent in the Dominican vital statistics show that the percentage of births beyond four Republic, Ecuador, and Venezuela. In Asia, excess fertility is ranges from about 2 percent in Korea to 30 percent in Egypt greatest in Nepal and Bangladesh. India has marginal excess and Pakistan. The UN data for the Philippines shows a rate of fertility. Malaysia and the Philippines have TFRs below four but 3 percent, whereas the survey data yields 19 percent. (See NRC have a RNI in excess of 2 percent. All the countries of the 1989, p 79, for a table showing how this proportion has Middle East and North Africa in our sample except Turkey declined dramatically in those countries which have had large have excess fertility by both measures. The Middle East coun- fertility declines.) Using data from the World Fertility Surveys tries, however, have higher excess fertility than those in North (WFSS), Hobcraft, writing in 1987, (cited in NRC 1989) has Africa. calculated the proportion of women with two or more births By using these measures for individual countries, it is possi- in which the preceding interval was less than two years. This ble to estimate the percentage of births in each region which proportion ranged from over 50 percent in Jordan, Costa Rica, fall into the excess category. These estimates are presented in and Colombia to about 20 percent in Senegal, Lesotho, and table 16-3. The three measures give different estimates of the Korea. These last three countries show two patterns. The two magnitude of high fertility. The TFR and the RNI indicate that African countries have long breastfeeding and postpartum about 14 to 16 percent of the births in the developing world abstinence to prevent close spacing of children, whereas Korea outside China and India are excess. The data on births by age has very high levels of contraceptive use. of mother indicate that about 30 percent of the births are in Therefore, we can identify excess fertility as the number of high-risk categories. By these measures the greatest excess births needed to reduce the rate of natural increase to 2 fertility is in Sub-Saharan Africa, followed by the Middle East percent, to reduce the TFR to four, or to eliminate births at and North Africa. There is little excess in Asia and Latin high-risk ages. Using the aggregate data in table 16-1 and data America, where fertility rates have already fallen substantially, on age-specific fertility for the countries in table 16A- 1, we but a fifth to a quarter of births are to women under twenty or over thirty-four. Table 16-2. Percentage of Women with Various EXCESS FERTILITY AS REPORTED BY INDIVIDUALS. The usefulness High-Risk Factors for Another Birth of parents' stated fertility preferences has been questioned for (percent) many years and is still questioned by many. One argument is Risk Factor Kenya Philippines Zimbabwe that parents tend to rationalize their actual fertility and thus are unlikely to report that they want fewer children than they Too youing 1.4 3.6 4.4 already have. This may be true. Nonetheless, in survey after Too old 36.2 35.8 33.6 Too many births3 6.5 survey in recent years, many women have reported lower BirThs too soon 61.5 42.4 40.1 desired than actual fertility or that their last birth was un- Birhs OO OO 48.3 32.3 29.6 Any risk factor - 79.7 69.7 wanted. Even more report that they want no more children. In addition, many who do want more children want to wait a -Not available. significant period before the birth of their next child. Evidence Note: World Bank estimates for TFR: Kenya, 7.7; Philippines,3,9o s kn d can befsed the a.9t firs t ar imi o Eeces Zimbabwe, 5.4. of this kind can be used to get a first approximation of excess a. Under eighteen in Kenya and Zimbabwe, under twenty in the Philip- fertility. In table 16-4 we give the number of children desired pines. by women according to surveys conducted in the late 1970s h. Thirty-five and older. c. Four or more births in Kenya and the Philippines, five or more in and early 1980s. We also report the actual TFR and the TFR that Zimbabwe. would have prevailed if preferences were realized or if all d. Birth in the preceding twenty-four months in Kenya; open birth inter- unwanted births had not occurred." val of less than fifteen months in the Philippines; less than fifteen months potpartum and not pregnant in Zimbabwe. Several observations emerge from the various data. Family Source: Kenya and Zimbabwe, DHs reports. Philippinies, Casterline 1990. size preferences and the proportion of families who want no 336 Susan Cochrane and Frederick Sai Table 16-3. Excess Fertility, Measured by National Demographic Data (millions) Births in women under 20 and over 34 TFR > 4 RNI > 20 Region Percent Number Percent Number Percent Number Latin America 20 2,473,000 1 86 11 1,034,000 Sub-Saharan Africa 34 3,337,000 42 4,474,000 41 4,329,000 Middle East and North Africa 39 3,421,000 22 2,663,000 25 2,938,000 a Asia 26 3,686,000 9 1,409,000 7 1,138,000 a Total 31 12,917,000 14 8,632,000 16 9,439,000 Excess births (millions) 12.9 12,900,000 8.6 8,600,000 9.4 9,400,000 Total births for countries covered (millions) 42 42,300,000 59.0 59,000,000 59.0 59,000,000 a. Excluding China and India. In 1985-90, there were 113 million births in the developing world on average annually and 65 million outside India and China. Source: Authors' calculations from World Bank data. more children vary greatly from place to place. In Sub-Saharan gion except Sub-Saharan Africa. Overall, in the countries Africa fewer women say they want no more children, and the covered there were 11.6 million excess births in the average desired TFR (6.7 or 6.4) in table 16-4 is very close to the actual year in the late 1980s: 5 million in Asia outside China and TFR (6.9). Even so, there are differences among countries, India, 4 million in the Middle East and North Africa, and 2 particularly among younger women. Ghana and Lesotho re- million in Latin America. If these countries were representa- port substantially lower desired fertility among the youngest tive of the entire developing world except China, which is a women. In Latin America, actual (4.7) and desired (3.7 or 3.8) special case, there were 27 million excess births a year by the family sizes are much lower than in Africa, but excess fertility individual women's definition.'3 is greater. In Asia, desired (3.7 or 3.3) and actual (4.7) TFRS are The figures above, however, probably underestimate excess very close to those in Latin America and the Caribbean. The fertility for several reasons: first, as mentioned earlier, women Middle East and North Africa have the most varied pattem are somewhat reluctant to report desired fertility below actual; among countries, actual and desired fertility being very low in second, fertility preferences are declining in many cases faster Turkey and very high in the Republic of Yemen. On average than actual fertility, as indicated by the fact that the propor- the TFR is 6.2 for this region, and desired fertility ranges tion of women wanting no more children is increasing;i4 third, between 5.6 and 4.7. Therefore, using the conservative desired many women wish to space their births;"5 and fourth, to the fertility measures cited above, we find that in a perfect contra- extent that fertility preferences themselves reflect the cost of ceptive world, fertility could be lowered by at least one child contraception these preferences would be reduced by increased per woman in Latin America and Asia, by between 0.2 and 0.5 access. children in Sub-Saharan Africa, and by between 0.6 and 1.5 Another way of estimating excess fertility is to measure what in the Middle East and North Africa. fertility would be if contraception were perfect in all women In table 16-5 we report the proportion of women who want who wish to stop childbearing or postpone their next birth. no more children according to World Fertility Surveys (funded This is a much more difficult measure to obtain because of the by UNEP and USAID) and more recent Demographic and Health rarer data on spacing and the need to run population projec- Surveys (DHS funded by USAID). We also give the proportion tions with different usage levels. who wish to postpone their next birth among those who do A possible way of analyzing the extent of unwanted fertility want more children. In every case in which there are data from is through model populations. In figure 16-1 we plot the two points in time, the proportion who want no more children relationship between the mortality level and the proportion of has increased over the period. Kenya, where the proportion women who want no more children. Forty percent of the wanting no more children increased from sixteen to forty-nine, women in the countries with a crude death rate of ten or below is the most dramatic example. want no more children, with the exception of Paraguay, where In table 16-6 we sum the individual measure of excess only a third want no more. The pattern is less uniform for the fertility by region. This estimate is obtained by taking the high-mortality countries. Twenty percent or less of the women percentage difference between the actual TFR and the desired in all the Sub-Saharan countries want no more children.6 This number of children and adding any births reported as un- also applies to Yemen, where 19 percent want no more. In most wanted and weighting it by the number of births in that non-African high-mortality countries 30 to 40 percent of the country in a recent year. For the countries covered, 30 percent women want no more children. Thus three model country of the births are considered excess by the individual women types would be needed: high-mortality African countries and themselves. This proportion exceeds 30 percent in every re- probably also high-mortality Middle Eastern countries, high- Excess Fertilitv 337 Table 16-4. Preferred Family Size and Total Fertility Rates in Relation to Desired Family Size Total fertility rate Desired family size Preferred family size Desired family size exceeded or last Women Women All Usual TFR exceeded and birth birth unwanted Country 15-19 45-49 women (no birth deleted) deleted and birth deleted Africa Benin 7.2 8.0 7.6 7.3 7.3 6.9 Cameroon 6.5 8.6 8.0 6.4 6.1 6.1 C6te d'lvoire 7.5 9.6 8.4 7.2 7.2 7.0 Ghana 5.2 7.3 6.0 6.1 6.0 5.6 Kenya 6.6 8.7 7.2 7.9 7.6 6.9 Lesotho 5.6 7.3 6.0 6.0 5.6 5.3 Mauritania 8.3 9.4 8.8 7.5 7.1 6.8 Senegal 8.3 8.4 8.3 7.1 6.9 6.7 Sudan (North) 5.4 6.5 6.4 5.6 5.0 4.8 Latin America and the Caribbean Colombia 2.7 5.7 4.0 4.6 3.4 2.6 Costa Rica 3.5 6.1 4.7 3.5 3.0 2.6 Dominican Republic 3.4 6.0 4.7 5.2 3.8 3.0 Ecuador 3.1 5.6 4.1 5.2 4.1 3.1 Guyana 3.4 5.9 4.6 4.4 3.8 2.8 Haiti 2.8 4.3 3.6 5.6 4.3 2.8 Jamaica 3.3 4.8 4.1 4.4 3.4 2.3 Mexico 3.8 5.8 4.4 5.7 4.5 3.6 Panama 3.4 5.1 4.3 4.2 3.9 3.4 Paraguay 3.7 7.1 5.2 5.0 4.5 4.2 Peru 3.1 4.6 3.8 5.3 3.5 2.6 Trinidad and Tobago 3.2 4.8 3.8 3.2 2.5 2.4 Asia Bangladesh 3.7 5.0 4.1 5.4 4.6 3.1 Indonesia 3.3 5.4 4.2 4.3 4.0 3.6 Korea, Republic of 2.7 3.8 3.1 3.9 2.8 2.5 Malaysia 3.9 4.5 4.3 4.5 3.3 3.1 Nepal 3.6 4.3 3.9 6.1 5.4 4.5 Philippines 3.0 5.6 4.3 5.1 4.1 3.6 Sri Lanka 2.6 4.8 3.7 3.4 2.9 2.2 Thailand 2.9 4.4 3.6 4.3 3.2 2.6 Middle East and North Africa Egypt 4.2 4.7 4.1 5.0 3.6 3.1 Jordan 4.9 7.5 6.2 7.0 6.0 5.1 Morocco 4.3 6.6 4.9 5.5 4.4 3.7 Pakistan 4.0 4.5 4.2 6.0 4.3 3.9 Syria 5.0 7.1 6.1 7.4 6.3 5.6 Tunisia 3.7 4.4 4.1 5.5 4.1 3.6 Turkey 2.8 3.1 3.0 3.8 cc 2.4 Yemen, Rep. of 4.5 6.9 5.5 8.9 8.2 7.4 Note: Preferred Family size based on direct question; TFR based on synthetic cohort estimates of desired stopping points. Source: Lightbourne 1987. mortality Asian and Latin American countries, and low-mor- underlying these models are fairly straightforward. Three pop- tality countries. Three artificial countries have been created ulation types were developed of I million population each in to represent these types of countries; high-mortality Sub- 1990. They all had age structures, age-specific fertility and Saharan countries, Libana; high-mortality Latin American or mortality rates, and patterns of marriage, breastfeeding, and Asian countries, Banglapal; and low-mortality countries of contraceptive use specific to real countries of a general type. Asia and Latin America, Colexico. The number of births per year with current contraceptive use The excess fertility in these three model countries of a was projected for 1990 and 2000 as a base case. Then two million population is given in table 16-7. The assumptions alternative projections were made. In the first, all women who 338 Susan Cochrane and Fredenick Sai Table 16-5. Currently Married Women Who Want No More Children or Who Wish to Postpone ChiLdren (percent) Want no more Wish to postpone' Counrr'q WFS DHS WFS/CPS DHS DHS Year Africa Benin 8 - 55 - 1982 Botswana - 33 - 55 1988 Burundi - 24 76 1987 Cameroon 8 - - - 1978 C6te d'lvoire 4 - 38 - 1980 b Ghana 12 23 - 70 1978/1988" Kenya 16 49 68 1978/1989 Lesotho 14 - - - 1977 Liberia - 17 - 48 1986 Mali 17 - 50 1987 Mauritania 11 - - - 1981 Morocco 42 48 - 53 1987 Nigeria (Ondo State) - 23 - 58 1986 Senegal 6 19 - - 1986 Sudan 15 - - - 1978 Togo - 25 - 71 1988 Tunisia 48 57 - 64 1978/1989" Zimbabwe - 33 - 61 1988 Latin America and the Caribbean Bolivia 72 - 48 1989 Brazil - 60 - 64 1986 b Colombia 61 69 - 64 1976/1986 Costa Rica 52 - 77 - 1976 b Dominican Republic 52 63 - 51 1975/1986k Ecuador 55 65 - 65 1979/1987b El Salvador - 63 - 68 1985 Guatemala - 47 - 68 1987 Guyana 54 - Haiti 42 Jamaica 48 Mexico 56 65 33 1976/1987 Panama 63 - Paraguay 32 - 68 - 1979 Peru 61 70 - 60 b Trinidad and Tobago 46 55 - 55 1977/1987 Venezuela 55 - - Asia Bangladesh 30 - 72 - 1976 b Indonesia 40 51 - 73 1976/1987 Korea, Republic of 74 - 36 - 1974 Malaysia 46 - 36 - 1974 Nepal 30 - - - 1976 Philippines 54 - - - 1978 Sri Lanka 62 65 - 60 1987 Thailand 61 66 - 61 1987 Middle East and North Africa Egypt 54 61 - 51 1988 Jordan 42 - - - 1976 Pakistan 43 - - - 1975 Syrian Arab Rep. 38 7.1 6.1 7.4 1978 Yemen, Rep. of 19 6.9 5.5 8.9 1979 - Not available. a. Percentage of women who wish to postpone children for two or more years among those who want more children. b. WFS/DHS. c. Does not include currently pregnant women. Those who wish to postpone is defined as the percentage of women who prefer to wait one or more years among those who want more children. d. Does not include currently pregnant women. Source; WFS/DHS Surveys. Excess Fertilit 339 Table 16-6. Total Excess Fertility, Measured from limiting is less important than spacing in the high-mortality Individual Responses countries. Excess over desiredfertility Excess of TFS ABORTION AS AN INDICATION OF EXCESS FERTILITY. The most Region (number) (percent) extreme statement that a woman can make about excess fer- tility is to seek an abortion. It was because of the high incidence Sub-Saharan Africa 467,000 9 of complications from illegal abortion that the family planning Middle East and North Africa 4,088,000 35 movement began in the United States and many countries of Asiaa 5,062,000 31 Latin America. Because the data are of such poor quality, measures of the extent to which actual births exceed desired Total excess births 11,632,000 30 births cannot be derived from estimating the magnitude of Total excess births among abortion. The data are sufficient to give an indication of the births_______covered _______38____971____000 ____n___a__ extent to which current programs delivering contraception n.a. Not applicable. have been insufficient for controlling fertility to levels desired a. Excluding India and China. by women. (This issue is also treated under the section on case Source: Author. management.) The worldwide estimate of the number of abor- tions is between 40 million and 60 million (Henshaw 1987). claimed that they wanted no more children were assumed It is estimated that at least 14 million occur in China and 11 to be using perfect contraception.'7 The second alternative million in the countries of the former U.S.S.R. Other indus- assumed that, in addition to those who wish to limit fertility, trial countries contributed about 4.5 million abortions. Esti- one-half of those women who wish to space their next birth mates of abortion in developing countries are much less precise are using perfect contraception. " It is surprising that by this because abortions are more likely to be illegal there than in measure high-mortality African countries have somewhat industrial countries. In developing countries, abortion appears higher excess fertility than the high-mortality non-African to be highest in Latin America and Asia and lowest in Sub- countries. The reason for this is that current contraceptive Saharan Africa, where, as indicated above, desired family sizes use in Libana is much lower than in Banglapal. If spacing are much greater. There is evidence that there are substantial demand for contraception is included, Libana and Colexico differences in the incidence of abortion in the countries of have almost 40 percent excess fertility and Banglapal 26 Africa and that such incidence is increasing. percent. The relative importance of unmet need for contra- ception for spacing and for limiting, however, shows that The Consequences of High Fertility The consequences of high fertility are many, ranging from Figure 16-1. Relationship between Life Expectancy health consequences for mother and child to consequences for and Proportion of Women Wanting No More Children economic development. Percentage wanting no more children HEALTH CONSEQUENCES OF HIGH FERTILITY. The health con- 80 sequences of high fertility for mother and newborns are dis- cussed in another chapter of this collection. In this chapter, 70 . Korea we will discuss the health consequences beyond the first week cOlombia Panama of life. There is little debate in the literature that mortality of 60 Thailand *. *Pe Srimanka neonatal and postneonatal infants, and of children is positively Eypt Ecado0 Mexco la correlated with women giving birth at too young an age, too 50 - PhTuPnPisinae;/ Roepublic .*oaRaica old an age, too closely together or too many times. There is Pakistan * . Morroco Maaysia ;rmnidad substantial debate about whether increased contraception will 40 Halndonesit Jyria n improve survival rates (Trussell and Pebley 1984; Bongaarts, Banglades . Mauldin, and Phillips 1987, 1988; Potter 1988; Trussell 30 - Nepal * *a/ Paraguay 1988).'9 In this section, we will first examine the evidence of 20 Yemen an association between high fertility and infant and child 2$udcn Kenya Percent waning no more mortality. We will then discuss upper and lower limits on the Maritania. Lexsotho 10 * Ghana = -73.7 + 1.86 LE costs of the deaths that may be averted by reduced fertility. nea Benin* * Cameroon R2 0.61 The hypothesized effects of high fertility on the survival of 0 , Ivory Coast , offspring arise from both biological and socioeconomic and 40 50 60 70 80 90 behavioral factors. The biological factors arise most noticeably Life expectancy in the period immediately after birth and are assumed to explain why both very early and very late childbearing are detrimental to children as well as why close spacing may be Source: Authors' calculations, problematic. High parity and close spacing are also believed to 340 Susan Cochrane and FredeTick Sai Table 16- 7. Excess Fertility for Three Model Countries 1990 2000 Scenario Births Excess (percent) Births Excess (percent) Libana-High-mortality African or Middle Eastern country 1. Current conception prevalence rate (CPR) 54,010 n.a. 74,810 n.a. 2. All who want no more children (using contraception) 46,470 n.a. 64,090 n.a. 3. Line 2 plus half of spacers 33,975 n.a. 46,655 n.a. 4. Excess: lines 1-2 7,540 14 10,720 14 5. Excess: lines 1-3 20,035 38 28,155 38 Banglapal-High-mortality Latin American or Asian country 1. Current contraception prevalence rate (CPR) 46,540 n.a. 62,400 n.a. 2. All who want no more children (using contraception) 42,900 n.a. 57,450 n.a. 3. Line 2 plus half of spacers 34,635 n.a. 46,260 n.a. 4. Excess: lines 1-2 3,640 8 4,950 14 5. Excess: lines 1-3 11,905 26 16,140 26 Colexico-Lou-mortality Asian or Latin American country 1. Current contraception prevalence rate (CPR) 33,565 n.a. 43,031 n.a. 2. All who want no more children (using contraception) 26,397 n.a. 33,766 n.a. 3. Line 2 plus half of spacers 20,740 n.a. 26,434 n.a. 4. Excess: lines 1-2 7,168 21 9,265 22 5. Excess: lines 1-3 12,825 38 16,597 39 n.a. Not applicable. Note: Calculated using target models with constant levels of proximate determinants of fertility, except contraceptive use. Current contraceptive preva- lence was estimate to calculate number of births. The effectiveness was rhe level currently observed with the existing method mix. Assumes that all women who want no more children and half of those who wish to space children use perfectly effective contraceptives. Source: Authors. have the effect ofdiluting the household resources of maternal parity seven or more are less disadvantaged in Asia and the time and attention as well as family economic resources. Middle East than elsewhere, but even there, there are excep- Therefore, we should expect different patterns of effects de- tions and the mortality of these high-parity births is more than pending on the age of the child as well as on the environment. 60 percent higher in Yemen and the Philippines than that for Data from the World Fertility Survey has been extensively children of second or third parity. analyzed by Hobcraft, McDonald, and Rutstein (1983, 1985) Because of the much higher mortality of first births than any to show the relationships between childbearing patterns and other group, the reduction of fertility through the reduction of the survival of offspring. In table 16-8 we show the average of higher-order births will not necessarily reduce infant mortality. thirty-five developing countries of the percentage increase in This is so because a larger percentage of births will then be first death rates from various reference groups for various categories births. We have found no conceptual resolution of this appar- of births. These estimated effects have been controlled for ent paradox (Bongaarts 1988 and Trussell 1988). maternal age, spacing, parity, sex of the child, and the educa- The effects of matemal age are also nonmonotonic. The tion of the household and thus cannot be attributed to different mortality of children of mothers under twenty is 30 to 60 fertility pattems of women of different educational groups. percent higher than that of children whose mothers are (Detailed estimates for each country in the study are reported twenty-five through thirty-four. Other evidence suggests that in tables 16A-2 through 16A-5.) for children of women under eighteen these risks are even The effects of birth order generally come to mind when the greater. The risks alsodifferbetween regions and are somewhat effects of high fertility are mentioned. These effects are non- less elevated in Africa than elsewhere. Even there, there are inonotonic and nonuniform across age groups. The first year of countries, such as Tunisia, where the mortality risks of the life has the highest risk for first births. The risk of dying for first neonates of these young mothers are greater than for the births is 80 percent higher than for second and third births in neonates of older mothers. These high risks persist until the the neonatal period and 60 percent higher in the postneonatal child reaches two. The different patterns of risk for very young period. The fourth through sixth births do not show elevated mothers reflect in part the selectivity of women who bear mortality in the first year of life but do have 20 to 30 percent children at these young ages, which may not be completely higher mortality in the second through fourth years. Births captured by the control for education. Mortality risks are also beyond six have 20 to 30 percent higher mortality at all ages. higher for the children of women over thirty-four, but again These pattems do vary substantially from country to country. this varies by region, being greatest in the Americas, less in Fourth through sixth births tend to be more disadvantaged in Africa, and much less in Asia. Postponing the births of women Latin America and the Caribbean than elsewhere. Births at until they reach the age of twenty and reducing the births after Excess Fertlity 341 the age of thirty-four could possibly make important contribu- an important correlation with higher mortality of toddlers and tions to reducing mortality in a number of countries, all else young children. One or more live births in the twelve months being equal, but until more is known about who has children after a birth raises toddler mortality by about 130 percent and at these ages and the behaviors that affect their survival child mortality by 40 percent (tables 16A-1 through 16A-5). chances, it is difficult to judge the magnitude of effects. Thus, the effects of spacing on child survival are both It has been hypothesized that close spacing of children is stronger and more consistent across countries than are the detrimental to their health for two reasons. First, the biological other effects of high fertility (Hobcraft, McDonald and Rut- effect on the mother of close spacing of children leads to stein 1985; NRC 1989). For this reason, to get a rough estimate depletion of her health and her ability to nurture and bring to of the effect of family planning on child survival, we will term a baby of normal weight and her ability to breastfeed that concentrate on the effect of spacing on mortality. These esti- baby adequately.2' The second factor is the competition for mates are tentative at best. We will estimate the effect of one household resources such as the mother's time and attention year of contraceptive protection. Assuming an average of three and the household's economic resources. From the analysis in months of natural protection from conception following a table 16-8, it is clear that having a birth in the twenty-four birth and nine-month gestation, it would take only twelve months prior to the birth of the studied child raises substan- months of contraception protection to extend the birth inter- tially the child's risk of death. If that previous child survived, val to two years. In table 16-9 we summarize for a sample of mortality of the subsequent child is increased by between 50 countries the number of deaths averted in the first five years and 90 percent, depending on the age group under consider- of a child's life from 1,000 years of contraceptive protection. ation. These relative risks show more consistency than those The first estimates of deaths averted, method 1, are arrived at for parity and maternal age, especially for infants. There are, by comparing actual mortality of those children where there however, substantial differences between countries. The rela- was no living birth in the preceding twenty-four months with tive risks in North Africa and the Middle East seem to be themortalityofchildrenwhentherewasonesurvivingchild.2' particularly high because of close spacing. The elevation in The second estimate (method 2), which is generally smaller, mortality is much more dramatic if a previous child died, but is based on the Hobcraft and others (1985) estimates that this does not necessarily so much reflect the effect of the short control for age, parity, education, and spacing simultaneously. birth interval as the high clustering of deaths within house- The low costs per death averted through the family planning holds. Close spacing of births after the birth of a child also show of one year are estimated on the basis of 100 percent effective- Table 16-8. Percentage by Which Mortality Rates of Births in Various Categories Exceed Those of Reference Group Births in previous 24 months' Paritlb Mother's age' Child One alive One dead Two or more 4-6 7 or more 1 20 or less 20-24 35 or more Neonatal (0-1 month) Africa 69 278 289 1 27 114 18 0 18 Asia 82 259 362 -12 0 98 37 6 3 Latin America 53 319 209 15 44 40 37 12 27 Developing world 70 290 250 0 20 80 30 10 20 Postneonatal (I -12 months) Africa 65 261 233 7 27 49 38 15 0 Asia 100 215 384 2 31 70 32 6 -3 Latin America 99 256 203 3 43 62 65 27 7 Developing world 90 240 240 0 30 60 40 20 0 Toddler (1-2 years) Africa 19 97 62 17 14 -I 43 15 -4 Asia 83 145 81 25 31 5 44 -1 -7 Latin America 27 62 99 23 45 -8 98 46 11 Developing world 40 100 80 20 30 0 60 20 0 Child (2-5 years) Africa 48 67 38 18 18 -5 35 10 12 Asia 42 67 124 29 17 -12 52 16 22 Latin America 59 47 23 35 42 0 25 3 -15 Developing world 50 60 60 30 30 -6 40 10 10 a. Reference category is no birth. b. Reference category is parity oif two or three. c. Reference category is age 25-34. Source Derived from Hobcraft. McDonald, and Rutstein, 1985. 342 Susan Cochrane and Frederick Sai ness and $7.7022 costs per couple-year protection.23 Because veloping countries" (NRC 1986, p. 90). The 1984 World Devel- each year of couple protection does not result in a prevented opment Report concluded, "In short, policies to reduce popula- birth, these figures are then adjusted for the average annual tion growth can make an important contribution to births per woman of childbearing age in the country under development (especially in the long run), but their beneficial consideration. The high cost estimate assumes $18.00 per effects will be greatly diminished if they are not supported by couple-year of protection.24 These are then converted into the right macroeconomic and sectoral policies. At the same disability-adjusted days of life gained. time, failure to address the population problem will itself There are substantial differences in the costs per death reduce the set of macroeconomic and sectoral policies that are averted and the costs per disability-adjusted life-year gained. possible, and permanently foreclose some long run develop- Not surprisingly, the lowest costs are in the highest mortality ment options" (wDR 1984, p. 105). countries in the group, Pakistan and Bangladesh. More surpris- The costs of excess fertility, like its measurement, can be ingly, Egypt, which has a life expectancy at least five years viewed by the family or society. The costs of children are above Pakistan, has similar costs, and Kenya, with life expec- primarily borne by the family. Therefore, the family's report of tancy 3.5 years less than Egypt, has costs 50 percent higher. what is the desirable family size would incorporate the family's Diarrheal diseases may be an important factor in explaining judgment of the desirable expenditure on children. The costs these different patterns, because close child spacing may be to the family of excess fertility would be measured by the costs linked to early weaning and higher mortality risks. of the marginal child to the family. Unfortunately, due to both In summation, increased spacing of births probably repre- methodological and data problems, the measurement of the sents the most important way to reduce mortality through costs of a child is rarely available for developing countries.26 In family planning. Elimination of births at the youngest age additiontothecostsoffood,clothing,medicalcare,schooling, groups is probably the second most important factor in reducing and housing, children in the family affect the amount of time infant and child mortality. High parity and births to women over a mother devotes to child care. These costs are borne by a thirty-four have an effect that is less well established. number of adjustments in the household. Expenditure per child is reduced with increasing numbers, resulting in many SOCIALANDtECONOMICCOSTSOFEXCESSFERTILITY. Population cases of poor health in the children (the excess child and its growth affects economic development through its effects on siblings) and reduced school participation of all children. In savings and investment, technological change, changes in addition to reduced expenditure per child, high fertility also 25 efficiency, and returns (increasing or diminishing) to scale. Z results in efforts to increase family income.27 The increase may In addition, population growth affects the resource base come about through child labor or through added labor of the through resource depletion and pollution. The precise rela- parents. An interesting body of evidence is accumulating on tionship between any one of these and population growth has the higher labor participation of men in households with larger yet to be firmly established and depends on current levels of numbers of children. Finally, the adjustment to higher un- population density, resource endowment, and the rate of pop- wanted fertility can be made by reducing the savings of the ulation growth as well as on myriad policies from property household. (See King 1987 and Cochrane, Kozel, and Alder- rights in resources to agricultural subsidies (Kelley 1988; The man 1990 for reviews of these issues.) The bottom line is that Worl Development Report 1984; and NRC 1986 for comprehen- it is impossible to document the negative effects of high sive reviews). fertility on every dimension or every country. Thus there is not The National Research Council concludes, "On balance, one cost of a child that can then be used to evaluate the savings we reach the qualitative conclusion that slower population from avertinga birth for the individual household.28 Therefore, growth would be beneficial to economic development of de- it must be left to the household to establish its own evaluation Table 16-9. Infant and Child Deaths Averted by 1,000 Couple-Years of Protection through Child Spacing and Associated Costs, Selected Countries Cost per DALY saved (1988 dollars) Deaths averted Actuala Adjusted" Country Actual5 Adjustedb Low cost High cost Low cost High cost Pakistan 85 63 405/41 944/33 544/19 1,268/44 Mexico 40 34 1,532/53 3,570/123 1,794/62 4,189/144 Bangladesh 96 125 445/15 1,037/36 341/12 795/27 Philippines 40 38 1,667/58 3,907/135 1,764/61 4,110/142 Kenya 61 64 784/27 1,827/63 746/26 1,738/60 Egypt 111 131 483/17 1,125/39 413/14 962/33 a. Actual differences in mortality rates for births with and without a birth within previous rwenthy-foir months. Other method uses adjusted rates. b. Estimates controlled for the education, and parity of the mother. Source: Author. Excess Fertility 343 of the costs and benefits of children. These evaluations are and mental health (David 1986). Other data on how parents reflected in their stated fertility preferences. The fact that at treat unwanted children have been documented by Shorter the family level the costs exceed the benefits in many cases is (1976), Ware (1976), and Scrimshaw (1978 and 1983). Evi- revealed by the evidence cited above that 30 percent of births dence of infanticide, abandonment, neglect, and the selective are in excess of stated family size preferences. provision of food and medical care to children has been drawn The evidence above refers to the effect of an added child on from historical and contemporary data from all geographic the family. If that child is unwanted, the negative effects, regions. Further evidence of the effect of unwanted children particularly for the child, are probably much more severe even comes from a recent study in Ethiopia. It was reported that though they are less well documented. Some work has been women whose births were described as unwanted were least done on this for the industrial world as part of the justification likely to seek antenatal care (Kwast and others 1985). for publicly subsidized family planning programs. For example, The unwanted fertility of the unmarried woman is even women denied abortion in Prague were followed up for a period more costly for both the mother and the potential child. of twenty years. Forty-five percent of these mothers were Women lose out on educational and employment investments dissatisfied with their child's development compared with 21 and are forced to choose between abortions (safe or unsafe), percent among controls. The children themselves perceived fostering out or adoption of children, and raising the child more problems in life and more disappointments in life, love, without economic support from the father. Little or no eco- nomic analysis has been done of the economic costs of fertility outside marriage in the developing world. The medical costs Table 16- 10. Savings per Birth Averted in Three of unsafe abortion are discussed later in the chapter. Types of Countries Society also bears some costs as a result of the birth of a child. (1987 U.S. dollars) The most obvious costs are those for education and health. If Expenditure 5 percent discount I0 percent discount the state also takes on responsibility for food, shelter, safe water, and so on, through subsidies or public provision, the Libana-high-morrality costs are commensurately higher. In the late 1950s and early Armcan counr 18 1960s efforts were made to calculate the savings from a birth Primary education 160' 84 Secondary education 147b 57 averted. These most frequently took the form of estimating the Health 129c 60 consumption by children and adults and the earnings of an Total 436 201 average adult and discounting the hypothesized streams of consumption and production. These efforts are reviewed by Banglapal-high- mortalitv Ohlin (1967). Since any period of production is preceded by a non-African country period of consumption, these estimates were highly sensitive Primary education 160d 84 to the discount rate. Enke (1960) used rates of 10, 15, and 20 Secondary education 193 74 percent for a country like India and found the value of a birth Health 129' 60 averted was 3.8, 2.6, and 2.1 times the per capita annual Total 482 218 income, respectively. Ohlin estimates that the value of a birth Co"enco-low-morrulirv averted would be zero at a 4 percent discount rate, but twice developingcountrn per capita income at 6 percent. Alternative methodology was Primary education 508e 354 employed by Demeny (1965), who projected income using a Secondary education 492f 273 macroeconomic model with different levels of fertility. He Health 564g 281 estimates that "gains from preventing a birth is of the order of Total 1,564 908 magnitude of two per capita income" (Demeny 1965 cited in ---- ---- Ohlin 1967, p. 116). a. Assumes universal six years of school. Capital costs at lowest 30 percen- More recent work has been less heroic and has focused only tile for World Bank primary school projects. Buildings last thirty years. Re- 2kNortman current cost 13 percent ofaverage poor country per capita income. on public expenditure saved by preventing a birth.2 Nortman b. Assumes 16 percent attend six years of secondary school. Capital costs and Lewis (1986) focused on the savings to the Mexican social at lowest 30 percentile of World Bank secondary school projects. Buildings security system from each peso spent on family planning. They last thirty years. Recurrent costs per year are twice those in primary educa- tion. documented that the cost per pregnancy per mother was c. Assumes $6 per capita (as in China and Sri Lanka). Life expectancy is 36,000 pesos and the cost of care for a child in the first year of fifty-one years. life was 34,000 pesos.'o In calculating a cost-benefit ratio they d. Assumptions as in note b., but 21 percent of children have six years of l w secondary education. also included the benefits of preventing incomplete abortions, e. Universal six years ofprimary school. Capital costs at median for which then had to be treated. The cost-benefit ratio was World Bank projects. Recurrent costs $100 p.a. calculated to be nine pesos saved for pesos spent on family f. Assumes forty-five percent secondary enrollment. Capital costs at me- dian for World Bank projects. School life thirty years. Recurrent costs tWice planning. For Indonesia, Chao and others (1985) analyzed the those in primary education. savings from expenditure averted for education and health by g. Assumes $28 per year for public expenditures per capita. Life expec- preventing a birth. A study by Kiranandana and others (1984) tancy is sixty-four years. Source: World Bank estimates. for Thailand estimated the savings in lower expenditure on 344 Susan Cochrane and Frederick Sal education, health care, housing and infrastructure, and social Widowhood is not a bar to remarriage, and polygamy ensures services. As long as programs are completely voluntary the that women do not remain without partners for too long. Thus benefits of a wanted child to the parents need not be included in most parts of Africa most fertile women are exposed to the in the calculations for obvious reasons. possibility of childbearing throughout their reproductive life, In table 16-10 we report the total savings and the savings except for periods of postpartum abstinence, which may be per birth averted in education and health for the three model quite long in some traditional societies. countries if all the unwanted births are averted.' The savings It is difficult to change behavior as basic as the initiation of per birth averted are dramatically different between the high- sexual activity and marriage, but age of marriage does rise and low-mortality countries but fairly similar for the two systematically with certain aspects of economic development. high-mortality countries. The rate of discount also makes a Developmental actions are taking place today which influence substantial difference in the savings. In a later section the these types of marriage and which can be considered among savings per birth averted will be compared with the costs per possible strategies for decreasing fertility in many communi- birth averted. ties. Among these are general education and paid employment away from home. General education, particularly of girls, helps Reducing Excess Fertility them to postpone getting married until they are in the later teens or early twenties. By postponing marriage they have less The most direct strategy for reducing excess fertility is family exposure to pregnancy and may have fewer children, if premar- planning, but delayed marriage, prolonged breastfeeding and ital fertility does not increase. Legislation may also have some abortion have significant effects as well. effect on the age of marriage, but it must be accompanied by general reform of women's rights and enforcement mecha- Elements of Preventive Strategies nisms (Duza and Baldwin 1974). Cultural taboos, such as a woman's leaving the husband's Strategies to prevent excesfetifamily after she has a baby and staying with her own family Strategies to prevent excess fertility have been very widely utltecidcnwl,o oa' o aigcnatwt discussed in the literature. It is now accepted that general until the child canwalkorawomans nothaving contact with socioeconomic development leads to a lowering of fertility and her husband until the child has grown its mslk teeth, are therefore any programs aimed at fertility control will ensure stopulations whrch ensured that children were spaced at nter the most rapid results if undertaken as a component of broader withm erzor more years. These taboos are breaking down development efforts. A strategy for preventing excess fertilityand as husband and wife stay together in woulde havelo n etortebased.on A carategyful analysi exss a reiatin nuclear families in urban situations. In such situations, there would have to be based on a careful analysis and appreciation is a clear need to replace the lost cultural taboos with technol- of the proximate determinants of fertility in any given country og t enecd spacing. or community and the relative value or relative contribution ogy tO ensure child spacmg. Breastfeeding is considered the most important natural con- of each determinant within the system at any given point in , A > . . . , ,. , . ~~traceptive. For breastfeeding to be a useful contraceptive on a time. Bongaarts (1978) has stated the following as the main community basis, it has to be prolonged and given on demand determinants of fertility in any community: with the child being at the nipple whenever he or she wants, * The patterns of marriage and consequently exposure to even at night. It is believed that the suckling at the nipple pregnancy produces a nervous stimulus which then triggers the hypothal- * Breastfeeding practices amus and pituitary axis to produce the necessary hormones * Abortions which prevent ovulation. Therefore, efforts to promote * Contraception or. direct fertility regulations activities breastfeeding must be considered an important contribution to efforts to help with the reduction of excess fertility. Recent These proximate variables directly affect fertility. Socioeco- guidelines indicate that breastfeeding is a highly reliable nomic factors, access to family planning, and economic oppor- method of contraception until the child reaches six months of tunities, as well as legislation, affect fertility through these age, supplementary feeding is introduced, or menses retums. proximate determinants. Education is the most pervasive and Under these circumstances breastfeeding is 98 percent effec- best-documented factor affecting fertility through these mul- tive. (See Consensus Statement on the Use of Breastfeeding tiple channels (Cochrane 1979; Cochrane and Zachariah as a Family Planning Method, Lancet 1988.) If any one of these 1983; United Nations 1988). events occurs, other forms of contraception should be intro- In many developing countries marriage pattems are varied duced. Breastfeeding beyond any of these points will lower the and not easy to define. In traditional societies, women are probabilities of conception, but the degree of that reduction generally expected to marry early and remain in marriage; their varies greatly from one woman to another. Thus, although period of exposure to childbearing is thus very long. In many prolonged breastfeeding, beyond six months, can suppress fer- parts of Africa, for example, young women marry as early as tility for society as a whole, it is an unreliable individual age fifteen or younger-among some groups as early as onset contraceptive. During the course of development and with the of menarche-and remain married either to the same person increase in education, the practice of breastfeeding to maxi- or another person until they stop childbearing naturally.32 mize its antinatal effects tends to decrease. To counter this, it Excess Fertilitv 345 Table 16-11. Appropriateness of Contraceptive Method, by Stage in Reproductive Life Cycle Method Before first birth (delay) After first birth (spacing) Completion of family Oral contraceptive Most appropriate Appropriate Inappropriate Injectable hormone Appropriate Most appropriate Least appropriate Implant Appropriate Appropriate Most appropriate IUD Inappropriate Most appropriate Appropriate Condom Most appropriate Least appropriate Inappropriate Vaginal spermicide Appropriate Appropriate Inappropriate Diaphragm/cap Appropriate Appropriate Inappropriate Periodic abstinence Appropriate Appropriate Inappropriate Sterilization Inappropriate Inappropriate Most appropriate Note: Menstrual regulation and abortion are backups throughout reproductive life cycle. Source: Hutchings and Sanders 1985. is necessary to make family planning more available and to selves need to create demand as well as provide services. provide for the promotion of breastfeeding where possible Demand creation can be divided into two types: indirect and (Huffman and Combest 1988; Green 1989; and Labbok and direct. Indirect demand creation depends generally on social McDonald 1990 in the supplement to volume 31 [19901 of the and economic development, issues within the domain of de- InternationalJournal of Gynecology and Obstenics)." velopment planning generally and national social mobiliza- The attitudes and practices of health personnel, particularly tion. Among specific efforts in this domain are activities that those attending delivery, can have an important effect on the improve general education of the people and especially those establishment and continuation of breastfeeding and on appro- which emphasize women's education, women's mobilization, priate weaning behavior. These practices are important as and the improvement in the status of women through various health interventions which have an externality of preventing activities, be these activities developed as direct projects for pregnancy. It is impossible, however, to determine the cost- family planning programs or activities developed from other effectiveness of such interventions in reducing excess fertility. programs but relating to the improvement in the status of Family planning programs depend on the availability and women or improvement of their economic opportunities. Such the acceptance of modem contraception. The programs them- programs can indirectly lead women to seek and accept more readily assistance to control their fertility. Their effect can be Table 16 12. First Year Failure Rates of Birth strengthened when accompanied by special messages or infor- Cotrole Method. FrtYaFaurRaeofBthmation and communications activities aimed at highlighting Control Methods awareness of the women about the importance of fertility Lowest Failure rate regulation for their lives. In education generally, opportunities observed In typical need to be taken to emphasize the relation between population in od bl need users Method failure rate' users - and resources and population and the environment. Popula- Chance (no method of birth control) 70 70 tion dynamics need to be a part of every school program in Tubal ligation 0.04 0.04 developing countries at the present time. Girls and young Vasectomy 0.15 0.15 women must also learn at an early age the consequences ofhigh Injectable progestin 0.25 0.25 fertility for their own health and welfare and that of their Combined birth control pills 0.5 2 children. Progestin-only pill 1 2.5 Direct demand creation is that incorporated in information, IUD 1.5 4 education, and communication (IEC) programs which provide Diaphragm (with spermicide) 2 10 information specific to family size preferences or family plan- Cervical cap 2 13 ning and contraception techniques and services. This can take Foam, creams, jellies and vaginal many forms, from mass media campaigns to a series of smaller, suppositories 3-5 15 more directed efforts. All available methods of communication Coitus interruptus 16 23 may be used from a variety of ministries and agencies, minis- Fertility awareness techniques (basal tries of health, youth, women's affairs, defense, industry, and body temperature, mucus method, agriculture, to name a few. All these need to go hand in hand calendar, 'rhythm," and douche) 2-20 20-30 with programs for the distribution of contraceptive supplies. a. Number pregnant by end of year among 100 women who start out the The delivery of services can take diverse forms. In many year using a given method and who use it correctly and consistently. environments they have been most effective as part of mater- b. Number pregnant by end of year among 100 typical users who start out nal and child health programs In other environments vertical the year using a given method. Source: CDC 1983. or single function programs have proven most successful. The 346 Susan Cochrane and Fredenick Sai Figure 16-2. Estimnated Annual Deaths Resulting from Pregnancy, Abortion, and Contraceptive Use, by Age of Woman Industrial countries Low-income developing countries Annual deaths per 100,000 ferlile married women Annual deaths per 100,000 fertile married women 150 - 150 _ 153 (186) 145 - 145 - 140 -140- 135 - 135 - 130 130 - 125 125 120 (104) 120 (69) 40 -40- 35 - 35 N, 30 30 - 25 -25- 20 2 20 - 15 -15 1 0 1 0. ...... ........ ....... 5 --- w5 . --- . -- ---- 15-19 20-24 25-29 30-34 35-39 40-44 15-19 20-24 25-29 30-34 35-39 40-44 Age Age - Absence of birth control-pregnancy-related deaths ---------- Legal abortionb Oral contraceptives-normal Oral contraceptives-women with high cardiovascular risk b,c Intrauterine devicesb,C womenb,C -- - Condoms and other barrier methods c - - - - Female sterilizationd Note: Maternal death estimates assume lower use-effectiveness rates in developing countries for all methods except the IUD a. Countres with per capita incomes less than $410 (U.S. dollars) and with average maternal mortality rates of about 350 deaths per 100,000 live births. b. Method-related deaths. c. Pregnancy-related deaths. d. Procedure-related deaths, a one-time risk. Data that reflect deaths per 100,000 procedures rather than per100,000 women overstate relative risks over time compared with other methods. Source: Population Crsis Committee 1985, adapted from Potts, Speidel, and Kassel 1977. Excess Fertility 347 Table 16-13. Relative Cost of Birth Control Methods Subsequent Relative cost per cYP product cost, Relative cost amortized over average Costs in initial per year per CYP in initial lifetime of methoda Method year (U.S. dollUars) (U.S. dollars) year' (percent) (percent) Oralcontraceptive 2.17 2.17 100 100 Injectable hormone 3.51 3.51 162 162 Implant 16.23 n.a. 748 212 IUD 3.45 n.a. 159 42 Condom 3.88 3.88 179 179 Vaginal spermicide 5.76 5.76b 265 265 Diaphragm/cap 4.75 0 219 77 Femalesterilization 8.91 n.a. 411 59 Male sterilization 6.68 n.a. 308 44 Menstrual regulation/abortion 4.45 n.a. 205 n.a. n.a. Not applicable. a. Percentages are based on estimated couple of years protection (cyp) initial year costs and are relative to cost of oral contraceptives. b. In actual practice, spermicide use would add to the yearly cost. c. Couple years of contraceptive protection. Source: Hutchings and Sanders 1985. one thing that has been learned is that there is no uniquely except the most recently developed implants.34 In figure 16-2 defined best delivery mode. Important though such clinic- we summarize the mortality risks from various methods. It is based programs are, their outreach may be rather limited, important to note how the relative risks between contracep- especially in communities where the health services them- tion and childbearingdiffer between developing and industrial selves are fairly restricted in their outreach. In such a situation countries. It is less easy to summarize the morbidity and other there is a need to build outreach programs such as social side effects of various methods. Most methods have some marketing or community-based programs which ensure that actual or perceived negative side effects. These effects and the the users themselves are closely involved with the family importance attached to them differ greatly from one woman planning programs. Irrespective of the form of delivery system, or couple to the next. It is also true that some methods have there is a need to have a medical service with trained staff in positive side effects, from reducing anemia (oral contracep- what may be termed clinical contraception, such as the inser- tives) to prevention of sexually transmitted diseases to the tion of intrauterine devices (lUDs) or terminal methods such as reduction of the risk of some cancers (Lee, Peterson, and Chu vasectomy and tubectomy. Such a service provides for referral, 1990). Thus, programs that offer a wide mix of methods are backup, and training of the staff of other programs. able to attract a larger number of women than programs with Primary prevention of excess fertility is thus based on a a narrower range of methods. It is also clear that side effects of strategy of general social development with equity, in particu- contraceptives, particularly oral contraceptives, differ accord- lar when such development targets women. Direct population ing to a woman's age. This is another reason to have a program education in communicating also helps, as does specific IEC for that has a wide method mix. Family planning methods vary men and women of childbearing age. Family planning services substantially in commodity costs and personnel and infrastruc- providing information and contraceptive services at affordable ture costs. The costs of various methods are too specific to be prices help to ensure that the population is able to control its reviewed in detail here (see table 16-13 for the supply costs of fertility. There are, however, other factors that inhibit that various methods and those costs in relation to the cost of oral ability that need attention, from the empowerment of men and contraceptives). Of greater relevance are the program costs of women to information on the real side effects of high fertility serving a family planning user or preventing a birth. Reviews and contraceptive risks. The role of abortion in preventing of program costs per user or per birth averted were compiled in excess fertility is more controversial and will be discussed later. preparation for the 1984 World Population Conference (Bulatao 1985 for a review and analysis of these costs). The Costs and Efficacy of Family Planning costs include or should include the entire costs of promoting family planning use and of commodities, personnel, equip- A wide variety of family planning methods is now available. ment, and facilities for the program. These cost estimates are They differ in their effectiveness and side effects as well as their in fact often simply derived by dividing expenditure in any costs (Holck and Bathija 1988; cited in WHO 1988). In table given year by the number of users. Detailed cost analyses to 16-11 we summarize appropriateness of various contraceptives give precise estimates are fairly rare. (Analyses by Barnum to different phases of the reproductive life of a woman. In table [1983] and Chemichovsky and others [1989] are exceptions. 16-12 we summarize the failure rates for most modern methods See Serageldin and others 1983 for a review of the issues in 348 Susan Cochrane and Frederick Sai cost-benefit and cost-effectiveness analysis and Cochrane, uncertainty about the most effective and efficient delivery Hammer, Janowitz, and Kenney 1990 for a review of newer cost systems for family planning in different environments. We will estimates.) briefly review changes that are currently on the horizon. For the purpose of this analysis the costs per user in the study Many contraceptive technologies are currently being devel- by Bulatao (1985) and the costs per birth averted in the study oped that may be useful eventually. The most promising for by Cochrane and Zachariah (1983) will be used. Both of these the 1990s is the implant. This provides highly effective con- are dependent on earlier work by Speidel (1983). The costs as traception for up to five years with few side effects except collected refer to 1980, but they have been inflated to 1987 in irregular menstrual periods. It is, however, relatively expensive table 16-14.35 If economies of scale exist in the program, the and requires skill in its implantation and even more skill in its average costs have probably dropped since 1980.36 The costs removal.40 Thus it is intensive in training and personnel at the per birth averted are related to the fertility preference of the beginning and end of use. Some research is under way on women in the society. Regression analysis shows that the cost developing a biodegradable sheath which would eliminate the per birth averted drops by $66 for a one-child difference necessity of removal. A once-a-month injectable contracep- between the actual and desired TiR (table 16-4). As shown in tive, Cyclofem, is nearing commercial production and would figure 16-3, the cost per birth averted also decreases with an eliminate the problemof irregular menstrual periods that occur increase in the proportion who want no more children ($4.6 with longer-term, hormonal methods. Other contraceptive per 1 percent increase in the number wanting no more chil- methods being developed have less chance for immediate dren).37 Therefore, for the three hypothetical countries we can breakthrough because they are in earlier stages. The male pill predict the costs of a birth averted by first determining the is being developed in China under a cooperative program with average life expectancy (fifty-one years for model countries the Rockefeller Foundation. Vaginal rings are also being de- Libana and Banglapal and sixty-four years for Colexico) and veloped, and the new RU486 provides some promise for future predicting the percentage wanting no more children (figure use. The Population Council, Family Health International, 16-1) and then projecting costs per birth averted from the the Ford Foundation, the National Institutes of Health in the equation in figure 16-3. One adjustment that must be made is United States, the Rockefeller Foundation, and the Human that on average the countries with lower mortality have higher Reproduction Programme (HRP) in WHO are the main actors in proportions of women who want no more children than do the coordinating research on contraceptive technology. (See NRC primarily African countries. Best estimates indicated approxi- 1990 for a review of the evaluation of contraceptive research mately 20, 30, and 45 percent of the women want no more and development.) children in the three models, respectively. The equation in Despite the efforts of these organizations, the low level of figure 16-3 implies the costs per birth averted of $259, $213, funding for contraceptive development for low-income coun- and $144, respectively, in the three countries." Comparing tries is a serious problem because private enterprise, which is a these costs with the savings of births averted in table 16-10, significant source of technological change, is not interested in one finds that at the 5 percent discount rate, the costs of a birth developing contraceptives for that market. Three factors ex- averted is justified on the basis of economic savings to the plain this lack of interest: (a) because many techniques that government alone in all three models. At the 10 percent rate need to be developed for poor countries must, of necessity, be of discount, the costs of a birth averted would be fully justified very low cost and low dosage, they will offer low profits; (b) in Colexico, marginally justified in Banglapal, and unjustified because most contraceptive research is done in industrial in Libana39. countries and must be approved for use there to be profitable, There are several caveats to be made with respect to the the research is skewed toward contraceptives that reflect the findings above: (a) the costs and savings are based on hypo- relative risks of using contraceptives and the maternal risks in thetical country types, and actual estimates would have to be those countries (see figure 16-2 ); and (c) product liability laws made in real circumstances; (b) the conclusions above apply are so stringent in many countries like the United States that only to the economic benefits of family planning, and the new products face too many potential liability costs even if a health benefits, which are substantial, as shown above, would product is developed and is sold. An example of how these risks be additional; and (c) the extent to which access to contracep- distort research is the recent development of the contraceptive tion would in fact alter family size preferences is yet to be sponge in the industrial world. This method is too expensive, established. To the extent that such an effect exists, the cost too inconvenient, and has too high a failure rate for poor functions are incorrectly specified. countries, but it has very few possibilities of side effects for which a company could be sued (Population Crisis Committee Technology and Future Changes 1985). An important question to be addressed in the area of con- All existing contraceptives have some negative side effects or traceptive technology in the next decade is the interaction of inconvenience; few methods allow male control or responsi- various methods and acquired immunodeficiency syndrome bility for the method, and some methods are unacceptable to (AIDS). The condom serves both to prevent births and reduce certain religious groups. Thus technological development is the probability of contracting AIDS. For other methods the needed for contraceptives. In addition there is continued linkage between the spread of the disease and the use of the Excess Fertiliy 349 Table 16 14. Costs of Averting a Birth through the development of a small, cheap vial-needle combination Family Planning that is impossible to reuse for delivery of Cyclofem. Research (1987 U.S. dollars) is also needed on how various contraceptives affect the risk of Cost per averted birth contracting the human immunodeficiency virus or developing CountrY Cost per user Low High the disease by either changing sexual behavior, increasing the - receptivity to the virus, or stimulating the development of the Bangladesh 29 102 109 disease. Through HRP the World Health Organization is spon- Colombia 7 21 29 soring some research in this area and has issued guidelines. Costa Rica 22 71 160 These risks must be set against the risk of the more rapid onset Dominican Republic 17 50 69 of the disease that is stimulated by pregnancy and the Indonesia 12 49 64 Jordan 31 88 108 transmission of the disease to the infant during pregnancy or Kenya 100 350 386 delivery. Korea, Rep. of 12 53 77 In the area of the management of the delivery system, Malaysia 21 69 92 attention is now being focused on two important aspects: Mexico 22 59 78 improvement of service quality (Jain 1989; Bruce 1990), and Nepal 80 330 364 developing delivery models in Sub-Saharan Africa, where Pakistan 1 5 77 8 1 Panama 36 136 231 demand for family planning is weaker and more oriented Peru 10 34 38 toward spacing births than stopping them and the health Philippines 20 63 77 infrastructure is very weak. Other themes that are receiving Sri Lanka 8 31 41 considerable attention are the necessity of having an appropri- Source: Converted from 1980 figures in table 9 of Cochrane and ate monitoring and evaluation system, the movement away Zachariah 1983, using standard indices of the Priorities Project. from one-method programs (India), the appropriate role of the private sector (Indonesia), and the appropriate mix of family planning delivery and promotion (issues involving the minis- methods is unclear and needs to be researched. All surgical tries of health, ministries of information and education, and methods require much close attention to cleaning of instru- other important ministries and coordinating bodies in popula- ments and equipment when the AIDS virus is at all prevalent in tion and family planning). The commitment of the medical society. This covers implants, sterilization, abortion, and IUD profession as well as national leaders to the delivery of family insertions. In addition, injections must be more carefully mon- planning for health, equity, and economic considerations is itored to be sure that needles are not reused or are properly also of crucial importance to the success of a delivery system. sterilized. The Human Reproduction Programme is sponsoring There exist serious shortcomings of actual practice over the best practice in family planning delivery, but it is also true that despite excellent synthesis work done by the National Acad- Figure 16-3. Relationship between Costs per Birth emy of Science in its 1987 review by Lapham and Simmons Averted and Percent Wanting No More Children and the Johns Hopkins University Population Reports there is still an enormous amount that needs to be done to clarify what Cost per birth averted the best practice is in different environments. 400 Kenya * Reasorable Delivery Systems for Model Countries * Nepal 350 \ In the two hypothetical high-mortality countries the lack of 300 health infrastructure is a fundamental underlying constraint. 300 \ Coa2t = 351 - 4.6 (percent wanting no more) In addition in Libana demand for contraception is less than in 250 - . Banglapal. In both models it is necessary to develop the health \ Panama delivery system. It is also essential to target efforts in family 200 planning, but in the second model the targets can be broader. * Costa Rica Providing condoms, pills, and other barrier methods through 150 health centers and pharmacies and perhaps social marketing 100 Bangladesh* Jordan * \MalYs exico may be desirable and effective in urban areas. Sterilizations and 1dPaeistan * * Korea IUD insertions should be done in district hospitals everywhere, 50 - IdniDominic nIPeruia Sri Lanka but in Banglapal they may also be done in clinics if they have Republic Peru ooi adequate staff and if staff members are available to handle side Coiombia 0 effects. Sterilization camps, particularly for vasectomies, may 0 10 20 30 40 50 60 70 80 90 100 also be cost-effective, but politically sensitive. Injections can Percent wanting no more children be done on a mobile team if a reliable system of delivery can be maintained to make regular trips. This latter function may Source: Authors' calculations. be better carried out by a nongovernmental organization if the 350 Susan Cochrane and Frederick Sai government logistics are weak. Such programs need to be up to that time. It estimated that, depending on the country, backed up by information, education and communications between 4 and 70 percent of maternal deaths in developing programs that can advise on location of service and appropriate country hospitals were the result of complications of illegal use and contraindications. This is particularly important abortion. A recent study on illegal abortion by Figa- where contraception use is low, because fear of side effects can Talamanca and others, assessed the medical costs of illegally seriously undermine the development of a program if there induced abortion in urban hospitals in four developing coun- exists no referral agency to handle them. tries. For Malaysia, 52 percent of the abortion cases admitted In Colexico all the above can be used, but delivery of IUDS, were estimated to be induced, whereas only 12 percent of those sterilization, and injection can be done more routinely at lower in Nigeria were. In Turkey, 41 percent were estimated to be levels of service because of higher demand and better health induced. There are no estimates immediately available of the infrastructure. It is probably also essential to have either ex- economic costs of those abortions, but the induced abortion plicit policies on legal abortion or IEC on the dangers of abor- cases experienced high costs in hospital days, units of blood tion and medical facilities for treating incomplete abortions. administered, and cost of medication (table 16-15). The cost This is likely to be more serious in Colexico because the much in disability and ill health from these illegal abortions has not stronger demand to restrict fertility will lead to abortion in been estimated. more cases of contraceptive failure. Priorities Case Management: Unwanted Pregnancies Priorities for resource allocation depend on the level of de- For pregnant women who want no more children or who wish mand for family planning and the level of mortality. to postpone the timing of a birth, the choice is between carrying the pregnancy to term or seeking an abortion. The Priorities for Resource Allocation type of costs of an unwanted pregnancy include: (a) the cost of abortion, (b) the cost of treating incomplete induced abor- A considerable amount of evidence has been compiled in tion, and (c) additional costs associated with unwanted preg- various sources on the unmet demand for family planning to nancies that are not associated with normal pregnancies. limit births and to a lesser degree to space births (see table 16-5 These latter costs are not well documented for industrial or for data; Westoff and Moreno 1989 for a deeper analysis for developing countries. five Latin American countries). As discussed above, we esti- There are enormous differences from country to country in mate that about a third of the births in the developing world laws covering abortion. Where it is legal, the method is rela- tively safe in the first trimester and the costs are mainly those associated with the abortion itself. Estimates from the UN Table 16-15. Cost-Related Indices of Induced and indicate that about 40 percent of the women in developing Spontaneous Abortions in Four Participating Centers countries have access to legal induced abortion. This ranges Induced Spontaneous from 10 percent in Africa to 50 percent in Asia (40 percent in Parameters, bw center abortion aboraon Latin America; United Nations 1988, table 38). The laws on abortion differ dramatically, however. In thirteen of ninety-six Mean length of hospitalization (days) developing countries reviewed, it was illegal under all circum- Malaysia 4.8 4.5 Nigeria 10.5 7.5 stances, and in only seven was it available on request. In the Turkey 1.7 1.0 vast majority it is available for health reasons, but in most, only Venezuela to save the life of the mother (forty-two of ninety-six). Thirty- Caracas 4.2 2.4 five countries permit abortion for health reasons. Such laws Valencia 5.3 2.5 permit wide latitude to doctors in preforming abortions and Mean units of blood administered give access to safe abortion to most women who can pay Malaysia 0.2 0.1 doctors fees. The price of private legal abortion ranges from Nigeria 0.6 0.2 $16 in Bangladesh to $966 in Iraq. The normal range is Turkey - - between $100 and $200 in countries for which data are avail- Venezuela able (Ross and others 1988). Publicly provided abortions are Caracas 0.5 0.2 free in a number of countries, but fees of under $100 are Valencia 1.0 0.5 charged in several countries. Relative cost of medication For women with no access to safe abortions, the cost of Malaysia 1.9 1.0 abortion complications must be added to the cost of the Nigeria - - Turkey 1.5 1.0 abortions themselves. It is impossible to get an estimate of the Venezuela 8.8 1.0 number of illegal abortions performed throughout the world or their cost. Evidence of abortion complications from hospital - Data not available. a. Data refer to cases classified as shown in Table 16-6. admissions is the best index. The "Population Reports" of July b. Computed by considering the spontaneous abortion cost equal to unity. 1980 by Liskin compiled data on complications of abortions Source: Figa-Talamanca and others 1986. Excess Fertiliry 351 outside China are unwanted, and excess fertility would be even Priorities for Research higher if births that come too soon were to be included. This implies substantial unmet need for family planning. Research in contraceptive technology is needed primarily in It is difficult to estimate what would be necessary to meet the following areas: (a) reversible sterilization, (b) male con- all the need for family planning as expressed by individuals. traceptive methods, and (c) understanding of the interaction The analysis above indicates that in areas where mortality is between various contraceptives and AIDS. Reversible steriliza- low or demand to limit family size is 30 percent or more the tion has high priority, not only to expand the range of choice savings from public health and education expenses alone are of individuals, but also to meet the requirements of Islamic sufficient to justify public expenditure on family planning to teachings on what is acceptable.4" Male contraceptives are avert a birth.4' Where thedemand to limit the numberofbirths needed for a number of reasons. Evidence is accumulating that is lower, as in much of Sub-Saharan Africa, the cost of averting in many parts of the world husbands do not want significantly a birth is much higher. Because of this the economic justifica- more children than their wives, and in some cases they want tion for supporting family planning is less compelling, but fewer (Mason and Taj 1987). Therefore it is important to given much higher maternal, infant, and child mortality in provide them with more methods from which to choose. The these areas the health justifications are more important (see interaction of contraception and AIDS is, as explained earlier, table 16-9 for the cost of averting an infant or child death an important question in gaining access to the effectiveness through family planning for spacing). In addition, reducing and safety in contraceptives. mortality is an important factor in stimulating more demand for Although the lack of a perfect contraceptive for all users family planning.42 Even in areas of low demand, however, at restricts use to some degree, there are important research rates of discount of 5 percent it is justifiable that the public questions that still need to be addressed in service delivery, not fully support the prevention of all births that are unwanted by to mention motivation. As indicated earlier, a large number of the family. women in the developing world who are motivated to limit The level of expenditure needed to eliminate all fertility their fertility or space their births are not using contraception, that is unwanted by the family is difficult to calculate. The and many of them say that they do not intend to do so in the estimate in table 16-6 that 11.6 million births were unwanted future. One important area of research is the ambivalence in the countries for which we have data is, of course, an toward family planning. There has not been a large compen- underestimate because those countries included only 39 mil- dium of information from surveys on why women say they are lion of the approximately 113 million births that occurred in not using or do not intend to use contraceptives, particularly the developing world in the average year in the late 1980s. The among the women who want no more children or want to estimates in table 16-14 show an average estimate of about postpone their next birth.44 Data that have been compiled $125 per birth averted in 1987 prices. Preventing the 11.6 indicate that there are wide differences in reasons for nonuse million births would cost $1.5 billion. If one can generalize of contraception among those who do not want another child. from the WFS data, 30 percent of all births in the developing In Nepal and Mexico the main reason was lack of knowledge world are unwanted and preventing them would cost $4.2 of a source.45 Fear of side effects was a prime reason in Asia and billion per year. The United States Agency for International Latin America. In three countries of Latin America in the late Development (usAIn) estimates that $1.5 billion was spent on 1970s, the cost of contraception was also a significant deter- family planning in 1980 and that $3 billion would be needed rent. Lack of access is not mentioned frequently in African in 1990 (Gillespie and others 1988). The USAIDestimates imply surveys, although it is probably important. In addition, few of that if the WFS figures are representative of the world as a whole, the surveys find the opposition of husbands a great problem. substantial increases in funding will still not be sufficient to In the recent Demographic and Health Surveys, lack of knowl- cover current users plus all those women who want no more edge of contraception was the main reason given in Ghana, children and are not using contraception. Our estimate is that conflict with religion and custom was given as an important this figure would be $4.6 billion in 1987 prices. This could not reason for nonuse of contraception in Senegal, and health covertheunmetneedforcontraceptionforspacing. Regardless concerns about contraceptives were most important in Nigeria. of exact current resource needs in 1990, by the year 2010 the Clearly more research is needed on why people who want to number of currently married women of reproductive age will limit fertility do not use contraception. This work should focus have doubled according to USAID estimates. Therefore the on trying to design strategies for family planning delivery which expenditure on family planning in the developing world is are specific to the concerns of the country. Increasing attention substantially below what is needed to eliminate unwanted is being given to the quality of service as a dimension of access fertility by the estimates of individuals, and those resource that affects not only use but efficency of use and continuation. requirements are increasing rapidly. The geographic distribu- A final area that needs attention is the determinants of tion of expenditure is more controversial because of the large family size preferences themselves. As indicated earlier, partic- difference in geographic distributions of excess fertility by ularly in Sub-Saharan Africa, there is a discrepancy between societal and individual estimates of excess fertility. Therefore what individuals may consider excess fertility and what might we know that a substantial increase in expenditure in family be excessive from the point of view of economic growth and planning is needed. There is, however, a large number of development. It may well be that one reason for the large factors that are not known. family size preferences is the lack of access to reliable family 352 Susan Cochrane and Fredenck Sai planning services. The effect of service access on family size ment of family planning services that are best suited to societies preferences is not well studied in the literature. There is good such as those in high-mortality countries of Africa, where there theoretical reason to believe not only that the access to family is low motivation to limit fertility but high motivation to space planning affects the use of contraception among those who births. The role ofbreastfeeding in the fertility decisionmaking want to limit their fertility, but also that it directly affects also needs to be more completely researched, especially with whether they wish to limit it (Cochrane and Cochrane 1971 respect to spacing. and 1974). The prior question of whether they perceive fertil- ity as a choice is probably related to access as well, but this has not been well researched. Appendix 16A. Tables Finally, it is essential to gain a better understanding of contraception for spacing: its determinants and its demo- The tables in this appendix show the relationships between graphic consequences. This knowledge is basic to the develop- childbearing patterns and the survival of offspring, by country. Table 16A-1. Total Fertility Rate, Crude Birth Rate, and Rate of Natural Increase, by Country 1960-65 1970-75 1980-85 1985-90 Country TFR CBR RNI TFR CBR RNI TFR CBR RNI TFR CBR RNI Sub-Saharan Africa Benin 6.8 4.8 16 6.8 50 23 7.0 51 30 6.5' 49 33' Cameroon 5.7 43 20 5.7 42 23 5.8 43 27 7.0a 48 35' Coe d'lvoire 6.6 43 18 6.7 45 24 6.7 46 30 7.0a 48 34a Ghana 6.5 48 28 6.5 47 30 6.5 47 32 6.3' 45 32' Kenya 8.1 57 35 8.2 57 3.9 8.1 55 41 7.7' 52 41 Lesotho 5.8 43 20 5.7 43 33 5.8 42 25 5.8, 41 28 Liberia 6.3 46 23 6.4 46 26 6.9 49 31 6.6a 46 33a Mauritania 6.9 50 23 6.9 50 26 6.9 50 29 6.5' 48 30Q Nigeria 6.9 42 28 7.1 51 31 7.1 50 33 6.9' 50 34a Senegal 6.7 47 21 6.7 47 25 6.5 46 37 6.5 45 282 Sudan 6.7 47 22 6.7 47 25 6.7 46 29 6.6 45 29a Latin America and the Caribbeani Brazil 6.1 42 30 4.7 34 24 3.8 29 22 3.4 28 212 Colombia 6.7 45 32 4.8 33 24 3.9 29 23 3.2 27 20 Costa Rica 6.9 45 36 4.3 31 26 3.5 31 24 3.3 28 24 Dominican Republic 7.3 48 32 6.3 42 31 4.2 33 25 3.8 31 25 Ecuador 6.9 46 31 6.0 41 30 5.0 35 31 4.3 33 26 Guyana 6.0 40 32 4.5 33 25 3.3 29 23 3.1 27 20 Haiti 6.1 44 23 6.1 43 23 5.7 41 27 4.7 35 22 Jamaica 5.5 40 31 5.4 33 25 3.4 28 23 2.9 26 20 Mexico 6.7 45 34 6.4 43 34 4.6 34 27 3.6 28 23 Panama 5.9 41 31 4.9 36 28 3.5 28 23 .3.1 26 21l Paraguay 6.6 42 30 5.7 38 30 4.9 36 29 4.6 35 29' Peru 6.9 46 29 6.0 41 28 5.0 37 26 4.1 32 23a TrinidadandTobago 5.0 38 31 3.5 27 20 2.9 25 18 2.8 26 19 Venezuela 6.5 44 35 5.0 36 29 4.1 33 27 3.8 31 26' Asia Bangladesh 6.7 47 25 7.0 49 28 6.1 45 27 5.5 40 25' China 5.9 38 21 4.7 31 22 2.4 19 12 2.4 21 14 India 5.8 42 23 5.4 38 22 4.3 32 19 4.3' 22 20Q Indonesia 5.4 43 21 5.5 41 24 4.1 32 19 3.6 29 18 Malaysia 6.1 43 30 5.1 35 24 3.9 31 24 3.5 28 22 Nepal 5.9 46 21 6.5 47 25 6.3 42 23 5.9 41 24a Philippines 6.6 44 31 5.3 37 26 4.4 33 25 3.9 30 23a Korea, Rcpublic of 5.4 40 27 4.1 29 20 2.6 23 17 2.1 20 14 SriLanka 5.1 35 26 4.0 29 21 3.4 28 21 2.8 24 18 Thailand 6.4 44 30 5.0 35 26 3.5 28 20 2.8 25 17 Middle East arnd North Africa 3 Egypt 7.1 45 25 5.5 38 22 4.8 37 25 4 S 34 24 Jordan 7.2 48 29 7.4 47 35 7.4 45 37 5.8a 38 32a Morocco 7.2 50 30 6.9 46 30 5.1 36 25 4.3' 32 23a Pakistan 7.2 48 26 6.5 44 26 5.8 43 28 6.7a 47 33a Syrian Arab Rep. 7.5 47 31 7.5 45 33 7.2 47 38 6.8' 45 38a Tunisia 7.2 47 29 6.1 37 24 4.8 33 23 4.3a 31 23a Turkey 6.0 41 26 5.5 37 25 4.0 30 21 3.7. 29 21' Yemen, Republic of 7.0 49 21 7.0 49 23 7.0 49 30 6.8 49 293 Table 16A,2. Estimate for Main Effects Parameters in Model of Neonatal Mortality Births in past 2 years Births in the past 2-4 years Mother's education Birth order Mother's age at birth One One Two One One Two Fourth Seventh Less 35 or Female Country Base alive dead or more alive dead or more Medium High to sixth or more First than 20 20-34 more child Africa Senegal -2.89 1.42 3.49 1.25 1.01 1.31 1.65 0.75 0.57 0.88 0.80 1.38 1.12 0.79 1.38 0.79 Benin -2.89 1.02 2.72 2.34 0.79 1.55 1.07 0.81 0.42 1.21 1.75 1.07 1.23 1.13 0.76 0.81 Egypt -3.71 2.46 4.10 4.22 1.21 1.86 1.88 0.74 0.53 1.28 1.28 3.06 1.19 1.30 1.57 0.75 Cote d'lvoire -2.90 1.07 2.36 1.39 1.06 2.01 1.60 0.69 0.29 0.91 1.07 2.27 1.26 1.06 1.20 0.69 Cameroon -3.46 1.36 3.10 2.39 1.06 2.08 1.68 0.87 0.51 1.09 1.08 1.88 1-25 1.14 1.05 0.96 Mauritania -3.60 1.75 3.67 5.75 1.03 2.53 1.62 1.63 1.14 0.79 1.21 2.05 1.06 0.77 1.21 0.61 Lesotho -2.90 1.70 3.60 6.82 0.93 1.73 1.25 0.90 1.11 0.81 1.32 1.27 0.91 1.02 1.20 0.90 Kenya -3.70 1.57 3.16 1.90 1.00 1.80 1.03 0.84 0.67 1.15 1.52 2.61 1.21 1.13 0.95 0.79 Morocco -3.80 2.77 3.53 2.92 1.08 1.28 1.52 0.57 0.78 1.01 1.39 3.10 1.09 0.91 1.27 0.84 Sudan -3.61 1.63 4.53 2.48 1.13 2.83 1.68 0.64 0.87 1.16 0.89 2.29 1.09 0.76 1.55 0.66 Ghana -3.77 1.49 3.10 3.90 1.13 3.06 1.19 0.82 0.83 1.19 1.88 2.44 1.05 0.97 0.79 0.86 Tunisia -5.50 2.03 8.08 11.36 1.65 3.97 3.46 0.96 0.01 0.64 1.07 2.25 1.65 0.97 1.26 1.06 Average -3.56 1.69 3.79 3.89 1.09 2.17 1.63 0.85 0.64 1.01 1.27 2.14 1.18 1.00 1.18 0.81 Asia and the Pacific Yemen, Republic of -3.22 1.25 2.72 2.77 0.91 1.02 0.96 0.63 0.83 1.02 1.62 1.49 1.73 1.39 0.61 0.72 Nepal -2.88 1.58 2.32 1.42 1.02 2.10 1.67 0.80 0.63 0.95 1.15 1.60 1.49 1.17 1.02 0.91 Bangladesh -3.01 1.99 3.94 3.71 1.14 1.54 1.65 0.89 0.76 0.71 0.90 2.59 1.09 0.85 0.57 0.94 Pakistan -2.87 1.32 3.06 2.46 1.02 1.62 1.51 0.80 0.66 0.93 0.93 1.84 1.42 1.07 0.97 0.76 Indonesia -3.50 1.99 2.36 3.19 1.13 1.90 1.77 0.76 0.73 1.05 1.22 1.49 1.72 1.17 1.02 0.82 Thailand -3.92 2.20 3.97 5.87 1.48 2.61 2.59 0.88 0.39 0.98 1.03 2.80 1.43 1.25 1.22 0.87 Philippines -3.91 1.62 2.32 2.53 0.68 1.88 1.07 0.90 0.68 1.57 1.65 1.54 1.25 1,05 1.25 0.72 Syrian Arab Republic -4.81 2.48 7.77 9.39 0.90 2.29 1.08 0.86 0.64 0.90 0.79 3.63 0.89 0.79 1.08 0.84 Jordan -3.76 1.92 4.26 4.95 0.87 2.05 1.35 0.89 0.73 0.70 0.88 1.90 1.06 0.72 1.00 0.99 SriLanka -3.21 1.36 3.67 3.32 0.95 2.83 1.51 0.77 0.54 0.76 0.61 1.79 1.11 0.85 1.77 0.71 Korea, Republic of -5.05 2.16 2.80 0.04 1.12 3.56 1.42 0.84 0.40 0.53 0.49 1.65 1.88 1.45 1.05 0.84 Malaysia -4.49 1.92 3.94 3.78 1.13 3.56 2.48 0.78 0.34 0.47 0.76 1.46 1.38 1.00 0.79 0.73 Average -3.72 1.82 3.59 3.62 1.03 2.25 1.59 0.82 0.61 0.88 1.00 1.98 1.37 1.06 1.03 0.82 Latin America and the Caribbean Haiti -3.00 1.43 3.74 1.03 1.02 1.67 1.21 0.90 0.41 1.58 1.12 1.40 1.43 1.17 0.87 0.90 Peru -3.46 1.97 3.29 2.83 1.19 1.88 1.67 0.55 0.42 1.13 1.05 1.67 1.36 0.84 1.06 0.70 Ecuador -3.49 1.68 2.86 2.61 0.87 1.67 1.25 0.73 0.64 1.19 1.16 1.43 1.38 1.00 1.46 0.79 Colombia -3.30 1.34 4.01 3.42 0.98 2.41 1.26 0.81 0.47 0.67 0.83 1.16 0.97 0.70 0.84 0.82 Mexico -3.56 1.31 3.13 3.10 0.88 1.84 1.08 0.73 0.62 1.35 1.82 1.42 1.70 1.27 1.07 0.76 CostaRica -3.70 1.54 3.86 5.53 0.88 1.86 1.21 0.78 0.63 0.87 1.02 1.08 1.35 1.17 1.49 0.78 Guyana -3.36 1.43 6.69 5.47 0.79 1.77 1.22 1.19 0.81 0.74 0.99 1.49 0.87 1.54 2.08 0.61 Panama -3.60 1.72 5.75 3.56 0.81 1.42 0.54 0.49 0.55 1.49 1.01 1.80 1.11 0.95 2.53 0.79 Jamaica -3.27 1.54 1.92 0.01 0.89 2.53 0.82 1.02 0.68 0.69 0.81 1.36 0.54 0.40 0.54 1.12 TrinidadandTobago -4.56 1.32 6.69 3.35 0.78 1.03 0.97 0.82 0.78 1.75 4.62 1.13 2.97 2.16 1.27 1.05 Average -3.53 1.53 4.19 3.09 0.91 1.81 1.12 0.80 0.60 1.15 1.44 1.40 1.37 1.12 1.27 0.83 Europe Portugal -4.30 1.21 9.49 5.58 0.75 2.51 2.23 0.86 0.84 1.30 0.72 2.59 1.12 0.94 1.67 0.61 Source: Hobcraft, McDonald, and Rutsrein 1985. Table 1 6A-3. Estimates_for Main Effects Parameters in Model of Postneonatal Mortality -Births in past 2 years Births in the past 2-4 years Mother's educa-ion - Bir-th or-der Moth-er's age at birith One One Two One One Two Fou-rth Seventh Less 35 oT Female Country Base alive dead or more alive dead or more Mediumn High to sixth or more First than 20 20-34 more child Africa Senegal -2.90 0.65 2.34 3.46 1.21 1.36 1.35 0.80 0.54 1.21 1.09 0.99 1.36 1.39 1.03 0.97 Benin -3.05 0.91 3.03 2.14 1.40 2.05 1.36 0.66 0.99 1.20 1.30 1.40 1.36 0.97 1.07 0.92 Egypt -3.26 2.08 2.83 3.16 1.17 1.67 1.84 0.99 0.70 0.96 0.93 1.75 1.36 1.06 1.14 1.17 C6te d'lvoire -2.88 1.21 2.16 1.46 1.04 1.63 1.70 0.86 0.67 1.17 1.13 1.45 1.32 1.09 0.84 0.89 Camneroon -3.12 1.84 2.80 3.42 0.92 1.51 1.46 0.59 0.46 0.99 1.12 1.65 1.17 1.12 1.22 1.13 Mauritania -3.47 0.90 2.92 2.10 0.66 1.40 0.81 1.04 1.52 0.88 1.86 1.00 1.20 0.94 1.04 1.17 Lesothio -3.50 1.73 2.18 2.80 1.32 2.20 1.93 1.02 0.76 1.77 1.68 1.26 1.93 1.27 0.79 1.03 Kenya -3.29 1.75 3.86 3.42 1.02 2.12 1.79 0.77 0.45 0.89 0.96 1.65 1.38 0.90 0.95 0.95 Morocco -3.79 1.82 3.06 2.94 0.97 1.43 1.46 1.13 0.87 1.16 1.73 1.68 1.46 1.23 0.92 0.96 Sudan -4.02 2.03 5.42 5.16 1.28 3.49 2.83 0.66 1.21 0.48 0.74 2.53 1.12 1.16 1.04 0.86 Ghana -3.60 2.39 4.01 3.86 1.02 3.32 1.72 0.58 0.68 0.88 0.84 1.14 1.28 1.22 1.17 0.78 Tunisia -4.91 2.51 8.67 6.05 0.80 1.58 1.58 0.81 0.01 1.23 1.88 1.42 1.57 1.42 0.73 1.01 Average -3.48 1.65 3.61 3.33 1.07 1.98 1.65 0.83 0.74 1.07 1.27 1.49 1.38 1.15 1.00 0.99 Asia anid the Pacific Yemen, Republic of -3.07 2.44 4.22 4.01 1.12 1.49 1.36 0.79 0.39 0.94 1.19 2.14 1.51 1.09 1.20 0.90 Nepal -2.88 1.27 1.97 3.82 1.19 1.84 1.67 0.88 0.76 1.07 1.23 1.39 1.54 1.30 0.84 0.99 Bangladesh -3.03 1.79 2.12 2.03 0.99 1.75 1.23 1.08 0.99 0.73 0.97 1.65 1.14 1.05 0.94 0.78 Pakistan -3.33 1.62 2.44 2.75 1.16 1.65 1.51 0.88 0.84 1.07 1.01 2.01 1.08 0.96 1.00 1.15 Indonesia -3.00 2.01 2.44 1.77 0.98 1.92 1.51 0.57 0.28 0.93 0.79 1.54 1.11 0.90 1.13 0.82 Th-ailand -3.70 1.95 3.39 2.64 1.11 2.12 1.99 0.82 0.33 0.82 1.19 1.62 1.04 0.89 0.76 1.06 Philippines -3.39 2.01 2.39 5.05 1.11 1.39 1.35 0.74 0.42 0.84 0.99 0.86 1.07 1.01 1.21 0.75 ~" Syrian Ar-ab)Republic -3.87 2.53 4.39 4.57 1.16 1.97 1.54 0.73 0.37 0.97 1.21 2.59 1.36 1.27 1.16 1.04 Jordan -4.29 3.94 3.97 8.17 0.99 1.26 1.84 0.65 0.73 0.93 1.05 2.48 1.34 0.92 0.59 1.16 SriLanka -4.28 1.46 1.52 2.69 0.98 1.86 1.63 0.91 0.80 1.45 1.95 1.06 1.30 1.17 0.66 0.92 Korea, Republic of -4.47 2.12 6.49 6.42 1.31 2.48 2.20 0.61 0.70 1.27 1.45 2.14 1.42 0.93 1.17 1.00 Malaysia -3.42 0.90 2.46 2.20 0.80 1.80 0.79 0.85 0.39 1.21 2.64 0.89 1.92 1.27 0.96 0.64 Average -3.56 2.00 3.15 3.84 1.07 1.79 1.55 0.79 0.58 1.02 1.31 1.70 1.32 1.06 0.97 0.94 Latin America and the Caribbean Haiti -2.97 1.80 2.80 2.80 1.20 1.73 1.99 1.92 0.87 0.84 0.86 1.46 0.95 1.06 1.09 0.84 Peru -3.21 2.14 3.03 3.03 1.08 1.82 1.72 0.58 0.32 1.03 1.07 1.49 1.46 1.12 0.98 0.90 Ecuador -3.52 1.68 3.22 3.13 0.93 1.67 1.60 0.95 0.39 1.20 1.52 1.68 1.35 1.04 0.83 0.82 Colombia -3.79 2.72 3.49 4.66 0.97 1.92 1.75 0.43 0.58 0.83 0.83 1.62 1.30 0.95 1.14 0.90 Mexico -3.66 1.84 2.05 2.64 0.81 1.63 1.65 0.87 0.27 0.97 0.95 1.20 1.34 1.12 1.03 0.84 Costa Rica -4.06 1.97 2.89 5.00 2.05 3.29 3.39 0.92 0.56 0.78 0.97 1.88 2.05 0.92 1.52 0.62 Guyan-a -4.26 1.70 4.06 3.39 1.73 2.94 1.60 1.03 0.88 0.70 0.76 1.42 0.95 0.89 0.82 0.94 Panama -4.88 1.16 0.77 0.73 1.21 2.27 0.68 0.74 0.39 1.23 2.80 1.13 2.61 2.86 1.23 1.13 Jamaica -5.34 2.64 9.21 1.63 1.70 2.29 1.23 1.07 0.90 1.01 1.65 1.43 3.22 1.27 1.40 0.91 Trinidad and Tobago -5.54 2.23 4.10 3.25 1.28 5.31 0.90 1.17 0.80 1.67 2.86 2.86 1.27 1.46 0.69 0.77 Average -4.12 1.99 3.56 3.03 1.30 2.49 1.65 0.97 0.60 1.03 1.43 1.62 1.65 1.27 1.07 0.87 Europe Portugal -3.85 2.01 0.42 4.35 1.06 0.99 2.05 0.59 0.30 0.85 0.87 0.70 2.48 1.72 1.38 0.90 Source: Hiobcraft, McDonald, and Rutstein 1985. Table 16A-4. Estimates fior Main Effects Parameters in Model of Toddler Mortality __________________ Births in past 2 years Births utp to one Birth order Mother's age at birth One One Two year later Mother's educatiOn Fourth Seventh Less 35 or Femnale Coun trv Base alive dead oyr moyre Births Preg-nant Medium High to sixth or more First than 20 20-34 moyre child Africa Senegal -2.44 0.54 1.11 0.75 0.01 0.95 0.75 0.44 1.21 1.36 1.05 1.28 1.00 0.87 1.02 Benin -3.45 0.79 1.45 0.40 2.18 2.12 0.61 1.00 1.73 1.16 0.75 2.18 1.25 0.97 0.91 Egypt -3.49 1.62 2.61 2.18 2.46 1.95 0.89 0.61 1.17 1.15 1.17 1.22 1.05 0.99 1.30 C6te d'Ivoire -3.12 1.00 0.91 3.42 1.92 2.05 0.75 0.67 1.00 0.87 1.16 1.22 1.01 0.86 0.83 Cameroon -3.30 1.46 2.39 0.61 3.35 1.42 0.77 0.59 0.90 0.78 1.11 1.42 0.97 0.73 0.90 Mauritania -3.66 0.79 1.80 1.7 3 1.86 2.36 0.89 0.76 1.36 1.15 0.84 1.15 1.20 1.23 1.14 Lesotho -3.23 1.04 1.58 2.48 3.39 2.86 0.87 0.56 0.78 0.83 0.61 1.36 1.51 0.87 0.62 Kenya -3.36 1.43 1.86 1.60 2.01 1.36 0.52 0.53 0.96 0.97 1.12 1.23 0.88 0.92 0.82 Morocco -3.95 1.82 2.77 1.60 2.01 1.48 0.38 0.00 1.19 1.26 1.09 1.49 1.45 0.98 1.04 Suidan -3.7 3 1.04 3.06 1.12 1.92 1.84 0.90 1.42 0.90 0.94 0.84 1.35 0.89 1.46 0.70 Ghana -3.67 1.26 1.65 0.00 4.95 1.79 0.73 0.55 1.22 1.28 0.87 1.43 1.04 0.84 0.82 Tunisia -6.62 1.46 2.48 3.56 3.94 4.31 1.57 0.03 1.65 1.97 1.21 1.75 1.58 0.78 0.93 Average -3.67 1.19 1.97 1.62 2.50 2.04 0.80 0.60 1.17 1.14 0.99 1.43 1.15 0.96 0.92 Asia anid the Pacific Yemen, Repuiblic of -3.62 2.80 5.42 2.14 2.12 1.99 0.55 0.75 0.94 0.64 1.84 0.62 0.77 1.02 1.22 Nepal -2.87 1.46 2.08 1.21 2.10 1.60 0.66 0.29 1.0.3 0.91 1.01 1.05 0.82 0.73 1.06 Bangladesh -4.60 1.79 1.45 2.92 2.69 2.69 0.86 0.76 1.40 1.38 1.28 1.77 1.46 1.03 1.31 Pakistan -3.91 1.48 1.73 0.99 2.14 1.86 0.79 0.21 1.32 1.40 0.83 1.60 1.31 0.89 1.40 Indonesia -3.73 1.70 2.41 1.90 3.13 2.75 0.83 0.26 1.27 1.45 0.96 1.46 1.12 0.48 0.88 Thailand -4.79 1.88 3.82 0.00 4.06 3.86 0.76 0.00 0.94 1.63 0.37 1.07 0.98 0.90 0.69 Philippines -4.06 1.27 2.20 1.32 1.58 1.63 0.68 0.36 1.20 1.46 0.56 1.55 1.04 0.73 0.98 Syrian Arab Republic -4.75 1.68 2.14 2.27 1.43 1.55 0.66 0.73 1.19 1.39 0.80 1.60 0.90 1.15 0.89 Jordan -4.18 2.01 3.1 3 5.26 1.48 1.55 0.65 0.30 0.76 0.68 1.04 0.99 0.88 0.44 1.08 SriLanka -5.67 1.52 1.34 2.34 4.14 0.78 1.06 0.68 2.61 2.61 1.57 2.66 1.77 0.81 1.13 Korea, Republic of -.6.49 2.97 2.56 0.07 4.66 5.47 0.71 0.23 1.58 1.16 2.32 1.88 0.70 1.86 2.72 Fiji -5.13 2.18 0.01 3.10 1.17 0.33 1.79 1.49 0.86 0.43 0.82 0.97 0. 12 0.33 0.66 Malaysia -5.35 0.98 3.53 0.01 1.62 1.38 1.17 0.00 1.17 1.90 0.30 1.43 0.99 1.77 0.53 Average -4.55 1.83 2.45 1.81 2.49 2.11 0.86 0.47 1.25 1.31 1.05 1.44 0.99 0.93 1.12 Latin America and the Caribbean Haiti -3.34 1.06 2.20 6.17 1.02 1.22 0.00 0.89 1.13 0.63 1.13 1.03 0.84 1.09 0.76 Peru -3.49 1.95 2.32 1.77 2.10 1.90 0.27 0.14 0.94 0.96 1.35 0.91 0.89 0.89 1.12 Dominican Republic -3.78 1.45 1.34 3.29 3.42 1.39 0.58 0.23 1.14 0.87 0.43 1.52 1.22 1.09 1.27 Ecuador -3.86 1.45 1.90 1.03 2.08 1.62 0.58 0.34 1.13 1.16 0.63 2.46 1.92 1.32 0.91 Colombia -4.46 1.62 2.2 3 3.00 1.67 1.08 0.76 0.22 0.84 0.61 0.61 1.54 1.51 1.40 1.60 Mexico -4.56 1.35 2.75 2.41 3.16 1.62 0.53 0.33 1.77 1.51 0.70 2.32 1.38 1.03 0.85 Paraguay -4.54 1.19 0.57 0.01 1.72 1.75 0.76 0.54 1.07 1.28 0.55 1.17 1.77 1.08 0.69 Costa Rica -4.76 0.96 2.12 2.92 2.34 1.17 0.44 0.28 1.30 2.14 0.85 2.61 1.62 0.92 0.98 Guyana -6.36 1.19 2.16 2.48 1.67 2.29 1.60 1.38 2.48 3.32 0.78 4.66 1.22 2.03 0.76 Venezuela -4.62 1.05 1.21 1.54 0.92 1.13 0.62 0.00 0.97 1.43 0.95 1.03 1.72 0.76 0.79 Panama -4.39 1.15 1.27 1.28 2.23 0.80 0.57 0.30 1.21 1.48 0.83 1.34 1.14 1.06 0.98 Jamaica -4.22 1.21 0.98 0.01 1.80 0.48 0.68 0.45 1.28 1.62 1.20 2.94 1.11 0.87 0.66 Trinidad and Tobago -6.55 0.95 0.01 0.01 2.64 1.79 1.43 0.91 0.78 1.90 0.66 2.20 2.61 0.90 2.44 Average -4.53 1.27 1.62 1.99 2.06 1.40 0.68 0.46 1.23 1.45 0.82 1.98 1.46 1.11 1.06 Europe Portugal -6.50 4.10 6.82 0.01 5.64 1.21 0.35 0.32 1.75 1.08 0.98 5.64 1.54 2.08 1.09 SOUrce: Hobcraft, McDonald, and Rutstein 1985. - ~ ~.,, -.~,* -. 4II VLUL&. UJ %,1I4LBU IVLUTLULLLL" BirthS in past 2 years Births up to I year later Mother's education Birth order Mother's age at birth One One Two One or None Fourth Seventh Less 35 oyr FemLale Countr Base alive dead or more more alive alive Pregnant Medium High to sixth or more First than 20 20-34 more child Af-rica Senegal -2.34 0.70 0.83 0.00 0.89 0.67 0.00 0.73 0.22 0.83 1.17 0.91 1.12 1.15 0.72 1.05 Benin -2.37 0.76 1,35 0.76 0.95 1.09 2.80 0.50 0.43 0.87 0.88 0.84 1.11 1.03 1.06 0.95 Egypt -3.54 1.73 1.60 1.99 1.40 1.54 2.39 0.82 0.57 1.43 1.30 0.64 1.22 0.94 0.83 1.15 C,&e d'lvoire -3.04 1.27 1.16 0.00 1.20 1.31 0.00 0.90 0.43 1.13 0.84 0.76 1.62 1.14 1.11 0.88 Cameroon -2.90 1.28 1.86 1.16 0.95 1.58 1.57 0.63 1.97 1.27 1.46 1.51 0.79 0.87 0.95 1.03 Mauritania -2.74 0.98 0.90 1.95 0.96 1.54 1.93 1.17 0.52 1.16 1.26 0.66 1.34 0.96 1.17 1.06 Lesotho -3.47 2.97 0.75 0.01 0.92 0.64 0.01 0.63 0.49 1.43 0.45 1.12 1.73 1.68 1.14 0.68 Kenya -3.57 1.21 1,70 1.75 0.99 1.09 0.8 3 0.59 0.74 1.05 1.49 0.78 1.40 0.96 1.16 0.95 Morocco -4.06 1.84 1.6.3 2.69 1.01 1.79 2.69 0.27 0.00 1.39 1.55 1.17 1.52 1.07 1.11 0.84 Suidan -3.68 1,19 1.36 1.73 1.19 3.03 1.93 0.52 2.29 1.21 1.23 0.71 1.26 0.86 1.54 1.27 Ghana -3.61 1.54 1.15 0.00 1.35 2.97 1.09 0.30 0.38 1.03 0.85 0.99 2.10 1.27 1.55 0.96 Tunisia -5.11 2.25 5.75 4.44 2.16 2.34 3.03 0.51 0.01 1.30 1.62 1.31 1.00 1.27 1.14 1.42 Average -3.37 1.48 1.67 1.38 1.16 1.63 1.52 0.63 0.67 1.18 1.18 0.95 1.35 1.10 1.12 1.02 Asia and the Pacific Yemen, Republic of -3.27 3.25 1.95 4.44 0.85 1.09 1.58 0.58 0.60 0.81 0.68 1.55 1.16 0.87 0.87 l.2 3 N"epal -3.36 1.31 1.08 1.90 1.31 2.14 3.13 0.65 0.24 1.52 1.58 0.78 1.67 1.21 0.59 1.07 Bangladesh -3.14 1.63 1.23 2.53 1.73 1.28 1.63 0.73 0.39 1.09 0.98 0.96 1.00 0.88 0.97 1.13 Pakistan -3.42 1.52 1.28 1.46 1.05 1.49 1.08 0.68 2.61 1.11 0.97 0.76 1.43 1.17 1.20 1.31 Indonesia -3.26 1.25 1.48 2.72 2.18 1.42 3.35 0.64 0.19 1.00 1.15 0.80 1.52 1.03 0.90 0.77 Thailand -4.34 1.55 0.85 3.10 1,65 2.27 3.16 0.73 0.19 1.58 1.60 0.64 1.14 1.17 0.56 1.57 Philippines -4.64 1.63 2.77 1.77 1.31 1.25 2.16 0.83 0.25 1.51 1.79 1.02 1.48 1.00 0.59 0.89 Syrian Arab Republic -4.91 1,23 2.75 1.16 1.17 1.32 1.67 0.31 0.16 1.62 1.35 0.84 0.91 1.09 0.41 1.32 ~" Jordan -4.94 1.93 1.90 3.46 1.11 1.51 0.84 0.12 0.22 1.14 0.64 0.7 3 1.65 1.07 0.53 0.79 SriLanka -4.10 1.08 0.79 0.50 1.72 2.32 0.46 0.68 0.35 1.21 1.73 0.56 1.14 0.85 0.55 0.96 Korea, Republic of -6.08 0.59 3.35 0.03 1.70 0.02 0.03 0,61 0.44 0.76 0.21 2.05 1.51 2.97 4.95 0.88 Fiji -5.38 0.67 1.04 301.87 0.81 0.00 0.00 1.12 1.73 1.14 0.77 0.28 0.54 0.97 1.04 1.05 Malaysia -5.80 0.77 1.19 6.89 1.58 0.01 0.02 0.58 0.38 2.25 1.79 0.47 4.57 0.77 2.69 0.90 Average -4.36 1.42 1.67 22.24 1.40 1.24 1.47 0.64 0.60 1.29 1.17 0.88 1.52 1.16 1.22 1.07 Americas and the Caribbean Haiti -2.80 1.40 1.03 0.00 1.06 1.02 3.13 1.27 0.52 0.73 0.37 0.76 1.55 0.94 1.39 1.06 Peru -3.93 1.35 1.21 1.48 0.98 1.42 4.35 0.30 0.15 1.46 1.65 0.73 2.27 1.39 0.95 1.20 Dominican Repuiblic -3.76 1.43 0.75 1.26 1.26 1.48 0.52 0.55 0.16 0.39 1.02 0.71 1.58 1.30 0.41 1.05 Ecuador -4.01 1.07 1.54 1.90 1.55 2.77 2.41 0.54 0.34 0.98 1.12 0.68 0.89 1.31 0.84 1.09 Colombia -3.90 1.31 1.52 0.85 1.28 1.93 4.62 0.62 0.18 1.08 0.88 0.92 1.04 0.83 1.55 1.02 Mexico -3.98 0.88 0.91 0.74 1.05 0.78 4.22 0.32 0.05 1.27 1.03 0.80 1.46 0.64 1.05 1.19 Paraguay -4.19 1.15 0.58 1.51 1.16 1.25 0.02 0.38 0.15 0.95 1.86 0.89 1.30 1.55 0.85 0.36 Costa Rica -5.32 1.17 0.57 2.46 1.30 2.59 4.35 0.63 0.00 1.79 1.21 0.80 0.76 0.68 1.42 1.49 Guiyana -4.63 1.27 0.80 0.00 1.04 0.00 0.01 1.07 0.47 1.20 0.85 1.49 0.00 0.42 1.04 0.91 Venezuata -5.56 2.14 0.01 5.75 1.09 3.25 4.85 0.38 0.12 1.88 1.36 2.41 1.30 1.63 0.01 1.23 Panama -5.20 0.85 0.78 0.01 1.86 0.01 3.39 0.35 0.32 1.58 2.0 3 1.45 2.16 1.34 0.53 1.31 Jamaica -7.51 6.11 8.76 0.04 2.59 0.01 0.01 0.19 0.21 3.74 2.61 24.53 0.78 0.56 1.04 1.51 Trinidad and Tobago -7.81 0.48 0.70 0.02 2.86 20.49 31.19 0.00 1.55 0.49 2.46 1.19 1.14 0.79 0.00 2.66 Average -4.82 1.59 1.47 1.23 1.54 2.85 4.85 0.51 0.33 1.35 1.42 1.00 1.25 1.03 0.85 1.24 Europe Portugal -9.86 4.57 0.00 0.00 2.18 0.00 0.00 3.00 6.75 6.36 15.64 1.14 16.12 6.17 1.25 0.80 Source: Hobcraft, McDonald, and Rutstein 1985. 358 Susan Cochrane and Frederick Sai Notes proportions currently using contraceptives are 2 5 percent, 12 percent, and 53 percent, respectively. The authors would like to thank Dr. Julie Da Vanzo of Rand Corporation 18. Because many of those who wish to space a child will go on to have Drs. Judith Fortney and Nancy Williamson of Family Health International further births, there is no correct assumption that would allow the conversion and Ms. Jane Nassim and Mr. Rodolfo Bulatao of the World Bank for helpful of spacing into excess births. Therefore, one-half was chosen as an arbitrary comments. Several anonymous reviewers also provided useul insights. Any figure. It is clearly incorrect to assume that all those who wish to space should romemaini. defcenc inythevapers a,of course, thefrespnsibilir oth be counted. Likewise it is incorrect to assume that the spacers account for no authors. excess fertility. In these models, 26 percent, 38 percent, and 14 percent of the authors. ~~~~~~~~~~~~~~~~women wish to space their next birth. I. The countries selected were, for the most part, countries in which survev y 9e Th to main th ent irth. dataallw usto ompre teseaggegat mesurs offeriliy an exess 19. The three main arguments why increased contraception will not Im- data allow us to compare these aggregate measures of fertility and excess provesurvival rates are (a) that reducing high parity births will result in a larger fertilitv with individual reports of excess fertility or the desire to cease proportion of births being first births and these births have even higher childbearing. Unfortunately, China, India. and Brazil have no World Fertility mortality than high-parity births; (b) that these correlations are not causative Survey data sponsored by tisAItI and UNFPA available tc) draw on for compari- but are associated with other characteristics of the mother or family, such as sons, but Brazil has participated in the more recent Demographic and Health low education and economic status; and (c) that changes in contraceptive use Survey, sponsoredl by USAID). Survey, sponsored by USA_DL may be associated with other changes in behavior, such as the reduction of 2. The dramatic nature of these declines in fertility can be observed by a decline by half in the TFR between the early 1960s and the late 1980s in breastfeeding, which will cause increased health risks. dColombinea,y Costalf Rica, theand t he Domincary R and byh40 percent or m 20. The biological mechanisms to explain these relationships have not been Colonbia, CostaRica, andthe DominicanRepublicandby 4percentormore identified (Haaga 1989). in Brazil, Ecuador, Mexico, and Venezuela. 21. This is an underestimate to the extent that it ignores the intervals in 3. Half or more of the developing countries of all regions,except central west which a child was bom in the preceding twelve months and then died. In those Asia perceived their population growth as being excessive. None of thle Asiauperceives their Ecopuaicand growthias Commission frcAriva NnESW tha cases, the causal issues are more complex. It also ignores the effect of the countr-ies in the Economic and Social Commission for Africa (E-CWA) had postponed birth on the mortality risk of previous children. that perception (United Nations 1989, p. 14). 22. Here and throughout, 1987 U.S. dollars are used. 4. Governments of 68 of 131 developing countries have reported to the 23. This estimate is based on the cost of Community-based distribution United Nations that fertility is too high in their country (United Nations (CBD) programs as estimated from a number of countries by Cochrane, Ham- 1989,p. 14). (s)rgasssiaefoaubrfonrebCcrn,Hm 5. See the National Research Council's review forderailed discussion of the mer, Janowitz, and Kenney (1990). h SealthcoNsequencReseforchwComeandchildsreniew finrdevalelpinscuiount ofh 24. This corresponds to a mid-range of costs for a couple-year of protection health consequences for women and children In developing countres of contraceptioni and reproduction (NRC 1989). forchnic-based distribution of oral contraceptives and intrauterine contracep- 6. This summation represents a simplification. Ina period of transition,such tive devices (Cochrane and others 1990). as thebabvboommtion thepUenited iStates afteriWorld Wnap tr the T ,Rscan be 25. The health costs of high fertility and close spacing of children are as the baby boom in the United States after World War 11, the TFR can be much higher than the average parity. Such a situation occurs if women have discussed in chapter 17, this collection. Family planning can reduce matemal mortalitv in two ways: (a) the fewer the births a woman has, the fewer her been postponing their first and second births and all women of different ages exposure e s to the risk of matemal mortality, and (b) by confining births to the are having low-parity births at the same time. healthiest age groups and spacing births to the best intervals, the risk of death 7. It should k e noted that in these microeconomic studies, the percentage associated with every birth that does take place is reduced. There exist various at risk eecause of high parity Is much larger than the percentage by which the estimates of the effect of family planning on mortality. It has been estimated hgur eeds f(ur.The reader should bear this in mindwheninterpreting in that 24 percent of matemal deaths could be averted by contraceptive use by figures ini table 16-3. 8. See footnotes to tabte 16-3 for descriptions ofhow the number ofwomen fecund women not currently using contraception but desiring no more births 8. Seeis fotngotes totable 16-3gned for datesrip s of howess birthe . numbero(Sai and Nassim 1987). in various categories are assigned to categories of excess births. There has been somewhat of a revision of position in the development 9. Asia's pattem is distorted by the verv low fertility in China. community in the United States and the development agencies away from the 10. More recent data from Kenya would give lower excess fertility, 43 and dire predictions of the negative consequences of rapid population growth (NRC 83 percent, respectively, by these measures butt higher excess fertility by 1986) just at the time that some Latin American and French scholars and preference measures. African politicians are becoming more concerned (Blanchet 1988; Paiva 11. Although most surveys report data only for women, a number ofsurveys 1988). report men's familv size preferences. Contrary to what is generally believed, 26. See Birdsall, Cochrane, and Van der Gaag (1987) for a review of the men do not systematically report higher fertility preferences than women methodological issues and estimates of child costs in industrial countries. See (Mason and Tat 1987). Perhaps the men do ear some costs of higher fertility Lindert (1980) for a review of the estimates available for developing countries. by having to work harder to support larger families in those environments In addition Deaton and Muelbauer (1986) have recently estimated the costs where marriages are stable. of a child in Sri Lanka and Indonesia. 12. Time constraints have not allowed the recalculation of these artificial 27. One reason it is so hard to determine the consequences of high fertility TFRs for more recent years. is that the causation goes in two directions and income can affect fertility as 13. Bongaarts, Mauldin, and Phillips (1990) estimate unwvanted fertility well as the converse. preferences for thedevelopingworld, includingChina, at 21 percent. InChina 28. Neither is there a well-defined range of the cost of a child, since the and a number of countries in Africa, there is deficit fertility which has not effect of a child on his or her parents and other siblings varies substantially been netted out of these estimates. from one environment to another. Deaton and Muellbauer (1986) estimate 14. In every country the youngest women report lower fertility preferences that patents spend about 30 to 40 percent of what they spend on themselves than the oldest women. Part of this may result from reporting bias mentioned on a child in Sri Lanka and Indonesia. Because they do not include the time above and part from genuine declining preferences. or opportunity costs that the child imposes on others in the family, these 15. A birth postponed will reduce the rate of growth in the short term. 1I1 expenditures are only part of the costs to a family of an additional child. addition, many births that are postponed never take place. 29. See King 1970 for an example for Jamaica. 16. More recent Demographic and Health Surveys show that more than 20 30. The exchange rate between the peso and the U.S. dollar was 120 in percent of the women wish to cease childbearing in Burundi, Ghana, and 1983. Ondo State in Nigeria. 31. These estimates are too large to the extent that there are economies of 17. The proportions of women who wish to limit their fertility in the three scale, and thus marginal costs ofa child are below average costs. The estimates countries are 30 percent, 23 percent, and 64 percent, respectively. The aremadeassumingagovemmentcommitmenttouniversalprimaryeducation Excess Fertility 359 and a progression rate from primary to secondary education similar to current 45. The data for Mexico were from 1978. It is interesting to note that once patterns in the country. Costs are obtained from two World Bank intemal the govemment undertook support for family planning about that time, there documents: "Comparative Education Indicators" for recurrent costs and "Unit was a rapid decline in fertility. Cost Estimates" for the capital costs. These costs have been inflated to 1987 as a base. Because no uinit recurrent costs are available for secondary school we have assumed that the ratio of recurrent secondary recurrent costs to primary recurrent costs reflects the ratio of their respective capital costs per References school place. There is no comparable measure of the health costs to be saved by a birth averted. Financing Health Services in DevelopingCountries: An Agenda Ainsworth, Martha. 1985. Family Planning Programs: The Clent's Perspectve. for Reform (World Bank 1987) provided the per capita health expenditure for World Bank Staff Working Paper 676, World Bank, Washington, D.C. a range of countries that can be used as a first approximation. Forthe two poor, Barnum, Howard N. 1983. Cost-Effectiveness of Family Planning Services in high-mortality countries the annual per capita public expenditure on health Nepal. Technical Notes, RES 9. Population, Health, and Nutrition Division, of Sri Lanka and China have been used, $6. For the low-mortality country an World Bank, Washington, D.C. average of seven countries has been used, yielding $28 per capita. The Birdsall, Nancy, Susan H. Cochrane, and Jacques van der Gaag. 1987. "The education costs have not been adjusted to reflect the age structure of mortal ity. Cost of Children." In George Psacharopoulos, ed., Economics of Education: The education figures would be at least 17 percent lower if the probability of Research and Studies. New York: Pergamon Press. survival to age five were used to adjust the figures. The health expenditure for Blanchet, Didier. 1988. "Estimating the Relationship between Population each country does reflect mortality to the extent that the number of years Growth and Aggregate Economic Growth in LDCs: Methodological Prob- included depends on the life expectancy in the model country. A more lems." Paper presented at the meeting of the United Nations Expert Group sophisticated method would weigh health cost in each year by the stirvival on Consequences of Rapid Population Growth in Developing Countries. cohort for that age. This would require information on expenditure by age, New York, N.Y., August. which is not availahle. The authors of chapter 17 in this collection are preparing estimates of costs that could be used in the calculations here as well Bongaarts, Jon. 1978. "A Framework for Analyzing the Proximate Determi- ifa- g profil of exedtr wer toh ,sd nants of Fertility." Populauion and Development Review 4: 105-3 2. If an age profi(e of expend1rnuTe were tro ne Lised- 32. In Northem Nigeria in 1981 the medianageoffirst marriage wasfifreen, . 1987. "Does Family Planning Reduce Infant Mortality?" Population which means that half the women are married by that age or earlier (Nigena and Development Review 13:323-24. Ferility Survey 1981/1982, published in 1984 by the National Population . 1988. "Does Family Planning Reduce Infant Mortality? Reply." Pop- Bureau). ulation and Development Review 14:188-90. 33.Thehealthbenefitsofbreastfeedingaresubstantialbut mustbebalanced Bongaarts, Jon, W. Parker Mauldin, James F. Phillips. 1990. "The Demo- against the time and nutritional costs to the mother. This is a topic that has graphic ImpactofFamily PlanningPrograms."PopulationCouncil Working been extensively discussed in the literature. Paper 17. New York. 34. A more comprehensive list of failure rates hy method and study is Boulier, Bryan L. 1986. "Family Planning Programs and Contraceptive Avail- available in Trussell and KrOst (1987). Their estimates include an estimate of ability: Their Effects on Contraceptive Use and Fertility." In Nancy Birdsall, failure rates for Norplant (a contraceptive implant) of 0.2 compared with 2 to ed., The Effects of Family Planning Programs on Fertility in the Developing 2.5 for oral contraceptives. World. World Bank StaffWorking Paper 677, Population and Development 35. These estimates are taken from general data sets. For actual policy 2. Washington, D.C. analsis in a country, detailed analysis o.f the family planning delivery system Bruce, Judith. 1990. "Fundamental Elements of the Quality of Care: A Simple would be necessary. Framework." Studies in Family Planmng 21(2).61-91 36. There is sonie evidence that real costs per user have dropped since 1980 in Indonesia even when costs of the health delivery of the family planning Bulatao, Rudolfo A. 1985. Expenditures on Population Programts in Developing program are included (World Bank 1990). Regions: Current Levelsand Future Requirements. World Bank StaffWorking 37. There are two main reasonis why cosLs would drop the more women are Paper 679, Population and Development 4. Washington, D.C motivated to cease childbearing. First, the more moitivated the women, the Casterimne,John B. 1990. "IntegratingHealthRiskConsiderationandFertility more likely they are to be using contraception. If there are economies of scale, Preference in Assessing the Demand for Family Planning." World Bank then costs would fall. Second, the more motivated women are, the less money consultant report. Washington, D.C. needs to be spent on motivation and the less extensive the deliverv system CDC (Centers for Disease Control). 1983. "Family Planning Methods and needs to be, because presumably the more motivated womnen will travel further Practices: Africa." Health Promotion and Education, Division of Reproduc- to get services. tive Health, Atlanta, Ga. 38. Using the lower estimate of costs per birth averted, one gets estimates Cliao, Dennis N. W., John Ross, and David Piet. 1985. "Public Expenditure of $238, $191, and $ 121, respectively. Impact: Education and Health, Indonesian Family Planning.": BKKBN/USAID, 39. The conclusions remain unchanged if the lower estimate of the costs Jakarta. per birth averted is used rather than the higher. Chemichovsky, Dov, Henry Pardako, David de Leeuw, Pudjo Raherje, and 40. The cost of the materials alone is $17, which is high compared with Charles Lerman. 1989. In "The Indonesian Family Planning Program: An approxiniately $2 a year for the oral contraceptive but almost identical to the Economic Perspective." Typescript. East Asia and Pacific Country Depart- five-year costs of injections. That cost has to be incurred up front, thus ment I (Philippines). discouraging many programs. Cochrane, Susan. 1983. "The Effects of Education and Urbanization." In R. 41. These expenses cover iniversal primary edtication and public health A. Bulatao and Ronald D. Lee, eds., Detertminants of Fertility: A Sumtnary of expenditures per capita on the level of Sri Lanka and China. 42. Cochrane and Zachariah (1983) showed that reducing infant and child Knowledge. Washington, D.C.: National Academy of Sciences. mortality was a more cost-effective way to reduce fertility than family planning Cochrane, Susan, and J. L. Cochrane. 1971. "Child Mortality and Birth: A in some very high mortality countries that had a low proportion of women Micro-Economic Approach." American Statistical Association and Proceed- wanting no more children. The data in that case applied to Kenya. ings. 43. In Indonesia, for example, religious leaders have ruled that any irrevers- .1974. "Child Mortality and Desired N umber of Children and Births." ible change cannot be )ustified except on health grounds (World Bank 1990). Australian Economic Papers, 13. 44. Johnson-Acsadi and Szykman (1980) and Ainsworth (1985) compiled Cochrane, Susan, Valerie Kozel, and Harold Alderman. 1990. Household data on the reason for nonuse of contraception for six and ten countries, Consequences of ligh Fertility: The Case ofPalostan. World Bank Discussion respectivelv, from data from the late 1970s or early 1980s. Paper 111. Washington, D.C. 360 Susan Cochrane and Frederick Sai Cochrane, Susan H., and K. L. Zachariah. 1983. Infant and Child Mortality as Kelley. Alan. 1988. "Economic Consequences of Population Change in the a Deterninant of FerUlity: The Policy Implications. World Bank Staff Working Third World." Journal of Economic Literature 26:1685-1728. Paper 556. Washington, D.C. King, Elizabeth M. 1987. "The Effect of Family Size on Family Welfare: What Cochrane, Susan, Jeffery Hammer, Barbara lanowitz, and Genevieve Kenney. Do We Know?" In D. Gale johson and Ronald D. Lee, eds., Population 1990. "The Economics of Family Planning." Typescript. Population, GrowthandEconomicDevelopment:IssuesandEvidence. Madison: University Health, and Nutrition Division; World Bank, Washington, D.C. of Wisconsin Press. David, Henry P. 1986. "Unwanted Children: A Follow-Up from Prague." King, Timothy. 1970. The Measuremnent of Economic Benefits from Family Family Planning Perspectives 18(3):143-44. Planning Projects and Programs. World Bank Working Paper 71. Washing- Deaton, Angus S., and John Muellbauer. 1986. "On Measuring Child Costs: ton, D.C. With Applications to Poor Countries." Journal of Poltical Economy Kiranandana, Thienchay, Suchada Kiranandana, Suwanee Surasieng Sunk, 94(4):720-44. Rachanivran Uthaism in collaboration with others. 1984. "Planning for Demeny, Paul. 1965. "Investment Allocations and Population Growth." De- Family Planning Programs and Social Services in Thailand." Chulalong- mography 2:203-32. kom University. DHS (Demographic and Health Surveys). Various years, Various countries. Kwast, Barbara, W. Kidav-Mariam, E. M. Saed, F. G. R. Fowkes. 1985. Columbia, Md.: Institute for Resource Development. "Epidemiology ofMatemal Mortality in Addis Ababa: A Community-Based Duza, M. B., and C. S. Baldwin. 1974. Nuptiality as an Area for Policy Interven- Study." Paper presented at the World Health Organization International tion: A Comparative Perspective. New York.: Population Council. Meeting on Prevention of Maternal Mortality, November 11-15, 1985, tion A ompaadv Perpecive.NewYork: PputaionCounil.Geneva. Enke, Stephen. 1960. "The Gains to India from Population Control: Some Money Measures and Incentive Schemes." Review of Economics and Statistics Labbok, M. H., and M. McDonald, eds. 1990. "Proceedings of the Interagency 17(2):175-81. Workshop on Health Care Practices Related to Breastfeeding." International Figa-Talamanca, Irene, T. A. Sinnalhuvary, K. Yusaf, and Chee Kin Fong. JournalofGynecologyandObstet-ics 31(supplement 1). 1986. "Illegal Abortion: An Attempt to Assess Its Cost to the Health Lancet. 1988. "Breastfeeding as a Family Planning Method." ii: 1204-5. Service and Its Incidence in the Community." nternanonalJournal of Health Lapham, Robert and George Simmons. 1987. Organizing for Effectve Family Services: Health and Social Policy 16(3):375-89. Planning Programs. Washington, D.C.: National Academy Press. Gillespie,Duff.H.E.Cross,J.G.Crowley,andS.R.Radloff. 1988. "Financing Lee, Nancy, Herbert Peterson, and Susan Chu. 1989. "Health Effects of the Delivery of Contraceptives: The Challenge of the Next Twentv Years." Contraception." In National Research Council, Contraceptive Use and The Demographic and Pnogramnmaizc Consequences of Contraceptive Innovatons. Controlled Fertbty. Washington, D.C.: National Academy Press. Committee on Population, National Academyof Sciences, Washington, D.C. Lightboume, Robert E. 1987. "Reproductive Preferences and Behavior." In Green, Cynthia P. 1989. "Media Promotion of Breastfeeding: A Decade's John Cleland and Chris Scott, eds., The World Fertlity Survey: An Assess- Experience." Nutrition Communication Project, conducted by the Acad- ment. London: Intemational Statistics Institute. emy for Educational Development for U.S. Agency for Intemational De- Lindert,Peter H. 1980."ChildCostsandEconomicDevelopment."lnRichard velopment, Washington, D.C. A. Easterlin, ed., Population and Economic Change in Developing Countries. Haaga, John G. 1989. "Mechanisms for the Association of Maternal Age. Chicago: University of Chicago Press. Parity, and Birth Spacing with Infant Health." In National Research Liskin, L. S. 1980. "Complications of Abortions in Developing Countries," Council, Contracepnve Use and Controlled Fernlity. Washington, D.C.: Na- Population Reports Series F 7 (July). tional Academy Press. Maine, Deborah. 1985. "Maternal Mortality in the Third World." People Henshaw, Stanley K. 1987. "Induced Abortion: A Worldwide Perspective." 12(2):6-8. International Family Planning Perspective 13(1):12-15. Mason, Karen Oppenheim, and Anju Malhotra Taj. 1987. "Gender Differ- Hobcraft, John. 1983. "Child Spacing Effects on Infant and Early Child encesinReproductiveGoalsinDevelopingCountries."PopulationStudies Mortalit'." Population Index49(4):585-618. Center Research Report 87-105. University of Michigan, Ann Arbor. 1987. "Does Family Planning Save Children's Lives." Paper prepared National Population Bureau. 1984. The Nigerian Fertlity Survey Principal for the Intemnational Conference on Better Health for Women and Chil- Report. 2 Vols. London: International Statistical Institute. dren through Family Planning. Nairobi, October 5-9. Nortman, Dorothy. 1986. "Cost-Benefit Analysis of the Mexican Social Hobcraft, J. N., J. W. McDonald, and S. 0. Rutstein. 1985. "Demographic Security Administration's Family Planning Program." Studies in Family Determinants of Infant and Early Child Mortality: A Comparative Analy- Planning 17(1):1-6. sis." Population Studies 39(3):363-86. Nortman, Dorothy L., and Gary L. Lewis. 1986. "A Time Model to Measure Holck, S., and H. Bathija. 1988. "Safety and Efficacy of Fertility Relating Contraceptive Demand." In John A. Ross, ed., Survey Analysis for the Methods." In E. Diczfalusy, P. D. Griffin, and J. Khanna, eds., Research in Gudance of Family Planning Programs. Liege: Ordina. Human Reproduction: BiennialReport(1986-1987). Geneva: World Health NRC (National Research Council). 1986. Working Group on Population Organization Special Programme on Research, Development, and Research Growth and Economic Development, Committee on Population, Commis- Training in Human Reproduction. sion on Behavioral and Social Sciences and Education, eds., Populaton Huffman, Sandra, and C. Combest. 1988. "Promotion of Breastfeeding: Yes, Growth and Economic Development: Policy Quesnons. Washington, D.C.: It Works!" Center to Prevent Childhood Malnutrition, Bethesda, Md. National Academy Press. Hutchings, Jane, and Lyle Sanders. 1985. "Assessing the Characteristics and - . 1989. Contraception and Reproducton: Health Consequences forWomen Cost-Effectiveness of Contraceptive Methods." Paper 10. PIACT (Program andChildrenin the Developing Wortd. Washington, D.C.: National Academy for the Introduction and Adaption of Contraceptive Technology), Seattle, Press. Wash. _- 1990. Developing New Contraceptives: Obstacles and Opportunities. Jam, Anrudh K. 1989. "Assessing the Fertility Impact of Quality of Family Washington, D.C.: National Academy Press. Planning Services." Population Council, New York. Ohlin, Goran. 1967. "Population Control and Economic Development." Johnson-Ascadm. Gwendolyn, and Maurice Szykman. 1980. "Selected Char- Development Center for the Organization for Economic Cooperation and acteristics of 'Exposed' Women Who Wanted No More Children but Were Development (OECD), Paris. Not Using Contraception." In Survey Analysis for the Guidance of Family Paiva, P. D. 1988. "Rapid Population Growth and Economic Development in Planning Programs. Liege: Ordina. Latin America." Paper presented at the meeting of the United Nations Excess Fertility 361 ExpertGrouponConsequencesof Rapid Population Growth in Developing Experience wlith Cost-Effectiveness Analysis and Cost-Benefit Analysis. New Countries, New York, August. York: St. Martin's Press; London: Croom Helm. Population Crisis Committee. 1985. "Issues in Contraceptive Development." Trussell, James. 1988. "Does Family Planning Reduce Infant Mortality? An Population. Briefing paper 15 (May). Exchange." Population and Development Review 14:171-78. Population Information Program. 1980. "Complications of Abortion in De- Trussell, James, and Kathryn Krost. 1987. "Contraceptive Failure in the veloping Countries." In Population Reports: Pregnancy Terminatons. Series United States: A Critical Review of the Literature." Studies in Family F 7. Baltimore: Johns Hopkins University. Planning 18(5):237-83. Potter, Joseph E. 1988. "Does Family Planning Reduce Infant Mortality?" Trussell, James, and A. R. Pebley. 1984. "The Potential Impact of Changes in Population and Development Review 14:170-87. Fertility on Infant, Child, and Maternal Mortality." Studies in Family Plan- Potts, Malcom, J. J. Speidel, and E. Kassel. 1977. "Relative Risks of Various ning 15(6):267-80. Means of Fertility Control when Used in Less Developed Countres." In J. J. tiN (United Nations). 1986. DemographicYearbook. 38. New York: Department Sciarra, G. Zatuchni, andj. J. Speidel, eds., Risks, Benefits, andControversies in of Intemational Economic and Social Affairs, Statistical Office. Fertilirs Control. Arlington, Va. (1978. Hagerstown, Md.: Harper and Rowe.) - 1988. World Population Trends and Policies: 1987 Monionng Report. Ross, John A., Marjorie Rich, Janet Molzan, and Michael Pensak. 1988. New York. "Family Planning and Child Survival: 100 Developing Countries." Center . Population Division. 1987. Fertlity Behavior i the Context of Develop- for Population and Family Health, Columbia University, New York. ment: Evidence from the World Fertiity Survey . New York. Sai, Fred and Janet Nassim. 1987. "The Role of Intemational Agencies, .__ S 1 N Governments, and the Private Sector in the Diffusion of Modem Contra- -1989. Trends in Population Policy. Population Studies 114. New York. ception." Technology In Society 9:497-520. Ware, Helen. 1976. "Motivations for the Use of Birth Control: Evidence from Salkever, David, and Ismail Serageldin. 1983. "Cost-Benefit and Cost-Effec- West Africa." Demography 13(4):479-93. tivenessAnalysisinPopulationPrograms:ItsRoleinProgramPlanningand Westoff, Charles F., and Lorenzo Moreno. 1989. "The Demand for Family Management." In Ismael Abdel-Hamid Serageldin, David S. Salkever, and Planning: Estimates for Developing Countries." Paper presented at the Richard W. Osbom. eds., Evaluating Population Programs: International Ex- Intemational Union for Scientific Investigation of Population (lussp) perience with Cost-Effectveness Analysis and Cost-Benefit Analysis. New York: Seminar on The Role of Family Planning Programs as a Fertility Determi- St. Martin's Press; London: Croom Helm. nant, Tunis. Scrimshaw, Susan. 1978. "Infant Mortality and Behavior in the Regulation of WFS (World Fertility Surveys). Various years. International Statistical Insti- Family Size." Population and Development Revieu 4:383 404. tute. The Hague. - 1983. "Infanticide as Deliberate Fertility Regulation." In R. A. World Bank. 1984. Population Change and Economic Development. (World Bulatao and R. D. Lee, eds., Determinant of Ferility in Developing Countries. Development Report, 1984). New York: Oxford University Press. New York: Academic Press. _- 1987. Financing Health Services in Developing Countries: An Agendafor Serageldin, Ismail. 1983. Evaluating Population Programs: International Experi- Reform. Washington, D.C. ence with Cost-Effectiveness Analysis and Cost-Benefit Analysis. New York: St. . 1990. Indonesia: Family Planning Perspectives in the 1990s. World Bank Martin's Press; London: Croom Helm. Country Study. Washington, D.C. Shorter, Edward. 1977. The Makingof the Modem Family. New York: Basic Books. WHO (World Health Organization). 1988. "Research in Human Reproduction, Speidel, J. J. 1983. "Cost Implications of Population Stabilization." In Ismail Biennial Report (1986-1987)." Special Programme on Research, Develop- Serageldin and others, eds., Evaluating Population Programs Internationial ment, and Research Training in Human Reproduction. 17 Maternal and Perinatal Health Julia A. Walsh, Chris M. Feifer, Anthony R. Measham, and Paul J. Gertler More than 20 percent of the population in developing coun- before pregnancy was ever contemplated, will result in com- tries are women in their reproductive years (United Nations plications many years later. The factors that affect the mother 1988). During a woman's life, one of the greatest risks to her and newbom can be divided into three: those that occur prior health is childbearing. Pregnancy brings high risks of sickness, to conception, so that a woman enters into the pregnancy in complications of delivery, disability, and death. Moreover, a precarious state of health; those during pregnancy (prenatal when pregnant women have complications, the infants these care may eliminate these); and those during delivery. Good women bear are at increased risk of low birth weight, illness health care will decrease these hazards. To reiterate, illness in resulting from the complications of delivery, disability, and the mother puts the newborn at greater risk. perinatal mortality. This chapter treats maternal and perinatal Unfortunately, prevention is relatively difficult: the lengthy health problems together because they are inseparably linked: list of causes of disease and social conditions associated with the main risk factors for disease and death among mothers and problems makes this evident. Obviously, no single, highly their newborns are the same. Not surprisingly, many of the effective control measure, such as a vaccine, drug, or vector interventions simultaneously improve both maternal and fetal control method, exists. Attention to women's general health health. care needs and readily available prenatal and obstetric care, Maternal and perinatal mortality rates are indicators of the however, can prevent the preponderance of perinatal and health of women of reproductive age and an indirect measure maternal morbidity and mortality. of the quality of the health care system. In many places health care during pregnancy is the only contact with modern medi- Public Health Significance cine that a woman seeks. Obstetric services, therefore, repre- sent a link to the general health care system and thus have an The size of the problem of maternal death and disability has effect on present and future pregnancies and on the well-being been underestimated and poorly recognized. Data on deaths of the entire family. High-quality supportive health services and complications of mothers have not been collected, nor reinforce the family's use of preventive and promotive services have data on the consequences for family and child health. in the future. In table 17-1 we detail the risk factors for maternal and Current Levels and Trends in the Developing World perinatal health problems. The main ones include illiteracy, poverty, poor nutrition, and low weight prior to pregnancy, Maternal deaths and illnesses, low birth weight, and perinatal minimal weight gain during pregnancy, first pregnancy or deaths are underreported in most countries. The best figures higher than fourth pregnancy (grand multipara), maternal age for the developing world are merely estimates pieced together younger than twenty or older than thirty-four years, poor from multiple sources of varying degrees of reliability. Little outcome of prior pregnancies, infections and illnesses during information exists which would allow quantification of the pregnancy, smoking, and inadequate health care during preg- severity and duration of disability. nancy and delivery. This complex web of social and medical factors suggests that health service solutions will be inadequate MATERNAL MORTALITY AND MORBIDITY. The World Health without concurrent attention to the other areas mentioned. Organization (WHO) estimates that 500,000 women annually The health problems discussed in this chapter are more die of complications of pregnancy and delivery. Ninety-nine complex than many of the others covered in this collection. percent of these deaths occur in the developing world. Mater- In the first place, they result from pregnancy, which is not a nal mortality is usually defined as death occurring while the disease but is often a sought-after and highly desirable condi- woman is pregnant or within forty-two days of termination of tion. Further, the lives of two people, mother and infant, are pregnancy. The "maternal mortality ratio," the number of involved. Finally, events early in the life of the mother, years maternal deaths per 100,000 live births, ranges from 25 to 363 364 Julia A. Walsh, Chris M. Feifer, Anthony R. Measham, and Paul . Gertler Table 17-1. Risk Factors for Maternal Morbidity and Mortality, Low Birth Weight, and Perinatal Mortality Adverse Outcome Maternal morbidity Penrnatal Preventive or treatment Risk factor and mortality Low birth weight mortality measures available Prepregnancy-Demographic risk factors Age (bless than twenty; more than thirty-four) Yes Yes Yes No Race I No Yes No No Low socioeconomic status Yes Yes Yes No Unmarriedb Yes Yes No No Low level of educationh No No Yes No Medical risk factors Number of children (none or more than four)a Yes Yes Yes Yes Low matemal weight-for-height (poor nutritional status)a Yes Yes Yes Yes Short patemal height No Yes No No Diseases such as diabetes and chronic hypertension Yes No Yes Yes Poor obstetric history2C Yes Yes Yes No Gynecologic abnormalities Yes Yes No No Matemal genetic and related factors No Yes No No Conception Male fetusa No Yes Yes No Pregnancy-Medical risk factors Multiple pregnancy5 No Yes Yes No Poor weight gain (maternal and fetal)a Yes Yes Yes Yes Anemia/abnormal hemoglobin Yes Yes Yes Yes Malaria Yes Yes Yes Yes Streptococcus (group B) infection Yes Yes Yes Yes Sexually transmitted diseasesA Yes Yes Yes Yes Urinary tract infection No Yes No Yes Rubella, cytomegalovirus No Yes Yes No Respiratory and diarrheal disease No Yes Yes Yes Short interpregnancy interval Unknown Yes Yes Yes Induced abortion (especially illegal) Yes No No Yes Ectopic pregnancy Yes No No Yes Hypotension No Yes No No Hypertension/pre-eclampsia/toxemia Yes Yes Yes Yes Low blood volume No Yes No No First or second trimester bleeding No Yes No No Placental abnormalitiesaf Yes Yes Yes No Hyperemesis No Yes No No OligohydT-amnios/polyhydramnios No Yes Yes No Inadequate health care Yes Yes Yes Yes Isoimmuni7ation No Yes Yes No Fetal anomalies No Yes Yes No Incompetent cervix No Yes Yes Yes Spontaneous premature rupture of membranes No Yes Yes Yes Behavioral and environmental risk factors Unwanted pregnancy Yes Yes Yes Yes Smoking/tobacco use" Unknown Yes Yes Yes a Alcohol and other substance abuse No Yes Yes Yes Diethylstilbestrol (DES) and other toxic exposures No Yes Yes No High altitude No Yes Yes No Absent or inadequate prenatal care Yes Yes Yes Yes Delivery Inadequate obstetrical care Yes Yes Yes Yes latrogenic prematurity No Yes Yes Yes Maternal and Perinatal Health 365 Adverse Outcome Maternal morbidity Perinatal Preventive or treatment Risk factor and mortalitY Low birth weight mortality measures available Postpartum/neonat period Inadequate care for mother~l Yes No No Yes Inadequate care for infant' No No Yes Yes a. High relative risk and/or very common. Retlects the relative risk and the proportion of the poptilation with che risk factor. b. Closely associated factors with increased risk in different studies. The independent contribution of each factor is difficult to disaggregate. c. Previous low-birth-weight-infant, maternal morbidity, perinatal morbidity, multiple spontaneous abortions, infertility treatment. d. Small pelvis, female circumcision, uterine disease, tubal scarring secondary to sexually transmitted diseases and p(otentially leading to ectopic pregnancy, and similar problems. e. For example, low maternal weight associated with own birth. f. Such as placenta previa and abTruptio placentae. g. Including occupational hazards. h. Prenatal care appears to decrease matemal and infant disease However, it may have less effect than expected from comparisons of the outcome of preg- nancy between women who attend prenatal care and those who do not. Those women who voluntarily choose prenatal care usually are healthier and have fewer risk factors. Source: Authors. 1,660 in studies from developing countries, and averages 10 in The "maternal mortality rate" is the number of maternal industrial ones. As reported in table 17-2, the highest overall deaths in one year per 100,000 women of reproductive age, ratio occurs in Africa (640); the ratios are lower in Asia (420) usually age fifteen through forty-nine. This rate combines the and in Latin America (270). Differences between countries, fertility rate (births per thousand women of reproductive age) and between urban and rural areas, are blurred by the regional and the maternal mortality ratio defined above, so it is influ- statistics, however. A recent study of the rural areas of the enced both by the likelihood of becoming pregnant and by the Gambia found a maternal mortality ratio of 2,200 (Greenwood risk of dying from that pregnancy. Improvements in both and others 1987). The maternal mortality ratio measures the family planning and obstetric services affect the matemal obstetric risk in a given pregnancy. mortality rate (Fortney 1987). The lifetime risk of maternal mortality is many times greater than the ratio indicates because the ratio ignores the effect of Table 17-2. Estimiated Maternal Mortality, repeated pregnancies. Each pregnancy adds to the total life- by Region time risk. In the developing world (excluding China) an Maternal average woman faces a lifetime risk of one chance in thirty- Maternal mortality rate three that a pregnancy will result in her death. For those with Live births deaths (per 100.000 serious risk factors or for those living in areas with inadequate Region (millions) (thousands) live births) health services and high fertility rates, the lifetime risk esca- Africa 23.4 150 640 lates greatly. The difference between industrial and developing North 4.8 24 500 nations is much larger for maternal mortality than it is for West 7.6 54 700 infant mortality: the risk of infant death is about 9 times greater East 7.0 46 660 in the least industrialized countries, but for maternal mortality Middle 2.6 18 690 the risk can be more than 100 times as great (wHiO 1986). Southern 1.4 8 570 Three-quarters of all maternal deaths can be attributed to Asia 73.9 308 420 one of three causes-hemorrhage, sepsis, or eclampsia (con- West 4.1 14 340 vulsions resulting from hypertensive disease during preg- Southeast 12.4 52 420 nancy)-though the route to these ends can vary. Most East 21.8 12 55 countries list five main causes of maternal mortality: hemor- Latin America 12.6 34 270 rhage, sepsis, eclampsia, abortion, and obstructed labor. Ob- Middle 3.7 9 240 structed labor and abortion, however, usually lead to death Caribbean 0.9 2 220 from sepsis or hemorrhage. Moreover, differences in reporting Tropical south 7.1 22 310 can obscure underlying events. For example, Reich suggests Temperate south 0.9 1 110 that about one-quarter of all pregnancies worldwide end in Oceania 0.2 2 100 induced abortions and result in up to 200,000 deaths (Reich Developing countries 110.1 494 450 1987); however, these deaths may be coded as hemorrhage or sepsis. In table 17-3 we summarize the leading causes of death Industrial countries 18.2 6 30Qnsvrlcutis in several countries. World 128.3 500 390 Death from hemorrhage can occur in less than an hour. Note: Estimates for 1980-85 from L.N demographic indicators ofcoCUntries. Women far from health services are exposed to the greatest Source: WHO 1985. danger because they may not reach a hospital in time for 366 Julia A. Walsh, Chris M. Feijer, Anthony R. Measham, and Paul . Gertler Table 17-3. Major Causes of Maternal Deaths, 1980-85 (percent) Obstructed labor/ruptured Region Ifemorrhage Sepsis Eclampsia Abortion uterus Other United States 10 8 17 6 3 56 CLba 6 19 12 15 48 Jamaica 23 9 30 10 3 25 Zambia (Lusaka) 17 15 20 17 .. 31 Egypt(Menoufia) 29 11 5 4 .. 51 Tanzania (four regions) 18 15 3 17 6 41 Ethiopia (Addis Ababa) 6 2 6 25 4 57 Bangladesh 22 3 19 31 9 16 Indonesia 46 10 5 7 .. 32 India 18 14 16 14 3 35 Negligible. Source: Calculated from Herz and Meashain 1987. transfusion or surgery. A study in China found, for example, The remaining one-quarter of maternal deaths include com- that 60 percent of the maternal deaths in rural areas were from plications of illness that existed prior to pregnancy, such as hemorrhage, in contrast to 25 percent in the urban areas hypertension, diabetes, and heart disease. Hepatitis, for exam- (Zhang and Ding 1988). ple, causes hemorrhage or liver failure in pregnant women and is a significant cause of matemal death in many countries (China [Chen 19851; Ethiopia [Kwast and Stevens 19871; India Figure 17-1. Major Causes and Consequences [Rao 1985; Bhatia 1985]; Nigeria [Ojo and Savage 1974];) and of Prolonged or Obstructed Labor in refugee camps in Somalia and Sudan (cDc 1987). Anemia impedes a woman's ability to resist infection or survive hem- Social factors Inadequate Early Multiparity Female orrhage and may increase the likelihood of her dying in child- childhood circumcision birth by a factor of four (Llewellyn-Jones 1965; Chi and others 1981). Additionally, latent infections, such as tuberculosis, Biological Stunted Early Scarring malaria, and sexually transmitted and other genital infections, consequences growth/ childbearing can become active during pregnancy and severely threaten the distorted or small pelvis health of mother and baby. The same complications which cause death can lead to Complications Cephalopelvic Weak Abnormal chronic disability when they are less severe. The complications disproportion uterine fetal of obstructed or prolonged labor, postabortal or puerperal contrac- presenta- tions tion sepsis, and hemorrhage, for example, include fistulae (tears in the vaginal wall), stenosis (narrowing of the vagina), and uterine scarring or prolapse (Howard 1987). Fistulae (rectal, Consequences Prolonged or urethral, and vaginal) can cause foul-smelling discharge and if good care obstructedl,advgna)cncus olsmligdscag n lacking labor social ostracism. Uterine scarring and prolapse can cause in- - Urinary fertility. In figure 17-1 the antecedents and consequences of incontinence obstructed labor are presented, clearly illustrating the inter- Ruptured .Vaginal uterus damage connectedness of reproductive complications. Other chronic disabilities include hypertension, chronic renal failure, and Hemorrhage Infection Vaginal Prolapsed urinary incontinence. Some of these illnesses can cripple a and/or uterus woman, both physically and socially. for the rest of her life. rectal fistula The incidence of pregnancy-related complications is poorly defined. The authors of the most-quoted study, that of a small Later Infertility Painful village in India, found sixteen illnesses for every death (Datta ectopic intercourse o .Im pregnancy and others 1973). If the icidence is similar elsewhere, then 3 to 12 percent of all pregnancies result in episodes of serious ill Divorce, Later health in women. Other studies, from Zaria, Nigeria, and from Death social obstructed China and Egypt, present widely differing results, ranging from disgrace labor I to 37 percent, depending on the population and survey methods (Li and others 1982; Mekhemar and others 1984; Source: Lettenmeier and others 1988. Harrison 1985). Matermal and Perinatal Health 367 Thus, complications of pregnancy and delivery cause an the quality of care during pregnancy, delivery, and the neona- enormous burden of illness. Most of this burden can be averted, tal period. It is a key determinant of health and well-being for however, by perinatal and obstetric care. Maternal mortality the rest of an individual's life. Compare the prospects of an has declined substantially with increasing use of hospitals for infant of low weight born after a prolonged, difficult delivery delivery. In Latin America and the Caribbean region, it has (with substantial risk of hypoxic brain damage, or limb weak- fallen by half since 1960 (Walker and others 1986). Despite ness from birth trauma, or pneumonia acquired from amniotic these declines, deaths from illegal abortion represent a larger fluid infection) with those of a normal infant. In developing percentage here than in other regions (Royston and Lopez countries very little is known about the incidence of these 1987). Trends in Asia and Sub-Saharan Africa are not known. perinatal disabilities and their consequences for the future About 25 percent of the deaths of women between the ages of well-being of both the child and family. twenty and thirty-four years result from matemity-related Seven million perinatal deaths occur annually-almost all causes. in developing countries (Lopez 1990). These perinatal deaths include stillbirths (also called late fetal deaths) and deaths in PERINATAL HEALTH PROBLEMS. At no other age is life so the first week of life. The perinatal mortality rate includes all tenuous and the risk of death so great as in the perinatal period. births (stillbirths and live births) in its denominator. In most Figure 17-2 illustrates the daily risk of death throughout in- developing countries, it ranges between 40 and 60 per 1,000. fancy and childhood, from the perinatal period, which extends High rates of 80 to 100 are found in the least industrialized and from delivery through the first week of life, until five years of most disadvantaged countries. In industrial countries, rates age. Perinatal health reflects both the healthofthe woman and range from 6 to 10 (Belsey and Royston 1987, quoting from WHO data banks). Perinatal deaths, particularly stillbirths de- livered at home, are frequently underreported. As infant mor- Figure 17-2. Daily Mortality Rate for Infants tality declines, postneonatal deaths (up to twenty-eight days and Children in Chaco Province, Argentina postpartum) decline more rapidly and perinatal deaths thus comprise a larger proportion of infant deaths. For example, the infant mortality rate in Mauritius declined from 70 in 1967 to (per 1,000 children) 29 in 1982; the perinatal mortality rate declined from 67 to 34; however, perinatal deaths as a percentage of all late fetal deaths and infant deaths increased from 61 to 70 percent during the same period. In general, in developing countries, the perinatal mortality rate is almost the same as the infant mortality rate. Neonatal deaths, shared by both the perinatal and infant 10 mortality statistics, approximate 50 to 60 percent of each of these statistics (Edouard 1985). Perinatal mortality is largely determined by delivery care and the maturity of the fetus, as reflected by birth weight and gestational age. Studies of the relative effects of birth weight 1 and gestational age on mortality suggest that birth weight is the predominant factor (McCormick 1985). Figure 17-3 demonstrates the dramatic rise in perinatal mortality among infants with a birth weight of less than 2,500 grams and more than 4,000 grams. These infants at the ends of the curve usually 0.1 had gestational ages of less than twenty-eight weeks or more than forty-two weeks. The main causes of perinatal mortality are infection of the amniotic fluid, congenital syphilis, abruptio placentae, fetal hypoxia of unknown cause, compression of the umbilical cord, 0.01 premature rupture of membranes, obstructed labor, birth trauma, and congenital malformations. Up to 30 percent of deaths are due to "other" causes (Naeye 1980; Lucas and others 1983; Oyedeji and others 1983; McCormack and others 1987). Virtually all these deaths occur within the first year of life, 0.001 _ primarily during the first few days of life. 1 1-6 7-27 1-12 13-24 25-26 Morbidity resulting from perinatal problems is difficult to day days days months months months qLuantify. The long-term consequences of low birth weight, one Age of the most severe perinatal health problems, are discussed below. Even less is known about the disabling effects of (a) lack Source: Puffer and Serrano 1973. of oxygen during labor and delivery that may cause cerebral 368 Julia A. Walsh, Chns M. Feifer, Anthony R. Measham, and Paul1. Gertler Figure 17-3. Neonatal and Postneonatal Deaths rupture of membranes (Lucas and others 1983). Many of these by Birth Weight same problems also cause matemal disease and death; im- proved obstetric care can prevent both matemal and perinatal Death rate morbidity and mortality. 1,000 Only a few researchers have examined the incidence of 800 morbidity among neonates, but they have demonstrated its 600 \ high frequency. In a Madras matemity hospital, with efficient 400 - obstetric and pediatric care, 20 percent of all babies suffered 300 - Neonatal monality some illness and 4 percent died neonatally. The most common 200 - Under 28 days conditions were low birth weight (below 2,000 grams in 7.5 (per 1,000 births) percent) and infections (in 8.3 percent). Obstetrical injuries 100 and asphyxia affected only a few (2 percent) as a result of the 80 quality of obstetric and pediatric care (Thirugnanasamband- 60 ham and others 1986). 40 \ \ Reproductive tract infections from both sexually transmit- 40 \ \ ted diseases and other genital infections are common and result 30 in a substantial amount of perinatal as well as matemal disease 20 (Kundsin and others 1988). In parts of Africa, more than 10 Postneonatal percent of pregnant women are seropositive for syphilis, but 10 mortality few are treated prior to delivery (Brunham and others 1984). In a study done in the 1940s prior to the availability of 6 28 days to 11 months (per1000 survivors penicillin, syphilitic mothers experienced over 10 percent 4 stillbirths and 20 percent infant deaths; more than 20 percent 3 ..- - of the infants who survived had evidence of congenital syphi- 2 _ lis, and only a third of the pregnancies resulted in a healthy child (Brunham and others 1984). Gonorrhea, chlamydia, and l l l l X ". mycoplasma cause stillbirths, preterm delivery, and intrauter- <1.000 1,001- 1,501- 2,001- 2,501- 3,001- 3,501- 4,001- ,4,501 ine growth retardation. Group B streptococci and bacterial 1.500 2,000 2,500 3,000 3,500 4,000 4,500 vaginosis are associated with neonatal sepsis and low birth Birth weight (in grams) weight (Brunham, Holmes, and Eschenbach 1984; Berman and others 1981; Investigators of the Johns Hopkins Study of Note: Deaths of single infants alive at birth, 1974-75, based on data from Cervicitis and Adverse Pregnancy Outcome 1989; Walsh and eight areas of the United States. Hutchinson 1989; Wasserheit 1989). Most of these infections Source: Shapiro and others 1980. can easily be identified and treated before and during preg- nancy to prevent these consequences. palsy, mental retardation, or learning disabilities; (b) pulmo- As infant mortality has declined and prenatal and obstetric nary dysfunction from scarring, respiratory infections, and care have improved, the perinatal mortality rate also has prematurity; (c) congenital syphilis causing multiple organ declined but to a smaller extent. Among all infant deaths, and brain damage; (d) birth trauma injuring limbs, nerves, and however, the proportion from perinatal deaths has increased. internal organs; and (e) amniotic fluid infections which re- As more women obtain skilled prenatal and delivery care, sult in pulmonary, ocular, brain, and other organ damage. perinatal morbidity and mortality should continue to decline. These acute and chronic disabilities have large social costs in health service use, family disruption, lost earnings, and long- LOW BIRTH WEIGHT. The World Health Organization defines term care. low birth weight as a birth weight less than 2,500 grams, Many infant deaths are potentially preventable by provision because below this value birth weight-specific infant mortality of prenatal and obstetric care to women. A clinico-pathologi- begins to rise rapidly. Two main processes determine birth cal study of 702 perinatal deaths in Nairobi revealed that weight: duration of gestation and intrauterine growth rate, and one-third were potentially preventable with better obstetric both affect fetal, neonatal, postneonatal, and child mortality. care (pregnant woman coming to the hospital earlier in the When gestation lasts less than thirty-seven weeks, the infant delivery process), because 38 percent of these were fresh still- is considered preterm. Intrauterine growth retardation (IUGR) births. More than 40 percent of deaths resulted from problems is also called "small-for-gestational-age" or "stnall-for-dates" avoidable by early cesarean section or assisted delivery: birth but has no standard definition. Commonly used definitions trauma, ruptured uterus, cord prolapse, obstructed and pro- include: birth weight less than the tenth (or fifth) percentile longed labor, placenta previa, placental abruption, and for gestational age; birth weight less than 2,500 grams and eclampsia. In addition to these deaths, nearly 10 percent of the gestational age greater than or equal to thirty-seven weeks; and neonates died of infection acquired intrapartum, potentially birth weight less than two standard deviations below the mean avoidable with early treatment and prompt delivery following value for gestational age.3 Maternal and Perinatal Health 369 Unfortunately, most studies from developing countries do Table 17-4. Mean Birth Weight and Prevalence not distinguish IU(:R from preterm infants (and even fewer of Low Birth Weight, Selected Countries disaggregate types of Iuc,R). Analysis of a small number of Mean birth Low birth weight studies suggests that more than half of the cases of low birth Country weight (g) (percent) weight in developing countries probably result from IUGR, whereas in industrial countries most of such cases result from AE2ica preterm delivery. In 1982, of the 127 million infants born, 16 Kgeynpta 3,243 127.8 percent (20 million) weighed less than 2,500 grams and more Nigeria 2,880-3,117 18.0 than 90 percent of these infants were born in developing Tunisia 3,210-3,376 7.3 countries (see table 17-4 for reported mean birth weight and Tanzania 2,900-3,151 14.4 prevalence of low birth weight in selected countries). Western Zaire 3,163 15.9 European countries have the lowest proportion (5 percent) Of Asia infants of low birth weight (Kramer 1987). China 3,215-3,285 6.0 The lowest infant mortality rate occurs among infants India 2,493-2,970 30.0 weighing 3,000 to 3,500 grams (see figure 17-3) but rises Indonesia 2,760-3,027 14.0 dramatically among those below this weight and only slightly Iran, Islamic Rep. of 3,012-3,250 14.0 above this. In industrial countries the prevalence of neu- Iraq 3,540 6.1 Japan 3,200-3,208 5.2 rodevelopmental handicaps among low-birth-we ight infants is Malaysia 3,027-3,065 10.6 three times that for those of normal birth weight. The risk for Pakistan 2,770 27.0 infants of very low birth weight (1,500 grams or less) is ten times that for normal weight infants. Eight to 19 percent of Brazil 3,170-3,298 9.0 infants of very low birth weight may be severely affected Chile 3,340 9.0 despite the availability of neonatal intensive care. Low birth Colombia 3,912-3.115 10.0 weight strongly predicts school failure, but the household Guatemala 3,050 17.9 setting modifies the effect (which is less pronounced in advan- Nlexico 3,019-3,025 11.7 taged households) (McCormick 1985). Comparable statistics North America are not available for developing countries. Canada 3,327 6.0 Infants of low birth weight have twice the risk of congenital United States 3,299 6.9 anomalies than those of normal weight, and infants of very low Europe birth weight have three times the risk. Congenital anomalies Czechoslovakia 3,327 6.2 and neurodevelopmental handicaps are not mutually exclu- France 3,240-3,335 5.6 sive. The proportion of infants affected with one or both ranges Hungary 3,144-3,162 11.8 from 19 percent in infants of normal birth weight to 42 percent Norway 3,500 3.8 in those of very low birth weight; the proportion of infants Sweden 3,490 4.0 severely affected is 2 to 14 percent (McCortnick 1985). Low- United Kingdom 3,310 7.0 birth-weight infants also suffer more pulmonary disease, use Source. Kramer 1987. health services extensively (for neonatal care and subsequent physician visits), increase family health costs, and disrupt the normal functioning of the family. developing countries prior to 1979 are fragmentary; few neo- Though the cause of the majority of low-weight births nates were weighed except in urban referral hospitals, and only occurring in both developing and industrial countries remains a small proportion of births was under surveillance. National unexplained, there are many important risk factors (see table averages cannot be accurately estimated from these figures, 17- 1). They are believed to include infant gender, race, ethnic and in any event, few of these statistics have been compiled. origin, socioeconomic status, maternal height, prepregnancy weight, paternal weight and height, matemal birth weight, Maternal and Perinatal Morbidity and Mortality in 2000 parity, history of prior low-birth-weight infants, gestational weight gain and caloric intake, general morbidity and episodic Future patterns of matemal and perinatal morbidity and mor- illness during pregnancy, reproductive tract infections (includ- tality probably depend more on the number and proportion of ing sexually transmitted diseases), malaria, cigarette smoking women of childbearing age in the population and fertility rates (and tobacco chewing to a lesser extent), alcohol consump- than on specific factors such as hospital access. The World tion, and in utero exposure to diethylstilbestrol. Health Organization made a series of projections assuming (a) From 1979 to 1982, the proportion of low-birth-weight no change in fertility and maternal mortality, (b) a reduction infants appeared to decline globally from 16.8 to 16 percent. in fertility of 25 percent, and (c) halving of the maternal In developing countries it was reported to have declined from mortality ratios. In the first case, 600,000 maternal deaths 18.4 to 17.6 percent, whereas in industrial nations it dropped would occur in the year 2000. If fertility were reduced by 25 from 7.4 to 6.9 percent. These are minimal declines and percent, the number of deaths would drop to 450,000. By probably represent only improved data collection. Data from halving the mortality ratio, deaths would decrease to 300,000. 370 Juila A. Walsh, Chris M. Fezfer, Anthony R. Measham, and Paul J. Gertler If fertility declined by 25 percent and the mortality ratio by on the health of the other. Death and sickness among these half, then maternal deaths would shrink to 225,000 in the year population groups are common, and the effects ofthis ill health 2000, fewer than half the deaths experienced in 1985. are pervasive and costly to society. The World Health Organization estimates continued de- clines in rates of low birth weight without citing the reasons Risk Reduction prior to Conception for the expected decrease. Reduced fertility would result in fewer perinatal deaths and the decline in rate of low birth Reduction of risk prior to pregnancy is the most feasible way weight would reduce the perinatal mortality rate. By 2015 the to improve maternal health. perinatal mortality rate should decline from the 1985 level of 98 for males and 75 for females to 42 and 31 in developing Elements of Prevention Strategies countries (Lopez, chapter 2, this collection). This decline depends on an increase in skilled delivery services along with In order completely to eradicate reproductive risk, pregnancy the decline in low birth weight. would have to be eliminated. Because families and society obviously value and desire children, however, the goal should Indirect Economic Costs be to avoid unwanted pregnancies and to lower the risks when pregnancy occurs. Many steps can be taken to do this even There is a dearth of information both on treatments used for before conception. maternal and infant morbidity occurringduring pregnancy and delivery and on the effects this morbidity has on the women BETTER GENERAL HEALTH FOR WOMEN. In general, when and infants, theirfamilies, thecommunity, and healthservices. women are healthier they are better equipped to handle preg- Though chronic morbidity may be more expensive than a nancy and their infants will be healthier. Good health for all quick death, mortality is complicated in that the death of the women requires an integrated set of actions, including health mother is usually accompanied by the death of the infant from services, community development, and education for female that pregnancy. Ninety-five percent of the infants born to children. mothers who died in childbirth also died, according to a small Women need betterservices throughout their life. Nutrition study in Bangladesh (Chen and others 1974). A disabled is important at all ages so that a woman enters childbearing motherconfinedtobedforsevereanemiaorrenalfailuremight age with normal height, weight, pelvic size, and nutritional drain the family resources and compromise the health of family status. Nutrition programs can include a variety of compo- members, particularly the very young children, but there would nents, for example, nutrition and health education, anemia also be long-range effects on the well-being of the entire family screening and prevention (including iron and folate sLIpple- from the mother's death. ments), iodine supplementation, food supplementation, pro- Although their efforts are often undervalued, women sup- motion of community and household gardens, and income port families through their productive labor: cash crop labor, supplements (Lettenmeier and others 1988). The choice of subsistence farming, and other remunerative work. Poorer intervention depends on the most important nutritional women are more likely to be solely responsible for their fami- deficiencies experienced in an area and the availability of lies. Many regions experience high male outmigration for jobs, resources. in effect leaving the women alone as the sole parent. The loss It is also important to pay attention to those conditions of the mother through death or disability then means the loss affecting girls early in life that make pregnancy more risky. of the nurturer, provider, and de facto household head. Rheumatic fever, for example, causes heart disease which can Treatment of low-birth-weight infants and neonatal illness complicate childbirth. In Menoufia, Egypt, 16 percent of ma- can also drain the resources of the family and health service, temal deaths were due to cardiovascular disease; one-half of particularly when neonatal intensive care is required. Indeed, them involved a history of rheumatic fever (Fortney and others when health resources are limited, health planners should 1988). Infections of the reproductive tract and sexually trans- weigh the cost of neonatal intensive care, which may benefit mitted diseases (STDs) also must be diagnosed and treated. only a few infants, against the cost of improved preventive care These can cause scarring of the fallopian tubes that may result for matemal and perinatal health, which would confer wider in infertility or ectopic pregnancy (an important cause of benefits. Even where neonatal intensive care units are absent, matemal death from hemorrhage) when conception occurs. the increased health needs of sick and disabled infants are These infections are also responsible for much morbidity as costly and time consuming. discussed earlier. Women are often uninformed about the symptoms, risks, and means to prevent STD infection. Summary The delay of reproduction until a woman is fully grown is very important and may be achieved by raising age at marriage, Maternal and perinatal health (including low birth weight) are reducing the need for women to prove their fertility early, and closely linked, and efforts to improve the health of either giving more attention to male and female responsibilities in pregnant women or the newbom will have synergistic effects avoiding adolescent pregnancies. Economic and social devel- Maternal and Pennatal Health 371 opment also makes it easier for people to obtain the resources low, family planning is likely to have more effect on mater- required for good health and to maintain hygienic conditions nal deaths, especially if deaths due to abortion are high (Fort- to safeguard it. Careful attention to development policy ney 1987). Family planning is somewhat less effective choices can increase the probability that women will have and more costly, however, in societies in which women want lighter workloads and better food, both of which are extremely large families. important in reducing the wear and tear on health, which is As the total fertility rate declines, the proportion of grand most felt during pregnancy and childbirth. multiparas declines, but the proportion of primiparas increases. The status ot women in society directly affects their health These two trends have opposite effects on maternal mortality in many ways. For example, in some societies women are the in industrial and developing countries. In industrial countries, last to get food in the family and the least likely to use health a slight increase in maternal mortality results; however, in services. Education can be an important means to improve developing countries, the improvement in maternal health status. Most important, increasing the education of women from the decline in grand multiparas outweighs the negative indirectly increases the motivation and the means to attain effect of the increase in primiparas. improved health status and instills in each woman an aware- Wider birth spacing has been recommended for improving ness of her own health needs and the methods by which she maternal morbidity and mortality. Too many births too closely may handle them. Educated women tend to marry later and spaced seem to cause a general decline in the health and are more likely to use family planning, prenatal, and obstetric nutritional status of some women; this decline is called the services and to avoid dangerous traditional health practices maternal depletion syndrome. Whether this syndrome actu- (Harrison 1980; Kwast and others 1984; London and others ally exists has been questioned (Winikoff and Castle 1987). 1985; and Monteith and others 1987). Maternal education Nonetheless, the freedom to plan the timing of offspring, substantially affects infant mortality rates (figure 17-4); one and the education required to help a woman make this might expect similar results for maternal health, but this is not choice, are undeniably appropriate goals for any matemal well documented. health program. FAMILY PLANNING. Family planning acts on maternal health through several mechanisms. First and most important, with Figure 17-4. Infant Mortality by Mother's fewer unwanted pregnancies, fewer women resort to illicit Education, Selected Countries abortion. Second, a birth interval of more than two years has Infant mortality rate no proven effect on maternal morbidity and mortality, but it (per 1,000 live births) does improve the health of the infant. Still, longer birtlh 150 intervals usually result in fewer total pregnancies for a woman and, therefore, fewer grand multiparas (more than five preg- nancies). Third, targeted family planning can play a significant 125 - role by reducing the number of pregnancies in women most at risk of pregnancy complications, especially those under eight- een or over thirty-four years of age. Family planning includes 100 _ education and access to a variety of contraceptive methods such as pills, condoms, intrauterine devices, and so on, plus availability of safe abortion. Because approximately 25 percent 75 - of maternal deaths result from unsafe abortion, the provision of safe abortion services to back Up other contraceptive ser- vices is a highly effective and low-cost way to reduce maternal 50 - mortality (Blacker 1987). The effect of family planning on maternal mortality depends on a country's stage in the demographic transition (Fortney 25 _ 1987). If the number of unwanted pregnancies and the need for abortion are reduced, potential declines in maternal mor- tality range from 5 percent in C6te d'lvoire to 62 percent in 0 L Bangladesh (Lettenmeier and others 1988). A general rule of Egypt Kenya Peru Trindad Sri Lanka Indonesia thumb is that in countries in which fertility is already low, Tobago increased use of contraception will have relatively little effect Mother's education on fertility and maternal mortality. These countries should (years) focus maternal and perinatal health care strategies on obstetric 1_3 _14-6 M_7+ improvements. It is also useful to look at both the maternal mortality ratio and rate; where the rate is high and the ratio is Source: Starrs 1987. 372 Julia A. XValsh, Chrs M. Feifer, Anthoir R. Measham, and Pau J. Gertler Targeted family planning can help avoid pregnancies matters, nutritional deficiencies are passed from generation to among women who are too young (less than eighteen), too old generation. The poorly nourished woman is more likely to give (over thirty-four), have already had many births (more than birth to an infant of low birth weight who in tum may never four), have an obstetric or medical history which places them reach fully normal stature. When this infant becomes an adult, at higher risk, or who do not want another pregnancy she may give birth to other low-birth-weight children or have (Lettenmeier and others 1988). Locating such women and obstructed labor from small pelvic size. Contrary to expecta- providing them with contraception will substantially reduce tions, there has not yet been any evidence of change in the matemal mortality rates. Estimates for this reduction range incidence of low birth weight in developing countries as a from 25 to 50 percent of all maternal deaths (Maine 1981; result of the large-scale food supplementation programs for Rinehart and others 1984; Blacker 1987; Maine and others girls and women prior to conception. 1987; Winikoff and Sullivan 1987). In table 17-6 we summarize possible actions of a maternal and child health program before conception occurs. The pri- Good Practice vs. Actual Practice-Where Are the Gaps? orities for services include family planning, including contra- ception and safe abortion; integration of family planning into Many developing countries have made remarkable progress in health services which can screen for and treat anemia, infec- reducing maternal mortality. Their experiences provide clues rions of the reproductive tract, sexually transmitted diseases, to the most effective actions to take. Lettenmeier and others malaria, and other illnesses common in the particular locale; (1988) report that "Sri Lanka, through nearly universal edu- and health education through health centers and other ave- cation, raised the average age at marriage and increased the nues for the prevalent illnesses and for health risks such as poor use of family planning, thereby reducing the number of risky nutrition and tobacco and alcohol use. Many of these screen- pregnancies in adolescents and in older women with many ing, treatment, and education services obviously are needed children (Henry and Piotrow 1979). At the same time health during pregnancy as well. Wherever possible, all these priority care improved. The Sri Lankan maternal mortality ratio services should be integrated into pregnancy care. Another dropped from 555/100,000 in 1950to95 in 1980 (Royston and alternative is to provide family planning and treatment for Lopez 1987)." Lettenmeier and others (1988) also report that reproductive tract infections and STDs together and pregnancy "China lowered maternal mortality by substantially lowering care separately. birth rates, raising the age at marriage, and improving health care for pregnant women (Chen and Kols 1982). The maternal Pregnancy, Delivery, and the Neonatal Period mortality ratio, at 25, now rivals that of industrial countries (WHO 1986)." These two success stories demonstrate the The strategy for maximizing good health during pregnancy potential of multifaceted efforts to reduce maternal mortality requires access to health care. and morbidity. Despite recent gains in a few countries or among women of Case Management of Pregnancy and Complications certain classes, most women in developing countries face social constraints in status and decisionmaking power which limit The ability of the health system to decrease the effect of their ability to safeguard their own health (pcC 1988). Women pregnancy complications depends upon the capacity to iden- in developing countries still lag behind men in educational tify high-risk women, prevent complications or treat them, and attainment and literacy, as is evident in table 17-5, in which refer for skilled obstetric care. When a complication presents are also shown the regions of the world where women marry at itself in spite of prenatal care, back-up obstetrical referral young ages and have large families. Africa scores lowest for facilities are required. It is difficult to identify all women who educating its women and age of marriage, and it has the highest will have complications; for example, in a group of women with fertility; maternal mortality is also highest in Africa. The no risk factors receiving prenatal care in Canada, 10 percent Indian subcontinent (Middle South Asia) scores the next of the women had obstetric complications and 19 percent of lowest and has matemal mortality rates nearly as high as Africa. the neonates had complications (Moutquin and others 1987). Poor nutrition is still a widespread problem among women A matemity care program in rural Bangladesh which posted despite recent efforts to provide food supplements and nutri- trained midwives at village health posts and referred compli- tion education. A survey of eighty developing countries in the cated deliveries to a central maternity clinic reduced matemal early 1980s showed that between 20 and 45 percent of the mortality by 60 percent (Fauveau and others 1991). Even women age fifteen through forty-four consumed insufficient though the maternal mortality ratio remained high in the calories daily (Hamilton and others 1984). Anemia incidence program area (1.4 per 1,000), the reduction suggests that is even higher; up to 80 percent of the women of reproductive midwives can improve matemal survival when given proper age in developing countries may be anemic (Royston 1982). facilities, supervision, and referral facilities. The program was Food supplementation programs have proved expensive and undertaken in an area with an already existing high-quality difficult to manage logistically. Furthermore, they have not primary health care system, probably a key element in its generated data on their contribution to the reduction of ma- effectiveness. The effect on perinatal mortality and costs must ternal mortality or morbidity (Weston 1986). To compound still be determined. Matemal and Penrnatal Health 373 Table 17-5. Selected Sociologic Indicators, by Region Literate adults Marned women Region Male (percent) Female (percent) age 15-19 (percent) Total fertility ratea Africa 33 15 44 6.2 North - 18 34 5.1 West 20 6 70 6.9 East 29 14 32 6.9 Central 35 9 49 6.2 Southem 55 56 2 4.7 Asia 56 34 42 3.5 Southwest 58 31 25 5.1 Middle south 44 17 54 4.7 Southeast 75 53 24 3.7 East 97 92 2 2.4 Latin America 76 70 16 3.6 Central 75 67 21 3.9 Caribbean 67 66 19 3.0 Tropical south 74 67 15 Temperate south 93 91 10 3.5 Oceania 90 88 10 2.5 North America 99 99 11 1.8 Europe 96 93 7 1.7 Former U.S.S.R 100 100 10 2.4 World 67 54 30 3.5 Industrial countries 98 97 8 1.9 Developing countries 52 32 39 3.9 a. Number of children per woman. Source: Starts 1987; United Nations various years. The generally accepted strategy for dealing with the com- in pregnancy and providing them with special care: for exam- plications of pregnancy and childbirth involves a regional ple, encouraging an appropriate diet, treating infections and network of community risk assessment through prenatal care other illnesses, and arranging skilled delivery for those requir- and use of facilities at the first-referral level, usually a district ing it. A study in Indonesia found that women who received hospital, for the management of high-risk cases and treatment no prenatal care were more than five times as likely to die than of obstetrical emergencies. Due to the nature of transportation those who attended a prenatal clinic (Chi and others 1981). in the developing world, maternal health programs must ad- It is estimated that in the United States each dollar invested dress the need for emergency transfer. In some cases, maternity in equitable access to comprehensive prenatal care results in a waiting homes for those with expected complications are pro- savings of $3.38 in subsequent expenditure for the care of posed; in others, innovative methods forproviding transporta- low-birth-weight infants (loM 1985). Low birth weight and tion are involved. Herz and Measham (1987) advocate perinatal problems are less common among women who have stronger community-based prenatal, delivery, and family plan- prenatal care (Chi and others 1981; Donaldson and Billy, ning services (incorporating the preventive strategies de- 1984; Brown 1985; Harrison 1985; IOM 1985; Trivedi and scribed earlier), stronger referral facilities, and an alarm and Mavalankar 1986; Winikoff 1988). Nevertheless, it has not transport system. been possible to identify the specific components of prenatal care which reduce these problems (IoM 1985; Winikoff 1988). PRENATAL CARE. Prenatal care is cost-effective because it Women who avail themselves of prenatal care tend to be better reduces the number of women requiring skilled obstetrical care educated and in better general health beforehand, so any by screening and treating women at risk of complications and comparison of women who take advantage of prenatal care referring them to other facilities if necessary. Prenatal care can with those who do not will show that the former group is better also substantially reduce the proportion of low-birth-weight off. Where general health care is not available to prevent or infants and the incidence of perinatal disease. These reduc- treat risk factors before conception, prenatal care becomes tions depend, however, on identifying high-risk women early even more important. 374 Julia A. Walsh, Chnis M. Feifer, Anthony R. Measham, and Paul J Gertler Table 17-6. Preconception Semices to Improve Maternal and Perinatal Health Intervention Effectiveness Major drawbacks Programfeasibility Nutrition Food supplementation Benefit to mother has not been Difficult to ensure that pregnant High cost. Generally too evaluated. Supplementation woman does not share the expensive for large continuing increases birth weight 30 to 200 g. supplement with others or programs. Works best with Most effective in malnourished substitute it for usual diet. locally available, inexpensive women. foods and when combined with nutrition education. Having women eat supplement at local distribution site most effective. Iron/folate supplementation Highly effective if tablets taken Gastrointestinal side effects and Low cost. Easy to add to existing before and during pregnancy regularly. Malaria and parasite need to take tablets several times antenatal care or other health treatment needed in endemic a day make compliance a care that reaches women. No areas. problem. Hard to motivate special storage facilities needed. asympromatic women to take tablets. Iodized oil injections One injection prevents iodine To benefit infant, injections Low cost. Easy to add to existing deficiency for three to five years, must be given before pregnancy. antenatal care. shrinks existing goiter. Family Planning Change maternal age All major methods highly Long-term consistent use and Moderate cost. Variety of Allow for birth spacing effective if used regularly and continuity of supplies needed for methods allows delivery through Limit family size correctly. some methods. Rumors and many medical and nonmedical Avoid high-risk pregnancy concems about potential side delivery systems. Users have Avoid unwaanted pregnancies effects may deter use. Specialized choice of methods. Some training needed to perform methods can be distributed by sterilizations and insert IUDs and briefly trained workers. Elaborate implants. Cultural restrictions storage system not needed. against use of family planning in some areas. Health services Primary health care for STDs, Highly effective depending on Services depend on local Moderate cost systems often in reproductive tract infections, content and use. Prevents tubal resources and endemic illnesses. place due to Alma Ata hypertension scarring, which can lead to Specialized training, diagnostic conference and "Health for All" infertility and ectopic pregnancy. tests, and treatments required programs. locally. Health education Reproductive tract infections Effect on matemal and perinatal Community participation in Moderate cost systems often in and STDs health depends on content and education and adequate training place due to Alma Ata Poor health habits (tobacco, use. of health educators. conference and "Health for All" alcohol) programs. Nutrition Signs of premature labor Source: Lettenmeier and others 1988. To some degree, the content of prenatal care must be levels of the health care system at which this care can occur. adjusted according to local technologies, economics, and pop- At the community level, health workers can educate pregnant ulation needs. The prevailing health problems must be identi- women about good nutrition, dangerous habits (tobacco and fied and care must be targeted to those at risk so that they are alcohol use), STDS, and about signs of premature labor and other identified and treated. Prenatal programs in poor countries complications so that women can get to the referral centers cannot hope to be as comprehensive as those recommended in quickly (lam 1989). At the primary level, health workers industrial countries; managers of such programs must select should be trained and resources available for health and nutri- carefully the components they will include for maximum effi- tion education and for screening and monitoring for a variety ciency and benefit. In table 17-7 we list the important compo- of conditions. The choice of diseases to screen for depends nents of prenatal care for developing countries and the lowest upon the local disease burden. For example, in parts of Africa Maternal and Perinatal Health 375 Table 17- 7. Prenatal Interventions for Preventing Maternal Morbidity and Mortality, Low Birth Weight, and Perinatal Mortality in Developing Countries Level of care Activity Areas covered Community Educate Nutrition; tobacco and alcohol use; signs of premature labor and other serious complications; self-referral for care Primary Monitor and treat or refer for skilled care Uterine growth; weight gain; bleeding; presentation; hypertension; edema Primary Screen and treat or refer for skilled care Reproductive tract infections; sexually transmitted diseases; diabetes; urinary tract infections; cardiac disease Primary Treat intercurrent illness Diarrhea, respiratory infections, malaria, and other diseases Primary Provide preventive and nutritional care Malaria prophylaxis in endemic areas; tetanus immunization; iron and folate supplements for anemia; nutritional supplements for malnourished women Referral Detect and treat Premature labor; rupture of membranes Referral Skilled delivery Small pelvic size; poor obstetric history; open cervix; other risk factors (such as age, parity) Referral Treat Complications of spontaneous and induced abortions; ectopic pregnancies; hemorrhage Source: Authors. where STDs and malaria are prevalent, these problems must be (for example, oral ergometrine for hemorrhage), or have nurse addressed. Inexpensive diagnostic tests are available for many midwives do surgical procedures (for example, cesareans) diseases: syphilis, gonorrhea, malaria, and anemia, among oth- when emergency transportation is a practical impossibility or ers. Unfortunately, for sexually transmitted diseases, even likely to be slow enough that local emergency treatment is a these diagnostic tests may be too expensive for widespread use. necessity. More operational research on diagnosis and treatment of STDs The World Health Organization recommends that referral is urgently needed. All prenatal clinics should have the capac- centers provide the following eight essential obstetrical ser- ity to screen for high-risk pregnancy, perform pelvic examina- vices (Starrs 1987; Lettenmeier and others 1988): tions, immunize against tetanus, monitor uterine growth, * Surgical procedures: cesarean sections, draining provide iron and folate, and provide malaria prophylaxis. De- abscesses, repairing high vaginal and cervical tears, remov- pending on resources, other interventions, such as more com- prhnsv helt education, shul be poie. Trie , ng ectopic pregnancies, emptying the uterus following in- prehensive health education, should be provided. Trained complete abortion, symphysiotomies, and, in case of severe birth attendants, not just family members or untrained tradi- complications, rupturing the amniotic membrane to induce tional midwives, should assist all deliveries so that incipient or quicken labor. complications can be treated or the woman rapidly transported to a referral center. * Anesthesia: general, spinal, and local. Most countries have an inadequate number of facilities and * Medical treatment: providing fluids and medications trained personnel for prenatal care, normal deliveries, trans- intravenously, treating shock, infection, preeclampsia and port to referral centers, and care of complications. The cost eclamptic seizures, and inducing labor. and effectiveness of investments in training and facilities ex- * Blood replacement: transfusions of blood or other fluids pands when they can be used for other priority programs such for hemorrhage or surgery. as STD screening and treatment, family planning, and maternal * Manual procedures for diagnosis and treatment: remov- and child health care. ing placentas manually, delivering by vacuum extraction or forceps, and using partographs to monitor labor. REGIONALIZED CARE AND REFERRALS. Careful consideration of * Family planning: tubal ligations, vasectomies, implant- which health care personnel will deal with what level of ing and removing intrauterine devices and Norplant, other complications is critical. Table 17-8 is an outline of a proposed family planning methods. system for regionalized care and referral for prevention of matenalmoralit. Asimlarsystm cn b use topreent * Special care for newborn babies: resuscitation, keeping maternal mortality. A similar system can be used to prevent wam rvniglwbodsua,tetn netos low birth weight and perinatal mortality by adding screening for a small number of other fetal risk factors (for example, * Managing high-risk women: providing waiting homes tobacco and alcohol use, need for malaria prophylaxis, repro- close by. (This depends on prenatal screening for success.) ductive tract infections, and sexually transmitted diseases). Better care during pregnancy, labor, and delivery potentially Health care workers should be adequately trained to assess and could reduce maternal mortality rates anywhere from 50 to 80 handle any complication they may encounter. If referral is percent (Kwast and others 1984; Walker and others 1986; necessary, they must know how to stabilize and find transpor- Lettenmeier and others 1988) and perinatal mortality rates by tation for the patient. Those in charge of programs may decide 30 to 40 percent (Lucas and others 1983; Kaumitz and others to allow traditional birth attendants (TBAs) to dispense drugs 1984). Most of the improvement would result from additional 376 Julia A. Walsh, Chris M. Feifer, Anthony R. Measham, and PautJ. Gertler Table 17-8. Selected Interventions at Primary and First,Referral Levels for Prevention of Maternal Morbidity and Mortality Problem Intervention Health system level' All causes of matemal Family planning Primary, first referral mortality and morbidity Prenatal care Primary Supervised delivery Primary Hemorrhage Risk screening, referral Primary Other prenatal care, including treatment of anemia Primary Oxytocics when placenta delivered Primary Intravenous fluids Primary Transport to first-referral level Primary Manual removal of placenta Primary Blood typing of donors First referral Blood transfusion First referral Infection Risk screening, referral Primary Tetanus immunization Primary Clean delivery Primary Antibiotics when membranes riptured, if not Primary delivered within twelve hours Transport to first referral level Primary Hysterectomy First referral Toxemia Monitor symptoms, blood pressure, and urine Primary for protein Bed rest, sedatives Primary Transport to first-referral level Primary Induction or cesarean section First referral Complications of abortion Antibiotics Primary Transport Primary Oxyrocics Primary Evacuation First referral Hysterectomy First referral Obstructed labor and ruptured Risk screening, referral Primary uterus Partograph Primary Transport to first-referral level Primary SymphysiotOmy First referral Cesarean section First referral a. Primary level includes outreach programs and health dispensaries, posts, or centers. First-referral level usually is a district or cottage hospital with twentv or more beds and the capability of giving blood transfusions and performing cesarean sections. b. Recommended experimental approach ar the commtunity level. Source: Herz and Measham 1987. training for those who attend most births and by organizing about 10 per 1,000 births, the ratio of fresh-to-macerated emergency systems for unexpected complications. stillbirths is one to five. In contrast, in developing countries Recently, experts have suggested the use of three indicators like Kenya, the ratio isone to three (Belsey and Royston 1987). for assessing the effectiveness of prenatal and delivery care: the Other countries, such as Mexico, suffer from excess cesarean ratio of fresh stillbirths to macerated stillbirths, the percentage sections with poor quality perinatal care and have high peri- of cesarean sections, and the ratio of scheduled deliveries natal mortality (Bobodilla 1988). (women identified as high risk during pregnancy and hospital The quality of postpartum and neonatal care is another delivery arranged) to emergency deliveries (Mark Belsey, per- concern. Women should be treated for infections and checked sonal communication, June 1991). The ratio of fresh stillbirths for hemorrhage, the health needs of the infant should be may indicate potentially avoidable deaths; a low rate of ced-.ar- monitored, and counseling should begin to address future ean sections may suggest that additional maternal and perina- health needs to guarantee less risky deliveries the next time. tal disease and death may be preventable; and a low proportion Counseling should cover family planning, nutrition, need for of scheduled deliveries suggests that prenatal care and referral rest, child care, and hreastfeeding. This attention during post- should be improved to identify better the women at risk and partum visits of women to hospitals or health centers is espe- arrange for hospital deliveries prior to onset of complications. cially important in those areas where women use clinic services In European countries where the perinatal mortality ratio is only for maternal health needs. In table 17-9 the feasibility of Maternal and Perinatat Health 377 the proposed services for a maternal and child health program Figure 17-5. Births Attended by Trained is assessed. Neonatal intensive care is extremely expensive and Personnel, by Region therefore limited to highly specialized referral centers which can provide high-quality care. At the local level, birth atten- Births attended (percent) dants should know about clearing respiratory passages, early breastfeeding, sterile cutting and care of the umbilicus, and 100 hydration of the infant when needed. TRAINING OF TRADITIONAL BIRTH ATTENDANTS. About 70 percent of all babies in developing countries are delivered by 80 _ traditional birth attendants (TBAs) or relatives (Lettenmeier and others 1988). Many women prefer TBAs even when modern health care facilities are available. Health workers need to work with TRAS to improve the quality of maternal care deliv- ered and to increase the number of women obtaining prenatal 60 care. With appropriate training and supportive supervision, TBAs can provide the basic care required by women who have normal deliveries.They must learn to use hygienic techniques, avoid harmful practices (for example, excessive force on the 40 - woman's belly or pulling the umbilical cord to withdraw the placenta), and make the woman comfortable while encourag- ing her to adopt practices such as breastfeeding immediately after delivery, keeping clean, and watching for signs of 20 - postpuerperal infection or hemorrhage. Traditional birth attendants also need instruction in emer- gency care. The attendants must recognize warning signs, know how to stabilize the woman's condition, and be able to _ transfer her to the next level of care. The proportion of births, 0 by region, attended by trained personnel is shown in figure Africa Asia Caribbean South North NoEuhern 17-5. Africa uses trained personnel the least and has the highest matemal mortality rates. Note: Reflects experience about 1982. Source: Starrs 1987. Good Practice vs. Actual Practice-Where Are the Gaps? Existing reproductive care for women falls far short of needs control over to insensitive clinic staff. Some women may and lags far behind what we know is feasible (Ratnam and consider hospitals or clinics places to die and so refuse to use Prasad 1984). Access to high-quality family planning and them for preventive or routine health care; they resort to abortion services remains limited. Few women in developing prenatal care only if they develop complications (Lettenmeier countries receive prenatal care. Worse still, the women who and others 1988; see also Bamisaiye, Ransome-Kuti, and receive late or no prenatal care are most often young, primi- Famurewa 1986; Leslie and Gupta 1989). parous, poor, of a racial or ethnic mninority, and undernour- Systems to identify high-risk pregnant women in industrial ished. and they are more likely to smoke or drink (Kramer countries have had poor sensitivity and specificity (tom 1985). 1987), all of which puts them at higher risk than average and Even though most of the adverse pregnancy outcomes occur more in need of care. Of the women who do receive prenatal in women who have several of the risk factors mentioned in care, most make only one visit to a health care center and that table 17-1, many maternal and fetal diseases occur in women one relatively late in pregnancy (Williams and others 1985). with no easily identifiable risk factor. Because of this phenom- Rural women are much less likely to receive prenatal care than enon, the sensitivity and specificity of systems based upon urban women (table 17-1 0). In places where walking distance history and physical examination average about 60 percent is short, more use has been made of clinics. The best results (IOM 1985). These systems identify only about half of the have been where women are within three kilometers of the women who eventually have low-birth-weight infants. A rea- clinic (Williams and others 1985). sonablv reliable system for classifying women who will have In many places, clinics are available but women do not poor pregnancy outcome is needed to contain the cost of attend for various reasons. Because the regional referral net- services so that those at high risk can receive special, and more works rely on contact during pregnancy, low prenatal partici- expensive, care. An excess of women incorrectly identified as pation rates undermine the entire effort. Cultural barriers to high risk means that scarce resources are used for the care of services range from distrust of male clinicians to fears oftturning those who will not benefit from the service and may experience 378 Julia A. Walsh, Chris M. Feifer, Anthony R. Measham, and Paul J Gertler Table 17-9. Maternal and Child Health Services for Management of Pregnancy Intervtention Effectiveness Major drawbacks Programfeasibility Antenatal Antenatal care; screening and Women who receive antenatal Appropriate screening criteria High cost. Feasible if there are referral; monitoring of height, care are three to five times less difficult to select. Village health central referral hospitals. Special weight, blood pressure; checking likely to die and also less likely to care providers may have no training may be needed. for anemia, swelling, protein, lose their child than women who formal training. Existing services Community must be educated presentation, and bleeding. do not get care. Low birth weight underutilized. and involved so women will and perinatal complications also come for antenatal care. decline. Antenatal education: waming Difficulr to assess. Studies of Scheduling education into busy Low cost. Can be done as home signs, nutrition, health habits nutrition education found success clinic hours; getting women to visits, through neighborhood in avoiding toxemia. attend. organizations, TBAs. Antenatal treatment of anemia, Better to prevent these Cost of diagnosis, medication, Women already demand these hypertension, infection, STDs, conditions or detect them early. and training; getting services to services; they can be used as concurrent disease, Without this level of service, appropriate levels. point of entry for general complications of spontaneous women will die. prenatal care. Some services can abortion, ectopic pregnancies be done at primary level, others at referral level. Antimalaria education and Highly effective if taken Weekly or monthly regimen for Low cost. Easy to add to existing prophylaxis to endemic areas regularly. Reduces low birth taking tablets throughout antenatal care or other health weight, anemia. pregnancy makes compliance a care that reaches women. No problem. Only chloroquine is special storage facilities needed. approved for pregnant women; Women more likely to take best drug for areas with medicine if assured it will not chloroquine-resistant malaria is harm the baby or cause abortion. controversial. Tetanus immunization Eliminates most deaths from Pregnant women need two Low cost. Coverage generally lags neonatal tetanus. Woman injections at least four weeks behind other childhood immunized as well. apart, the second, at least two immunizations. Wider coverage weeks before delivery. Toxoid possible if women could be must be kept cold. Some immunized at all health centers. personnel may need special training to give immunizations. Food supplementation Benefit to mother has not been Difficult to ensure that pregnant High cost. Generally too evaluated; benefits of woman does not share the expensive for large, continuing supplementing women during supplement with others or programs. Works best when pregnancy only have been substitute it for usual diet. locally available, inexpensive evaluated. Increases birth weight foods used and when combined 30 to 300 g. Most effective in with nutrition education. Having malnourished women. women eat supplement at local distribution site most effective. Iron/folate supplementation Highly effective if tablets taken Gastrointestinal side effects and Low cost. Easy to add to existing before and during pregnancy regularly. need to take tablets several times antenatal care or other health a day make compliance a care that reaches women. No problem. Hard to motivate special storage facilities needed. asyinptomatic women to take tablets. Refetral Mothers/ waiting homes Reduced hospital matemal Women may have to stay several Low cost. Done in at least ten mortality rates in Nigeria and weeks before delivery. Lack of countries. Requires good Cuba; other programs not transport to homes. Women coordination between evaluated. more likely to come if they can community programs and bring their children. Need to hospital. Food, bedding, and maintain or possibly build homes. cooking utensils supplied by women's families. Matemal and Perinatal Health 379 Intervention Effectiveness Major drawbacks Program feasibility Transportation to hospital for Many deaths from complications Need to buy and maintain High cost. In many places obstetric emergencies may be prevented by getting vehicles. Needs constant supply adequate roads will not be built women to hospitals quickly. of parts, gasoline, drivers. Road fo years. Innovative building not planned to meet communicarion and transport maternal health needs. Woman systems possible where roads are may need husband's permission not available. to leave home. Deiivery Regionalization; coordinate TBA, Cost effective Getting appropriate level of care, Cost depends on existing level of local health center, and especially when unforeseen infrastructure. Can be added to specialized facilitiesb complications. any system. Neonatal care Detects and treats problems None High cost, depending on technology level. Postpartum care Most effective way to detect and Criteria for referral must be Low cost. Easy to add to existing treat postpartum infections and selected. Traditional practices antenatal care. Good opportunity prevent secondary infertility. may keep women at home for to offer family planning. Best several weeks after birth. Existing when home visits are a real services underutilized. possibility. Training Training TBAs to provide TBA performance improves or Trainers need special skills and Moderate cost. Done in many antenatal care, screening, safe remains the same. Effect on preparation. Supervision and countries. Most effective if delivery matemal complication rates has refresher courses needed for TBAs. community and formally trained not been evaluated. Delivery kits underutilized. health care providers are involved. Training doctors, nurses, and Matemal health outcomes have Supervision and refresher courses Moderate cost. In some countries others not been evaluated. needed. Degree of training nurses and nurse-midwives have needed that prepares personnel been trained to perform obstetric for rural posts. In-service training surgery. needs to be offered widely. a. There is insufficienr evidence ro rate individual components of antenaral care. b. Herz and Measham plan. Source: Lettenmeier and others 1988 complications from unnecessary diagnostic and therapeutic perinatal health system, based on risk assessment and referral procedures. A misclassification as low risk, however, denies care, improved screening criteria are needed. at-risk women access to beneficial services. Some of the sys- Delivery and postpartum care suffer from lack of use for tems evaluated classified more than half the women as high reasons similar to those causing women to ignore prenatal care risk when the incidence of poor pregnancy outcome was less services. Delivery costs are a special barrier to formal services, than 10 percent. These systems for classifying obstetrical risk whicharealsooftencompetingagainstculturalnormsfavoring were evaluated in populations where the average percentage home delivery. Postnatal care is by far the least-used service; of low-birth-weight infants is 6 percent. in 1982 only 5 percent of mothers in Costa Rica made one The proportion of false positives and false negatives should postpartum visit to a health care center, although nearly 97 decrease in developing countries where the incidence of the percent delivered in hospitals (Lettenmeier and others 1988). adverse outcome increases. For example, in developing coun- In Jamaica only 37 percent of new mothers received post- tries where the incidence of low birth weight may be two to partum care in 1981 despite the fact that the govemment there five times greater than in industrial countries, the predictive provides it free and locally through mobile clinics (WHO 1985). value of the screening systems should increase. Conversely, Various cultural factors cause the lack of demand for post- because maternal mortality occurs much less frequently than partum services, not the least of which is the widespread low birth weight and perinatal mortality, a risk assessment belief that new mothers must stay at home (Lettenmeier and system set up solely to prevent maternal complications will others 1988). misclassify (false positives and false negatives) a large propor- There have also been many problems with programs for tion of women.4 In order to have a successful maternal and training TBAS. For example, the traditional compensation may 380 Julia A. Walsh, Chns M. Feifer, Anthony R. Measham, and PailJ. Gertler be thought inadequate by the newly trained TBA, villagers may Table 17-10. Women Reporting at Least One believe the attendant is paid through the health care system Prenatal Visit during Most Recent Pregnancy, and not compensate her in the traditional way, and the train- Selected Countries ing may be of limited use without an additional investment in (percent) supervision. For best results, the training must be part of a Counitry Urban Rural Total serious attempt to use available resources, not a potentially - - cheap solution for remote areas. Sri Lanka 97 97 97 Emergency care suffers in many places from the lack of roads, Dominican Republic - - 95 .. . . ' ~~~~~~Liberia ... 91 77 83 reliable vehicles, restrictions on the use of gasoline, and lack Burundi 96 78 79 of public transportation which would integrate a referral sys- Thailand 94 74 77 tem. Some programs have developed innovative solutions to Brazil 86 51 74 this problem: one community in rural Somalia has a flag on the Ecuador 82 58 70 road used to signal passersby when a woman needs a ride to the Colombia 83 59 69 hospital (Lettenmeier and others 1988); Malawi developed Senegal a 95 46 63 bicycle-pulled stretchers; and Zambia gave women expecting Morocco 56 16 25 complications free tickets for bus transportation (Favin and __- -___-_-_----_____-_-__- others 1984). The issue always is whether an emergency, if one -Data not available. arises, can be dealt with quickly enough. Note; Based on demographic and health surveys in 1986 and 1987. Cov- ers onilv pregnancies within five yearS preceding the survev. Interviewers Many prenatal care and delivery interventions have been asked women: "When you were last pregnant, did you see anyone for a recommended depending upon local problems and resources. check on that pregnancy?"' Only those respondents checked by trained The actual health effect of an intervention, however, depends health care providers are indicated. a. Only women checked by a docior or nurse. on several factors: (a) rates of effectiveness, or the ability of an Source: Lettenmeier and others 1988. intervention to prevent or treat the problem; (b) accuracy of the diagnostic tests to identify those who would benefit from the intervention; (c) quality of care; (d) patient compliance; Lomort has a reasonably well developed health system, a (e) coverage-in this case, the proportion of women using the large proportion of women receive prenatal and delivery care service; and (f) frequency of the health problem. All these from trained birth attendants, and the average standard of factors plus cost and feasibility should be considered by coun- living is moderate. The effect of increasing contraceptive tries planning a system for improving maternal and perinatal prevalence on matemal mortality in Lomort would be low. health. Increasing the use of contraception beyond the current level would be prohibitively expensive and the gains are unclear. Strategies for Two Standardized Populations Therefore, we consider other avenues in Lomort. In order to bring the maternal mortality ratio down, perhaps In this section we use two hypothetical populations of one to 25 per 100,000 live births, as in urban China, more invest- million persons to illustrate maternal and perinatal program ment is needed in obstetrical care and perhaps some important strategies and priorities. The populations include a low- social reforms are required as well. China has achieved large mortality country and a high-mortality country. Because of the reductions in adverse maternal events through its radical one- close interrelationship of resource allocation for fertility con- child policy for population control, higher age at marriage, and trol and the management of pregnancy, we present the pre- better health care for pregnant women. The China example vention and case management illustrations in a unified underlines the importance of political will for health achieve- separate section. ments at this level of development. China has widely available family planning services and a strong political commitment to Low-Mortality Example maternal health, and it offers continuing education and moti- vational campaigns using different forms of persuasion (Chen The first population, referred to here as Lomort (for low and Kols 1982). Lomort could make substantial gains in ma- mortality), has already passed through much of the demo- ternal health with relative cost efficiency by improving the graphic transition. The fertility rate is 2.8, the infant mortality quality of existing services, focusing on use of the services, and rate is 51 per 1,000 births, the perinatal mortality rate is 35.7 coordinating with other sectors to implement social reforms per 1,000 births (because about half of perinatal mortality which promote women's health. occurs in the first week of life, about 35 percent of all infant Social reforms in particular can have a large effect on deaths occur in the perinatal period), and life expectancy at maternal health. Such reforms include legalizing abortion, birth is 64.4 years. We assume the maternal mortality ratio is raising the age of marriage, requiring universal education, and about 200 deaths per 100,000 live births (inferred from WHO providing welfare to the worst-off, highest-risk groups. Latin 1986) and that about 65 percent of women of reproductive age American countries have matermal mortality ratios similar to use contraceptives. Lomort's, and their revision of abortion policies could poten- Matemal and Penrnatal Health 381 tially save both lives and costs. In the early 1970s in Chile, Table 17-11. Additional Annual Operating and 3,250 hospital-induced abortions cost $30,000 during a brief Capital Costs for Maternal Health and Family period when abortions were legalized. Researchers calculated Planning in Lorort, a Standardized Population a savings of more than $200,000 above earlier costs for emer- of One MiUion, in Safe Motherhood Initiative: gency treatment of illegal abortion complications (Weston Limited-Effort Model 1986). Cultural changes are equally important: researchers (U.S. dollars) find that low use of maternal health services is often due to Expenditures Cost cultural perceptions of women's roles, which block the woman's ability to get care for herself (Leslie and Gupta 1989). Annual operating costs There may also be traditional practices which are dangerous, Staff 50,000 female circumcision, for example, and women's health advo- Transport 100,000 In-service training and supervision 75,000 cates are calling for national educational campaigns to discour- Equipment and supplies 75,000 age them. Health education, cultural campaigns 100,000 Lomort's health sector might implement a ten-year, limited Monitoring and evaluation 20,000 effort, safe motherhood program that emphasized health and Contingencies 30,000 nutrition education, hiring additional staff to extend commu- Total 450,000 nity outreach and screening and to improve existing services and increase their use. The program would offer additional Costpercapita 0.45 training for personnel; provide health and nutrition education Capital costs campaigns; and examine facilities, emergency systems, and Training 200,000 referral networks and renovate them as necessary. The cost Construction and upgrading 240,000 estimates fof this kind of effort are listed in table 17-11. Vehicles 160,000 On the basis of experience in countries which have im- Total 600,000 plemented similar efforts, we estimate that the program's effect a on adverse events could be as much as a 65 percent reduction Capital costs attriburab[e to materal health 300,000 in matemal mortality and morbidity, a 20 to 25 percent reduc- Annualized cost per capita 0.03 tion in the number of low-birth-weight babies, and a 35 percent reduction in perinatal mortality over the ten-year Toal costsc period. Table 17-12 contains before-and-after scenarios that Gross 480,000 use more conservative estimates of the progress which can be Per capita 0.48 achieved. The table provides two sets of calculations for the achievements of a limited effort, in part because current gaps a Assumes half of total capital costs attributable to maternal health. b. Assumes ten-year deprec[ation. in knowledge keep us from knowing just what can be accom- c. Annual operating costs plus annualized capital costs. plished. For the lower estimate, we assume that the matemal Source: Herz and Measham 1987. mortality rate will be reduced by 25 percent, low birth weight will drop in incidence from 8 to 7.5 percent, and the perinatal mortality rate will be reduced by 12.5 percent; for the more lights the fact that safe motherhood programs benefit both optimistic scenario, we assume that matemal mortality will be women and their infants. reduced by 50 percent, the incidence of low birth weight will drop to 7 percent, and the perinatal mortality rate will be High-Mortality Example reduced by 25 percent. The total cost of the program for a population of one million would be about $480,000 annually, Our second standardized population, Himort (for high mortal- or about $0.48 per capita. The estimated cost per death averted ity), can benefit substantially from relatively small investments (matemal and infant) ranges from $3,967 to $1,975, and the in women's health. The ideal matemal and child health pro- estimated cost per adverse event averted ranges from $1,103 gram would provide family planning services (including safe to $550. Adverse events include deaths, episodes of matemal abortions, prenatal care, and training of traditional birth at- morbidity, and low-birth-weight babies. tendants and health personnel) and regional referral and emer- The fertility rate and number of births have been held gency transportation systems. As mentioned earlier, efforts to constant in table 17-12 in order to isolate the effect of im- increase female enrollment in schools and improve female provements in prenatal and obstetric care alone. In reality, one nutrition would also contribute to better matemal and perina- would expect a further decline in fertility if services and tal health. community outreach were extended. The number of infant Himort is at an early stage of development. A large area is deaths and low-birth-weight babies avoided is much greater desert and people are scattered in villages and nomadic settle- than the number of adverse matemal events in spite of the ments. Agriculture is limited and transportation is sparse. greater percentage of reduction in rates for women. This high- Another area is forested and has high humidity and seasonal 382 Julia A. Walsh, Chris M. Feifer, Anthony R. Measham, and Paul J Gertler Table 17-12. Current Indicators of Matemal and Child Health in Lomort, a Standardized Population of One Million, in a Limited-Effort Safe Motherhood Program WVith limited-effort program Indicator Before program Conservative estimate Moderate estimate Demographics Population 1,000,000 1,000,000 1,000.000 Contraceptive prevalence (percent) 65 65 65 Fertility rate 2.8 2.8 2.8 Crude birth rate 24.5 24.5 24.5 Births (number) 24,500 24,500 24,500 Morbidity and mortality b Maternal mortality ratio (deaths per live 100,000 births) 200 150 100 Perinatal mortality rate (deaths per 1,000 births) 35.7 31.2 26.8 Matemal deaths (number) 49 37 25 Maternal morbidity (number)c d 784 592 400 Perinatal infant deaths (number) 875 766 656 Low-birth-weight babies (number) 1,960 1,838 1,715 Program effectiveness Births averted (number) n.a. 0 0 Maternal deaths averted (number) n.a. 1 2 24 Matemal morbidity averted (number) n.a. 192 384 Perinatal infant deaths averted (number) n.a. 109 219 Low birth weight averted (number) n.a. 122 245 Costs (in U.S. dollars) Program cost n.a. 480,000 480,000 Cost per capita n.a. 0.48 0.48 Cost per death averted n.a. 3,967 1,975 Cost per event averted n.a. 1,103 550 n.a. Not applicable. a. Among females age fifteen to forty-four. b. Assumes that improved obstetric services result in 25 percent and 50 percent reductions in conservative and moderate estimates, respectively. c. Number of matemal deaths multiplied by 16. Estimated ratio from Datta and others 1973. d. Assumes that improved obstetrIc services result in 12.5 percent and 25 percent reductions in conservative and moderate estimates, respectively. e. Predicted from regression (see appendix 17A). Assumes a 0.5 percent and I percent drop in low-birth-weight incidence in conservative and moderate es- timates, respectively, resulting in 6.2 percent and 12.5 percent reductions in the number of low-birth-weight babies. Source: Herz and Measham 1987. floods. People crowd together on meager farming plots; com- consequent infant deaths during the first year. This is an munities are often isolated from each other by washed-out extremely conservative estimate of perinatal mortality roads and unnavigable waterways. Some areas are more fa- [Edouard 19851). Matemal mortality is believed to range from vored: conditions there are less harsh, people live in larger 800 to 1,400 per 100,000 live births in the various regions; communities, health facilities are better developed, and roads health officials estimate a national average of 1,000 deaths. are adequate. The total fertility rate is 6.9 and the crude birth rate is 49.5 Health services are scarce in Himort, with a few exceptions, (see table 17-13 for a summary of statistics). Use of modern constrained by inadequate staffing, supervision, and supplies. contraception is low. Family planning is virtually unknown, so there are very few The problem facing Himort is how to improve maternal and modem contraceptive users except in the capital. Unsafe perinatal health with severe limitations on resources, inade- abortions are relatively common. Most women, about 90 per- quate facilities, and inadequate transportation. Discussing a cent, deliver at home; about 70 percent of births are attended hypothetical situation similar to Himort, Herz and Measham by untrained traditional midwives orfamily members. The few (1987) proposed that the choice between "providing a little to better-off places are served by health centers and a district many" and "providing more to a few" inadequately represents hospital, but matemal health is not given priority. the dilemma. Little progress can be made in a situation like Life expectancy at birth is fifty-one years, the infant mortal- this without additional investment in community-level care, ity rate is 129 deaths per 1,000 births and perinatal mortality referral facilities, and transportation. Each region must indi- is 51.6 deaths per 1,000 births (note that the ratio of perinatal vidually assess its restrictions and needs and may need to plan mortality to all infant mortality is 0.4, lower than in Lomort an appropriate program which differs from those of other because of the higher prevalence of infectious disease and regions. Maternal and Pernatal Health 383 Table 17-13. Current Indicators of Matemal and In table 17-14 we present demographic data and health Child Health in Himort, a Standardized Population statistics for three scenarios: increases in the prevalence rate of One Million for contraception to 20 percent, 40 percent, and 60 percent, Indicator 'Value respectively. Through the increase in contraceptive users, fertility rates decline and the number of births is reduced, Demographics thereby reducing exposure to pregnancy risk. The matemal Population 1,000,000 mortality ratio is held constant throughout table 17-14, how- CF ntracepttve pre6alence (percent) .9 ever, emphasizing that unlike the maternal mortality rate, it is Crude birth rate 49.5 unaffected by overall changes in fertility. In spite of the con- Births (number) 49,500 stant maternal mortality ratio, maternal deaths and morbidity Mforbidityv and Mortalit-y drop remarkably. Using the new fertility rates, calculated on Maternal mortality ratio (deaths per 1,000 the basis of the contraceptive prevalence rates, we have also live births) 1 ,00 derived revised rates of low birth weight. Perinatal infant death Perinatal mortality rate (deaths per 1,000 rates are assumed to drop 5 percent for every 20 percent births) 51.6 increase in contraceptive prevalence. The number of low- Maternal deaths (number) 495 birth-weight infants and infant deaths declines with reduc- Maternal morbidity (number) 7,920 tions in fertility for the following reasons. First, with fewer Perinatal infant deaths (number) 2,554 births there is a smaller risk pool; second, in order to provide family planning services, health personnel increase their con- a. Among females age fifteeni to forty-four. tacts with the community, which improves health outcomes; b. Number of matemal deaths multiplied by 16. Estimared ratio from third, part of the demand for family planning results from social Datta and others 1973. c. Predicted from regression (see appendix 1 7A). Represents 15 percent. improvements, which also improve health. Source: Authors. The success of family planning efforts depends on program quality and acceptability, a strong demand-generation effort, method effectiveness, and consistent availability of supplies. Among the options available to Himort are preventive Demand is increased by such changes as increasing the age at activities as described earlier. In contrast to Lomort, family marriage, increasing female education, and employment pat- planning is the most cost-effective preventive activity that temns which decrease the value or necessity of children. The could be provided. Addition of some other cost-effective strat- demographic changes presented in table 17-14 may occur over egies discussed in earlier sections would improve cost-effec- several decades. The fertility rate in Thailand fell by almost 40 tiveness, for example, STD screening and treatment. A second percent in twenty-five years-one of the most rapid reductions option would strengthen prenatal care and train birth atten- recorded. The crude birth rate was about 44 in the early 1960s dants for pregnancy management. We begin by illustrating the (Sherris and others 1985) and fell to 24.5 in 1985 (WHO 1986). likely costs and benefits of family planning alone and then The drop in birth rate predicted in the model, from 49.5 to discuss the implications of a broader matemal health program. 28.4, is consistent with experience in Thailand but could take many more years, depending on the setting. The assumption FAMILY PLANNING ALONE. If Himort did nothing to for the costs discussed in table 17-14 is that expenses increase strengthen prenatal and delivery care but did offer family exponentially as population coverage goals increase, but ig- planning services, substantial improvements in maternal and nored in these estimates are adjustment costs, which could infant health outcomes could result. There are a number of occur over long periods and which may be necessary for signif- methods of providing family planning services, though social icant increases in contraceptive use to take place. The data in marketing, community-based distribution, postpartum pro- table 17-14 do show that the cost efficiency of family planning jects, and voluntary sterilization services appear to be among for improving maternal and perinatal health decreases as con- the most cost-effective. Studies show costs in 1980 ranging traceptive prevalence increases. The cost per death averted from $1 per couple-year of protection for social marketing in ranges from $806 to $1,338 to $2,962 as contraceptive preva- Colombia to $90 for some clinic programs (Sherris and others lence reaches 20 percent, 40 percent, and 60 percent of couples 1985). A community-based delivery system, using traditional at risk.' Costs per adverse event averted range from $139 to birth attendants and local health personnel, might be a good $229 to $505. Adverse events include deaths, episodes of method for Himort. Such a program in rural Cheju Province maternal morbidity, and low-birth-weight babies. in the Republic of Korea between 1976 and 1979, which also included free sterilization, resulted in a 20 percent increase in COMPREHENSIVE MATERNAL HEALTH PROGRAM. In this section women who accepted contraceptives and a decline in the total two plans are proposed for Himort which go beyond family fertility rate of more than 1.5 births in a period of five years, planning by including various maternal care and structural from 1975 to 1980. This program cost $0.47 annually per capita improvements to the health system. They assume that Himort and $9.35 per couple-year of protection (Chen and Worth has adopted the goal of a family planning prevalence rate of 20 1982; Park and others 1982). percent and that it also undertakes additional reforms in the 384 Juiia A. Walsh, Chris M. Feifer, Anthony R. Measham, and Paiu J. Gertler Table 17-14. Effects on Matemal and Child Health of Three Scenarios Based on Contraceptive Prevalence Rates Indicator 20 percent prevalence rate 40 percent prevalence rate 60 percent prevalence rate Demographics Fertility rate 5.3 4.3 2.6 Births (number) 41,170 34,183 28,381 Morbidity and mortality Matemal mortality ratio (deaths per 100,000 livebirths) 1,000 1,000 1,000 Perinatal mortality rate (deaths per 1,000 births) 49 46.4 43.9 Maternal deaths (number) 412 342 284 Maternal morbidity (number) C 6,592 5,472 4,544 Perinatal infant deaths (number) 2,017 1,586 1,246 Low-birth-weight babies (number) 5,764 4,444 3,406 Program effectiveness Births averted (number) 8,330 15,317 21,119 Maternal deaths averted (number) 83 153 211 Matemal morbidity averted (number) 1,328 2,448 3,376 Perinatal infant deaths averted (number) 537 968 1,308 Low birth weight averted (number) 1,661 2,981 4,019 Costs (in U.S. dollars) Progra e 500,000 1,500,000 4,500,000 Cost per capita 0.50 1.50 4.50 Cost per death averted 806 1,338 2,962 Cost per event averted 139 229 505 Note; Prevalence rate is percent of women fifteen to forty-iour using contraceptives. a. Predicted from regression (see appendix 17A). b. Assumes decreases in perinatal mortality rate from the original high of 51.6 through contraceptive prevalence rates of 20 percent, 40 percent, and 60 per- cent, are S percent, 10 percent, and 15 percent, respectively. c. Maternal deaths times 16. Estimated ratio from Datta and others 1973. d. AssuImes incidence of low birth weight decreases for each 20 percent increase in contraceptive prevalence. With decreases in numbers of births, the re- sult is total decreases in number of low-birth-weight babies of 22 percent, 40 percent, and 54 percent, respectively, for the 20 percent, 40 percent, and 60 per- cent contraceptive prevalence rates. e. Longer time periods are nieeded to achieve higher rates of contraceptive prevalence and are likely to involve costs not included here. Source: Authors. area of maternal health care. The limited goal of a 20 percent Measham (1987), who provide a more detailed discussion of prevalence rate for contraception was chosen because it is a these items. reasonable accomplishment for a ten-year plan; higher goals The goals of the limited effort are to reduce the matemal would demand a longer time horizon and more resources and mortality ratio by 20 percent through prenatal and delivery would involve much greater uncertainty. care and trained birth attendants. In addition, we expect the The goals for a comprehensive matemal health program in plan to reduce the incidence of low-birth-weight babies to 13 Himort would differ according to the existing infrastructure, percent (dropping from 14 percent with family planning socioeconomy, and ecology of particular subareas. The first alone) and perinatal infant mortality by an additional 12.5 effort, a limited one, is most appropriate for the very poor, percent. These decreases would be the result of reducing the isolated desert and rain forest areas that make up the bulk of number of high-risk pregnancies and of providing better care the country. The plan includes resources necessary for a limited to those who become pregnant. The limited plan would in- prenatal care and birth attendant training program; this effort clude the following elements: would be carried out in addition to the family planning effort, aiming for 20 percent prevalence. The second, moderate effort * Upgrading of existing facilities to ensure the availability would best be applied in areas that are not as poor, are densely of matemal health cate and the establishment of four more populated, have existing health facilities, including a hospital, centers (two with cesarean section and surgical family plan- and where women use health services more. The moderate ning capacity) plan includes the resources necessary for a moderate prenatal * Investment in an emergency transportation system, one care and birth attendant training program; this is again in four-wheel drive vehicle for each new center, so that more addition to the previously mentioned family planning effort. women can reach the existing service areas The goals of each effort are described below. We have adapted * Introduction of risk screening and development of the program design and cost estimates, developed by Herz and plans for at-risk women to deliver in health facilities; Maternal and Perinatal Health 385 three mobile units equipped with radios and staffed by Table 17 15. Additional Annual Operating and three health care workers; and maternity villages where Capital Costs for Maternal Health in Himort Safe those referred to facilities for delivery can await the start Motherhood Initiative: Limited-Effort Model of labor (U.S. dollars) * Strengthening of all community-based services by train- Expenditure Cost ing all traditional birth attendants and providing them with basic medications, compensation for family planning activ- Annual opeatingcosts150,00 ities, and radios, so that prenatal and uncomplicated deliv- Transport 100,000 ery care is readily available In-service training and supervision 50,000 * Coordination of outreach so that facilities and trained Equipment and supplies 75,000 personnel are efficiently used Health education 25,000 * Conduct of research activities to identify the most effec- Monitoring anid evaluation 20,000 tive strategies ~~~~~~~Contingencies 30,000 tive strategies * Depending on the community or region, other potential Total 450,000 components include training nurse-midwives to enlarge the Cost per capitaa 0.45 pool of health workers able to provide blood transfusions, surgical family planning, and cesarean sections; strengthen- Capitnl costs ing health services management and giving higher priority Construction and upgrading 100,000 to matemal health; and encouragmg community groups to Vehicles 160.000 become involved in women's health and safe motherhood. Total 600,000 The total cost of the program for a population of one million, Capital costs attributable to maternal health. 300,000 including family planning program costs, would be about Annualized capital costc 30,000 $980,000 annually, or about $0.98 per capita (see table 17-15 Annualized cost per capitaa 0.03 for breakdown of costs for the maternal health component alone). The cost per death averted (matemal and infant) is Total costs about $1,303, and the cost per adverse event averted is about Gross 480,000 $179*6 Adverse events include deaths, matemal morbidity, and Per capita 0.48 low-birth-weight babies. The goals of the moderate effort are to reduce the maternal Note s a. Assumes population of 1 million. mortality ratio by 40 percent through prenatal and delivery b. Assumes half of total capital costs atrributable to maternal health. care, facility development, and training of birth attendants. In c- Assumes ten-year depreciation. d. Annual operaring costs plus annualized capital costs. addition, we expect the plan to reduce the incidence of low- Source; Herz and Measham 1987. birth-weight babies to 12 percent (dropping from 14 percent from family planning alone) and perinatal infant mortality by an additional 25 percent. These reductions are more conser- complex services offered at health posts, health centers, and vative than most cited in the literature. The plan would the hospital, respectively include the following elements: * Development of an emergency transport system and training of all personnel for appropriate referrals. * Establishment of a community outreach system for pre- natal care, which would provide nutrition advice and preg- The total cost of the program for a population of one million, nancy risk screening, make appropriate referrals, and including the cost of the family planning program, would be encourage use of health facilities about $2 million annually, or about $2 per capita (see table * An increase in the number of health posts to one for 17-16 for breakdown of costs of matemal health components every 10,000 population, built with community assistance, alone). The cost per death averted (maternal and infant) is and the training of all TBAs in outreach and routine care about $1,554, and the cost per adverse event averted is about activities $258.' In table 17-17 we summarize the effect of and cost information for the family planning effort alone and the family * An increase in the number of health centers by the plnnn efor plslmtdadmdrteoserc.rgas construction of five new ones in five years, to be used as For e limited and m o bstetrics programs,ewe assue. referra centes for pegnanc compliationsFor the limited and m-oderate obstetrics programs, we assume some community contributions, and both programs are based * Addition of ten maternity beds to the district hospital on assumed correlations between improved prenatal care and and an operating room with the capacity to handle high-risk birth outcomes. The call for emergency transport systems and deliveries, cesarean sections, and surgical contraception care is in recognition of the fallibility of current risk-screening * Training of additional health personnel at each level so methods and the likelihood that some women will continue to that a regional network of services exists with increasingly use health services only when there is a problem. 386 Julia A. Walsh, Chnis M. Feifer, AnthonN R. Measham, and Paul J. Gertler Table 17 16. Additional Annual Operating and Finally, a word of caution is in order. The calculations Capital Costs for Maternal Health in Himort Safe presented here depend on heroic assumptions, because reliable Motherhood Initiative: Moderate-Effort Model data are lacking. We have found virtually no data on the effect (U.S. dollars) of maternal health programs on maternal health, infant mor- ExpendLture Cost tality, or low birth weight, nor were we able to separate perinatal from neonatal or general infant mortality. There is Annual operating costs no clear scientific evidence that a particular component con- Staff 575,000 tributes to a specified decline in adverse events. We proposed Transport 125,000 In-service training and supervision 150,000 goals for the hypothetical populatons and presented program Equipment and supplies 300,000 components without clear evidence of the magnitude of the Health education 50,000 effect of these measures. In sum, the declines in adverse out- Monitoring and evaluarion 50,000 comes that we have suggested are no more than best estimates Total 1,250,000 of the likely effect of the measures proposed based on the limited evidence available from the literature. Cost per capita 1.25 Capital costs Priorities Training 800,000 Construction and upgrading 3,600,000 Finally, we recommend priorities for resource allocation, and Vehicles 600,000 for research. Total 5,000,000 b ' Priorities for Resource Allocation Capital costs attributable to maternal health 500,000 Annualized capital costc 25,000 Annualized capital cost 25,000 Maternal mortality, perinatal mortality, high fertility, and low Annualized cost per capita 0.25 birth weight are all high-priority problems. Technically feasi- Total costsd ble and affordable methods exist which can significantly re- Gross 1,500,000 duce the incidence of perinatal and matemal disease and death Per capita" 1.50 and increase the prevalence of contraceptive use. Furthermore, - -A- pu- -i - - - - -ml investments in family planning and maternal and perinatal a. Assumes populati(mn (if I aullion. b. Assumes half of capital costs attributable to maternal health. health compare favorably with other health investments, such c. Assumes ten-vear depreciation. as curative care, that are more costly but have a more limited d. Annual operating costs plus annualized capital costs effect on the population. indeed, because materal and pen- Source: Herz and.Measham 1987. natal health problems have long been neglected, they should be at the top of the priority list in most countries-especially Summary of Hypothetical Examples those in which maternal mortality, neonatal mortality, and low birth weight are high (that is, South Asia and Sub-Saharan As the examples illustrate, efforts to improve matemal and Africa). perinaral health should involve appropriate combinations of As the examples set forth in the last section illustrate, the family planning, prenatal care, and obstetric improvements. priorities for particular programs that target matemal and Health services should also give priority to screening and perinatal health depend on the demographic situation, partic- referral networks and emergency transportation systems. ularly the fertility rate and level of contraceptive use, and on Screening for potential complications is used to direct high- ecological and economic factors. When countries are deciding risk women away from home delivery and into adequately on the balance between matemal health and other health staffed and equipped delivery facilities. Emergency transporta- programs, they will find it worth considering that maternal tion systems are also needed for cases in which screening was health affects the health of infants, children, and the depen- absent or produced a false negative. dent elderly. The loss of a mother threatens family survival. In countries with high fertility, significant reductions in The choice need not be between matemal health services and matemal and infant deaths result both from reductions in the other programs; creative health administrators are able to number of pregnancies through family planning and from incorporate the needs of women into primary or other health improved maternal care. Where fertility is already low, reduc- care programs. tions result almost entirely from improved matemal and peri- The focus in this chapter has been on mortality and morbid- natal care. Altogether, the investments required for safer ity; the other side of the equation, the magnitude and quality motherhood and healthier starts to life are relatively low, and of health, has largely been ignored. When safe motherhood the potential gains are great. Still, the health system must have programs are implemented, the priority should be to promote the capacity to make services available to people in all parts of good health, not merely to avoid death. Countries should look the country. Extending service coverage is especially critical at the quality of life for newbom children and the quality of for those countries whose current budget is allocated largely to life for women both during and after the reproductive period. the urban centers. These priorities will require intersectoral collaboration, strong MatenoL and Perinatal Health 387 Table 17-17. Estimated Effect of Comprehensive Maternal Health Plan Alone and for Family Planning with Limited and Moderate Obstetrics Program in Himort Familv Familv planning Family planning Before planning and limited and moderate Indicator program only obstetrics program obsterrcs program Demographics Population 1,000,000 1,000,000 1,000,000 1,000,000 Contraceptive prevalence (percent)" 0 20 20 20 Fertility rate 6.9 5.3 5.3 5.3 Crude birth rate 49.5 41.17 41.17 41.17 Births (number) 49,500 41,170 41,170 41,170 Morbidity and Mortality b Matemal mortality ratio (deaths per 100,000 live births) 1,000 1,000 800 600 Perinatal mortality rate (deaths per 1,000 births) 51.6 49 47.8' 36.8c Matemal deaths (number) 495 412 329 247 Maternal morbidity (number) 7,920 6,592 5,264 3,952 Perinatal infant deaths (number) 5,643 2,017 1,968 1,515 Low-birth-weight babies (number) 7,425 5,764 5,352 4,940 Program effectiveness Births averted (number) n.a. 8,330 8,330 8,330 Matermal deaths averted (number) n.a. 83 166 248 Matemal morbidity averted (number) n.a. 1,328 2,656 3,968 Perinatal infant deaths averted (number) n.a. 537 586 1,039 Low birth weight averted (number) n.a. 1,661 2,073 2,485 Costs (in U.S. dollars) Program cost n.a. 500,000 980,000 2,000,000 Cost per capita n.a. 0.50 0.98 2.00 Cost per death averted n.a. 806 1,303 1,554 Cost per event averted n.a. 139 179 258 n.a. Not applicable. a. Among females age fifteen to forty-four. b. Assumes additional decreases of 20 percent and 40 percent for limited and moderate efforts, respectively, over the decreases achieved by family planning alone. c. Assumes additional decreases of 12.5 percent for limited and moderate efforts, respectively, over the decreases achieved by family planning alone. Source: Authors' calculations. political will, and a willingness to recognize the importance ductive tract infection and STDs in high-incidence areas; refer- and value of women. ral of women developing complications during pregnancy to Family planning and access to safe abortion as a backup are facilities able to provide higher-level care; and encouraging clearly a priority. Successful family planning programs require higher-risk women to deliver in hospital. Somewhat lower a demand for spacing children and limiting family size. Health priorities for prenatal care are education aimed at decreasing improvements must be accompanied, therefore, by concomi- alcohol and tobacco abuse; monitoring weight and blood pres- tant efforts to improve female education and women's eco- sure of women during pregnancy; educating for signs of prema- nomic opportunities. Declines in fertility will then make ture labor; screening all women for sexually transmitted reproductive health care services more affordable. diseases and treating identified cases; and performing pelvic Access to prenatal and competent obstetric care also merit examinations to check gynecological anatomy in all women. high priority. The content of these services, their quality, and High priorities for delivery care are providing hygienic delivery their use certainly influence their effectiveness. Indeed, merely kits to birth attendants; training TBAS for delivery emergencies improving the quality of existing care and increasing the use (includes use of oxytocics) and neonatal resuscitation; arrang- of existing services could have a substantial effect in many ing regional referral systems for skilled obstetrical care and countries. A study in Jamaica concluded that 68 percent of neonatal intensive care. In all countries, health personnel matemal mortality could be prevented through improvements should locate favorable traditional practices and build on these in the quality of care (Walker and others 1986). for health promotion. Some suggested priorities for prenatal and delivery care follow: screening for high-risk women; tetanus toxoid im- Priorities for Operational Research munization; iron and folate supplements; malaria prophylaxis in endemic areas; nutrition education and supplements for the Maternal and perinatal health are areas that have been se- most malnourished mothers; testing and treatment for repro- verely neglected by researchers. The focus of programs which 388 Julia A. Walsh, Chris M. Feifer, Anthony R. Measham, and Paul J. Gertler have affected the health of women (such as nutrition supple- 5. This calculation is very sensitive to the estimated perinaral mortality. If mentation) has more often been on infant mortality. We have a higher initial estimate for perinatal mortality is used, then the costs per death averted would decrease almost 50 percent. little evidence of the factors causing mortality rate declines, 6. See note 6. no measures of the economic effect of death or disability, nor 7. See note 6. any firm idea of the prevalence or duration of maternal and perinatal illness. There is insufficient information to relate different service components directly with effectiveness and References cost. It goes without saying that more field research is needed in this area, and more programs should include an evaluation Bamisaiye. A., Olikoye Ransome-Kuti, and A. A. Famurewa. 1986. "Waiting component. Time and Its Impact on Service Acceptability and Coverage at an MCH Clinic at Lagos, Nigeria." Journal of Tropical Pediamcs 32:158-61. Belsey. M. A., and Erica Royston. 1987. "Overview of the Health of Women and Children." Paper presented at the Intemational Conference on Better Appendix 17A. Regression Equations Used in the Health forWomen and Children through Family Planning,Nairobi, Kenya, Construction of Tables 17-11 through 17-17 October 5-9, 1987. Berman, S. N., R. Harrison, W. T. Boyce, W. W. J. Haffner, M. Lewis, and J. Data from World Development Indicators (I 988) were used to B. Arthur. 1981. "Low Birthweight, Prematurity, and Postpartum Endome- estimate the regressions used for tables 17-11 through 17-17. triosis."Jourrnalof the American MedicalAssociation 257:1189-94. All equations include gross national product (GNP) to control Bhatia, J. C. 1985. "Matemal Mortality in Anantapur District, India: Prelim- for societal effects like education, nutrition, and general health mary Findings of a Study." Family Health Division, World Health Organi- status. The fertility rates, rather than crude birth rates, were zation, Geneva. used as Independent variables because they are a betr 1i . Blacker,J. G. C. 1987. "Health ImpactsofFamily Planning."HealthPolicyand used as independent variables because they are a better indi- Planning 2193-203. cator of risks for adverse matemal events. By multiplying the B lanning 3 . Bobodilla, Fernandez 3. L. 1988. Quality of Pennatal Medical Care in Mexico fertility rate and the maternal mortality ratio, one can also City. Instituto Nacional de Salud Ptiblica. Mexico City. derive a matemnal mortality rate. Brown, S. 1985. "Can Low Birthweight be Prevented?' Family Planning Per- ln (birthrate) = 4.92-0.009 (% contraceptive prevalence) specnves 17:112-18. -0.17 In (GNP) Brnunham, R. C., K. K. Holmes, and D. Eschenbach. 1984. "Sexually Trans- (22.84) (4 97) (4.37) mitted Diseases in Pregnancy." In K. K. Holmes, P.-A. Mardh, F. Sparling, and P. J. Wiesner, eds., Sexually Transmitted Diseases. New York: McGraw- In (fertility rate) = 3.63 - 0.0035 (% contraceptive preva- Hill. lence) -0.32 In (GNP) CLC (Centers for Disease Control). 1987. "Enterically Transmitted Non-A, (21.65) (3.45) (13.24) Non-B Hepatitis-East Africa." Morbidity and Mortality Weekly Report 36:241-44. In (% low birth wt) = 3.24 + 0.26 In (fertility rate) - 0.19 Chen, K., and George Worth. 1982. "Cost-Effectiveness of a Community- In (GNP) Based Family Planning Program in Cheju, Korea." Paper presented at the (4.75) (1.53) (2.71) Intemational Health Conference, June 14-16, 1982, Washington, D.C. Chen, L. C., M. C. Gesche, S. Ahmed, A. I. Chowdhury, and W. H. Mosley. 1974. "Matemal Mortality in Rural Bangladesh." Studies in Family Planning 5:334-41. Notes Chen, R. J. 1985. "Matemal Mortality in Shanghai. China." Family Health Division. World Health Organization, Geneva. I. The World Health Organization uses the ratio but calls it a rate. Chen, P. C.. and A. Kols. 1982. "Population and Birth Planning in the People's 2. The number of fetal deaths included depends upon the lower limit set for Republic of China." Population Reports, J, 25. Johns Hopkins University, fetal viability. Recently, the World Health Organization has recommended that perinatal statistics for international comparisons be restricted to fetuses and newborn infants with a birth weight of at least 1,000 grams (or, when the Chi, 1. C., T. Agoestina, and J. Harbin. 1981. "Maternal Mortality at 12 birth weight is not available, the gestational age of twenty-eight weeks or Teaching Hospitals in Indonesia: An Epidemiological Analysis." Interna- crown-to-heel body length of 35 centimeters) for both numerator and denom- rionaljournal of Gynecology and Obstetrics 24:259-66. inator for the perinatal mortality rate (Edouard 1985; Lancet 1991). Datta, S. P., [>. K. Sriniivas, R. V. Kale, and R. Rangaswamy. 1973. "Evaluation 3. Intrauterine growth retardation can be subdivided into "disproportional" of Matemal and Infant Care in a Rural Area." Indian Jounial of Medical or "wasted" IUGR infants, whose length and head circumference are relatively Sciences 27:120-28. normal for their gestational age but who are thin, with low weight-for-length Donaldson, P. J., and J. 0. G. Billy. 1984. "The Impact of Prenatal Care on and skinfold measurements; and "proportional" or "stunted" IUGR infants with Birthweight." Medical Care 22:177-88. proportional reductions in weight, length, and head circumference. The Edouard, L. "The Epidemiology of Perinatal Mortality." World Health Statistics distinction seems to relate to an earlier and more persistent impairment in Quatterl 38(3):289. growth in the stunted group. "Wasted" infants seem to grow faster postnataily, catch up to normal size more rapidly, and have fewer severe cognitive deficits Fauveau, Vincent, K. Stewart, S. A. Khan, and J. Chakraborty. 1991. "Effect than the "stunted" ones. Unfortunately, large studies comparing the relative on Mortality of Community-Based Matemity Care Programme in Rural incidence of these two forms of IUGR have not been published. Bangladesh." Lancer 338:1183-86. 4. The highest matemal mortality ratios are about 8 per 1,000, the highest Favin, M., B. Bradford, and D. Cebula. 1984. "Improving Matemal Health in perinatal mortality is about 100 per 1,000, and the maximum incidence of low Developing Countries." World Federation of Public Health Associations, birrh weight is up to 500 per 1,000. Geneva. Matenial and Pennatal Health 389 Fortney, J. A. 1987. "The Importance of Family Planning in Reducing Mater- Llewellyn-Jones, D. 1965. "Severe Anemia in Pregnancy." Australian and New nal Mortality." Studies in Family Planning 18:109-15. Zealand]ournal of Obstetrics and Gynecology 5:191-97. Fortney, J. A., 1. Susanti, S. Gadalla, S. Saleh, P. J. Feldblum, and M. Ports. ILondon, K., J. Cushing, John Cleland, J. E. Anderson, L. Morris, S. H. Moore, 1988. "Matemal Mortality in Indonesia and Egypt." lnternatonal]ournalof and S. 0. Rutstein. 1985. "Fertility and Family Planning Surveys: An Gynecology and Obstetrics 26:21-32. Update." Population Reports, M, 8. Johns Hopkins University, Population Greenwood, A. M., B. M. Greenwood, A. K. Bradley, K. Williams, F. C. Information Program, Baltimore. Shenton, S. Tulloch, P. Bypass, and F. S. J. Oldfield. 1987. "A Prospective Lopez, A. D. 1990. "Causes of Death: An Assessment of Global Patterns of Survey of the Outcome of Pregnancy in a Rural Area of the Gambia." Mortality around 1985." World Health Statistcs Quarterly 43:91-104. Bullenn of the World Health Organization 65:635-43. ILucas S. B., J. K. G. Mati, V. P. Aggarwal, and H. Sanghvi. 1983. "The Hamilton, S., Barry Popkin, and D. Spicer. 1984. Women and Nutrinon in Third Pathology of Perinatal Mortality in Nairobi, Kenya." Bullenn of the Society World Countnes. New York: Praeger. of Pathologists 76:579-83. Harrison. K. A. 1980. "Approaches to Reducing Matemal and Perinatal McCormack, W. M., B. Rosner, Y.-H. Lee, A. Munoz, D. Charles, and E. H. Mortality in Africa." In R. H. Philpott, ed., Maternity Services Iti the Devel- Kass. 1987. "Effect on Birthweight of Erythromycin Treatment of Pregnant oping World-What the Commumty Needs. London: Royal College of Ob- Women." Obstemcs and G vnecology 69:202-7. stetrics and Gynecology. McCormick, Marie. 1985. "The Contribution of Low Birthweight to Infant - 1985. "Childbearing, Health, and Social Priorities: A Survey of Mortality and Childhood Morbidity." New' England Journal of Medicine 22,774 Consecutive Hospital Births in Zaria, Northem Nigeria." Brintsh 312:82-90. Journal of Obstetncs and Gynecology Supplement 5:1-119. Maine, Deborah. 1981. "Fami ly Planning: Its Impact on the Health of Women Henrv, A.. and P. T. Piotrow. 1979. "Age at Marriage and Fertilitv." Population and Children." Columbia University, Center for Population and Family Reports, X., 4. Johns Hopkins University, Population Information Program, Health, New York. Baltimore. Maine, Deborah, Allan Rosenfield, M. Wallace, A. M. Kimball, B. E. Kwast, Herz, Barbatra. and Anthony R. Measham. 1987. "The Safe Motherhood Emil Papiemik, and S. White. 1987. "Prevention of Matemal Deaths in Initiative: Proposals for Action." World Bank Discussion Paper 9. Wash- Developing Countries: Program Options and Practical Considerations." ington D.C. Paper presented at the Intemational Safe Motherhood Conference, Feb. Howard, Deborah. 1987. "Aspects of Matemal Morbiditv: The Experience of 10-13, Nairobi. Less Developed Countries." In Advances in International Maternal and Child Mekhemar, S., A. el Sharbibi, 1. Mourad, and H. Riad. 1984. "High Risk Health. Vol. 7. London: Oxford University Press. Pregnancy among Women under 20 Years of Age in a Rural Community." lam,J. D. 1989. "Current Starusof Prematuritv Prevennon."JANA 262:265466. In M. M. Fayad and M. 1. Abdalla, eds., Medical Educaton in the Field of Investigators of the Johns Hopkins Study of Cervicitis and Adverse Pregnancy Pnmary Maternal and Child Health Care. Cairo: Egyptian Ministry of Health. Outcome. 1989. "Association of Chlarnydia trachomatis and Mycoplasna Monteith, Richard S., Charles W. Warren, Egberto Stanziola, Ricardo Lopez hominis with Intrauterine Growth Retardation and Preterm Delivery." Urzua, and Mark W. Oberle. 1987. "Use of Matemal and Child Health Amnerican Journal of Epidemiology 129:1247-5 7. Services and Immunization Coverage in Panama and Guatemala." Bullenn lM (Institute of Medicine). 1985. Preventing Low Birthsweight. Washington of the Pan-American Health Organizaanon 21:1-15. D.C.: National Academy Press. Moutquin, J. M.. R. Gagnon, C. Rainville, L. Giroux, G. Amyot, R. Bilodeau, Kauiitz, A. M., C. Spence, T. S. Davidson, R. W. Rochat, and D. A. Grimes. and P. Raynauld. 1987. "Maternal and Neonatal Outcomes in Pregnancies 1984. "Perinatal and Maternal Mortality in a Religious Group Avoid- with No Risk Factors." Canadian Medical AssocianonJounsal 137:728-32. ing Obstetrics Care." American Journal of Obstetrics and Gynecology Naeye, R. L. 1980. "The Role of the Pathologist in the Developing World." 150:826-31. In R. H. Philpott, ed., Maternity Services in the Developing World-What the Kramer, M. S. 1987. "1)eterminants of Low Birthweight: Methodological Cominiunity Needs. London: Royal College of Obstetricians and Gynecolo- Assessment and Meta-analysis." Buletin of the World Health Organization gists. 65:663-737. jo, G.A., and V. Y. Savage. 1974. "A Ten Year Review-of Matemal Mortaliry Kundsin, R. B., and S. S. Hipp. 1988. Impact oni the Fetus of Parental Sexualls Rates in the University College Hospital, lbadan, Nigeria." Amencan]four- Transmitted Disease. New- York Academy of Sciences, New York. - nal of Obstetrics and Gynecolog 118:517-22. Kwast, B. E., W. Kidane-Mariam, E. M. Saed, and F. G. R. Fowkes. 1984. Oyedehi G. A.,S. K.Olamijulo,and K.T.Joiner. 1983. "Experienceat Wesley: "Report on Maternal Health in Addis Abaha, Sept. 1981-Sept. 1983: A 1,391 Consecutive Admissions into the Neonatal Unit (Hurfod Ward)." Comnsunity Based Survey on the Incidence and Etiology of Matemal Journal of Tropical Pediatrics 29:206-12. IMortalityand the Use of Maternal Services, and Confidential Inquiries Into Park, C. B., J. A. Palmore, and L. J. Cho. 1982. "The Korean Experience." Maternal Deaths in Addis Ababa, Ethiopia." Save the Children Federation, Paper presented at the 52nd Annual Meeting of the Population Association Addis Ababa, Ethiopia. of America, April 28-May 1, 1982, San Diego, Ca. Kwast, B. E., and J. A. Stevens. 1987. "Viral Hepatitis as a Major Cause of PCC( (Population Crisis Committee). 1988. "Country Rankings of the Status Maternal MXortality in Addis Ababa, Ethiopia." niteTnational Journal of of Wonsen: Posr, Powerless, and Pregnant." Briefing Paper 20. Washington Gynecology and Obstetrics 2 5:99-106. D.C. Lancet. 1991. "Perinatal Mortality Rates-Time for Change?" 337:331. P'ufler, R. R.. and C. V. Serrano. 1973. "Patterns of Mortality in Childhood." Leslie, J., and C. Gupta. 1989. "Utilization of Formal Sernices for Maternal Pan-American Health Organiation, Washington, D.C. NutritionandHealthCare."lntemationalCenterforResearchonWomen, Rao, A. B. 1985. "Matemal Niortality in India: A Review." Family Health Washington, D.C. Division, World Health Organization, Geneva. Lettennseier, C., Laurie Liskin, C. A. Cltirch. and !. A. Harris. 1988. Ratnam, S. S., and R. N. V. Prasad. 1984. "Inadequacies of Present Health "Mothers' Lives Matter: Maternal Health in rise Community." Population Services and Strategies to Improve Maternal and Child Health in Reports, L, 7. Johns Hopkins University, Population Information Program, Developing Countries." Intemrational Jourial of Gynecology and Obstetrics Baltimore. 27:463-66. Li, B. Y., A. M. Dong, and J. R. Zhuo. 1982. "Outcomes of Pregnancy in Reich, j. 1987. "The International Conference oni Better Health for Women Hong-qiao and Qi-Yi Communes." American Journal of Public Health 72 and Children through Family Planning." International Family Planning Per- Supplement 9:30-32. specutes 13:86-89. 390 Julia A. WaLsh, Chris M. Feifer, Anthony R. Measham, and Paul J. Gertler Rinehart, W., A. Kols, and S. H. Moore. 1984. "Healthier Mothers and Walsh,J.A.,and S.Hutchinson. 1989. "The Importance ofGroupBStreptococ- Children Through Family Planning." Population Reports, J, 27. Johns Hop- cal Disease in the Developing World." Pediatric Infectious Diseases 9:271-77. kins University, Population Information Program, Baltimore. Wasserheit, J. N. 1989. "The Significance and Scope of Reproductive Tract Royston, Erica. 1982. "The Prevalence of Nutritional Anemia in Women in Infections forThird World Women." lnrernationaljournalof Gynecologyand Developing Countries: A Critical Review of Available Information." World Obstetrics Supplement 3:145-68. Health Statisucs Quarterly 35:52-91. Weston, Lynn. 1986. "Reducing Maternal Deaths in Developing Countries." Royston, Erica, and Alan D. Lopez. 1987. "On the Assessment of Matemal Population, Health, and Nutrition Technical Note 86-19. World Bank, Melortality." World Hlealth Statisucs Quarterly 40:214-24. Washington, D.C. Shapiro, S., M. C. McCormick, B. H. Starfield, J. P. Krischer, and D. Bross. WHO (World Health Organization). 1985. "Coverage of Matemal Care: a 1980. "Relevance of Correlates of Infant Deaths for Significant Morbidity Tabulation of Available Information." Geneva. at I YearofAge."AmencanJournalofObstemcsandGysnecology 136:363-73. . 1986. "Maternal Mortality Rates: A Tabulation of Available Infor- Sherris, J. D., K. A. London, S. H. Moore,]. M. Pile, and W. B. Watson. 1985. mation." Family Health Division, Geneva. "The Impact of Family Planning Programs on Fertility." Population Reports, J, Williams, C. D., Naomi Baumslag, and D. B. Jelliffe. 1985. Mother and Child 29. Johns Hopkins University, Population Information Program, Baltimore. Health: Delivering the Services. London: Oxford University Press. Starrs, Ann. 1987. Preventng the Tragedy of Maternal Deaths. A report on the Winikoff, Beverly. 1988. "Women's Health: An Altemative Perspective for Intemational Safe Motherhood Conference, Nairobi, Kenya, February Choosing Interventions." Studies in Family Planning 19:197-214. 1987. New York: World Bank, WHO9, UNFPA. Winikoff, Beverly, and M. A. Castle. 1987. "The Maternal Depletion Syn- Thirugnanasambandham, C.. S. Gopaul, and T. Sivakumar. 1986. "Pattem of drome: Clinical Diagnosis or Eco-demographic Condition?" Population Early Neonatal Morbidity: Observations in a Referral Matemity Hospital." Council, New York. journalof Tropical Pediarrics 32:203-5. Winikoff, Beverly, and Maureen Sullivan. 1987. "Assessing the Role of Trivedi, C., and D. Mavalankar. 1986. "Epidemiology of Low Birthweight in Family Planning in Reducing Matemal Mortality." Studies Family Planning Ahmedabad.' Indian Journal of Pediatrics 53:795-800. 18:128-43. United Nations. Variouis years. Demographic Yearbook. New York. World Development Indicators. 1988. Washington, D.C.: World Bank. Walker, G. J. A., D. E. C. Ashley, A. M. McCaw, and G. W. Bernard. 1986. Zhang, L., and H. Ding. 1988. "Analysis of the Causes of Death in China." "Matemal Mortality in Jamaica." Lancet 1:486-88. Bulletin of the World Health Organization 1988, 66:387-90. 18 Protein-Energy Malnutrition Per Pinstrup-Andersen, Susan Burger, Jean-Pierre Habicht, and Karen Peterson More than 500 million people are unable to meet their energy children and pregnant women per child life saved was esti- and protein requirements for an active and healthy life. They mated at $1,942.00 and $733.00, respectively. These estimates are almost all poor and most of them live in developing correspond to $40.00 and $23.65 per disability-adjusted life- countries, particularly in South Asia and Africa. They cope year; costs that compare favorably with the costs associated with the deficiency by reducing energy and protein expendi- with interventions in the health area and reported in other ture in growth, work, and leisure. As a result of growth falter- chapters of this book. Food supplementation is also justified on ing, the weight of about 170 million preschool children, or grounds of economic efficiency. The benefit-to-cost ratio of about one-third of all preschool children, is currently below food supplementation of preschool children was estimated to two standard deviations of normal healthy weight for children be 17.4, implying a return of $17.40 for each dollar invested in at their age; they suffer from protein-energy malnutrition (PEM) such programs after discounting for differences in the timing and related diseases. About 24 million infants are bom under- of investment and returns. Other successful single interven- weight (weighing less than 2.5 kilograms) each year, in large tions include nutrition education based on the concept of measure because ofPEM in women during pregnancy and before. social marketing in Indonesia and the Dominican Republic, About 5 million infants are unable to cope with growth and promotion of breastfeeding in several countries. faltering and low birth weight resulting from PEM and associ- Still, because PEM is frequently a result of many interacting ated diseases; they die before they reach the age of five. This factors, the effect of single interventions may be limited. amounts to about one-third of all deaths of preschool children Integrated programs that are designed to deal with the adverse in developing countries. In addition, PEM during childhood and factors identified for a particular population and for which the adulthood increases the cost of health and education and necessary national and local capacity and infrastructure for reduces school performance, labor productivity, and general program implementation exist or can be developed are likely economic growth. The estimated economic losses worldwide to be more successful than single interventions. Several such resulting from reduced labor productivity alone are $8.7 billion integrated programs, including the Tamil Nadu Integrated annually. Thus, although the contribution to child mortality Nutrition Program in India and the Iringa Program in Tanza- is very serious indeed, it is only part of the damage that PEM is nia, have dramatically reduced PEM in preschool children in doing to individuals, households, and nations. the program areas. Protein-energy malnutrition is a result of either infectious To be successful, efforts to alleviate PEM must be based on a diseases or insufficient intake of energy and protein or-most solid understanding of the environment within which the commonly-a combination of the two. Each of these two nutrition problem exists, the factors causing the problem, and immediate causes of PEM is influenced by a number of factors, feasible solutions. Such understanding is best obtained by includingsanitary conditions, the quantity and quality of water direct participation of intended beneficiaries in program de- available, access to primary health care, household behavior, sign and implementation. Furthermore, a strategy to alleviate and access to resources. Although child care, breastfeeding, PEM should consider direct and indirect interventions as well and weaning practices are of particular importance in avoiding as broader govemment policies, which, although not directed PEM among children, poverty and lack of access to appropriate at nutrition, may have a significant nutritional effect. primary health care are the overriding constraints to good nutrition at the household level. The Public Health Significance of PEM Some single interventions, including food supplementation for preschool children and pregnant women, have been suc- Three indicators are commonly used to estimate the preva- cessful in alleviating PEM in a cost-effective manner. Thus, the lence of protein-energy malnutrition: (a) protein and energy cost of perfectly targeted food supplementation for preschool intake in relation to requirements; (b) growth, weight, and 391 392 Per PiLstrp-Andersen, Susan Burger, Jean-Pierre Habicht. and Karen Peterson height in relation to established standards; and (c) birth PROTEIN AND ENERGY DEFICIENCIES. Using the cutoff point of weight. Because PEM results from either insufficient intake of 1.2 times the basal metabolic rate, which, as already men- energy and protein or infectious diseases, or most commonly a tioned, is likely to fall considerably below desirable energy combination of the two, the first indicator measures a critical intake levels, the FAO (1 987) estimated that about 335 million determinant of the nutritional status rather than nutritional people were undernourished during the period 1979-81, up status itself. An intake of energy and protein adequate to meet from 325 million ten years before. In relation to the total requirements is necessary but not sufficient to alleviate PEM. population of the developing countries, however, the percent- The combined effect of energy and protein intake and the age of those undernourished decreased from 19 percent to 15 prevalence of infectious diseases is reflected in the rate of percent during the ten-year period (table 18-1). Using the growth of children and in the weight in relation to height in same cutoff point, the Subcommittee on Nutrition of the adults. Birth weight may be used as an indicator of the nutri- Administrative Committee on Coordination (ACC/SCN) of the tional status of women during pregnancy and before. The three United Nations estimates that the number of undernourished sets of indicators mentioned above are widely used for estimat- people had increased to 360 million by 1983-85 (United ing the prevalence of PEM in populations, and they are the only Nations 1987). Updating these estimates with the 1989 pop- ones for which information is available at country and inter- ulation figures and adding a rough estimate of food poverty in national levels. China, Chen (1990) estimates that in 1989, 465 million The interpretation of the figures for energy and protein people lived in households too poor to obtain the energy intake for the purpose of estimating the prevalence of PEM is necessary for minimal activity among adults and for the further complicated by variations in energy requirements healthygrowthofchildren. A morerecentACC/SSN publication among individuals and over time and by unobserved adjust- (United Nations 1992) estimates that 786 million people were ments in energy expenditure, whether in response to intake or consuming less than 1.54 BMR by the end of the 1980s, down not. Individuals are considered undernourished when their from 976 million in the mid-1970s. energy intake is less than the minimum required to maintain As shown in table 18-1, the majority of undernourished good health and desired activity. The Food and Agriculture individuals are found in Asia-notably South Asia-although Organization (FAO) suggests that the minimum energy require- when the number of undernourished people are measured in ment for adults is 1.2 to 1.4 times the basal metabolic rate relation to total population, the problem appears to be slightly (FAO 1987).' more severe in Aftica. During the 1970s, the prevalence of Although the rate of growth is a better indicator of the undernourished people increased considerably in Africa, de- current nutritional status of children than attained weight or creased in the Near East, and remained almost constant in Asia height in relation to the genetic potential for a particular age and Latin America. No reliable data are available by region for (usually reported as weight-for-height, height-for-age, and the period since 1981. weight-for-age), the former is not usually available for popula- tions. Children whose weight-for-height is below two standard ANTHROPOMETRY-BASED ESTIMATES. The authors of a recent deviations of growth standards for a well-nourished population study for the United Nations Children's Fund (UNICEF) esti- are considered "wasted," implying malnutrition resulting from mate that 150 million children below the age of five years in recent or current deficiencies in nutrient intakes or infectious developing countries, excluding China, or 36 percent of all diseases or both. Children whose height-for-age is below two such children, are underweight, that is, below two standard standard deviations of growth standards are considered deviations of standard weight-for-age (Carlson and Wardlaw "stunted," implying malnutrition resulting from past or longer- 1990). Chen (1990) estimates that the prevalence in China is term deficiencies or infections or both. Low weight-for-age, 18 million, which, added to the number in the developing frequently the only available measure of malnutrition, may be countries, yields a total of 168 million for the world as a whole. evidence of wasting or stunting or both (WHO 1986). In 1992 the United Nations ACC/SCN estimated that this num- Table 18- 1. Undernourished Population of Developing Countries, by Region, 1969-71 and 1979-81 Total population Undernounrished populacion' Proporion of population (millions) (millions) undernourished (percent) Region 1970 1980 1969-71 1979-81 1969-71 1979-81 Africa 282 376 57 70 20 19 Far East 986 1,232 208 210 21 17 LatinAmerica 278 357 36 38 13 11 NearEast 159 210 23 16 15 8 Total 1,708 2,179 325 335 19 15 a. 1.2 times BMR for adults and adolescents. b. Ninety-eight developing market economies. Source: FAO 1987. Protein-Energy Malnutrition 393 Table 18-2. Prevalence of Malnutrition in Children under Five in Developing Countries Unde-rweighte S tuntng1 Wastingc Region Millions Percent Millions Percent Millions -Percent Africa 29 26.6 39 35.3 11 10.2 South Asia 73 45.2 66 41.3 16 9.8 Rest of AsiaC 40 43.4 43 46.2 7 8.3 Americas 8 13.8 15 27.7 1 1.3 Totale 150 35.7 163 39.0 35 8.4 a. Children more than two standard deviations below the reference median for weight-for-age. b. Children more than two standard deviations below the reference median for height-for-age. c. Children more than two standard deviations below the reference median for weight-for-height. d. Afghanistan, Bangladesh. Bhutan, India. Maldives, Nepal, Pakistan. and Sri Lanka. e. Excluding China. SoUTce: Carlson and Wardlaw 1990. ber was valid for the mid-1970s and that it had increased to continents, it appears that the prevalence of wasting among 184 million (34 percent) by the end of the 1980s. According African children below the age of five has increased from 4 to Carlson and Wardlaw, stunting is more prevalent than million to 11 million children, or from 7 to 10 percent of all underweight, whereas wasting is estimated to affect about 35 such children in the region, whereas the absolute prevalence million children, or about 8 percent of all children under the stayed constant for Asia and decreased in Latin America. In age of five years in developing countries, excluding China view of the differences between the two sets of estimates with (table 18-2). As shown in table 18-3, more than 15 percent of regard to data sources and methods, the results of such direct the underweight children are severely underweight. Severe comparisons are prone to large errors and should be inter- stunting affects more than 40 percent of all stunted children. preted as only rough indications of change in the prevalence Underweight and stunting are more prevalent in Asia than by region. in the other regions both in absolute numbers of affected The prevalence of malnutrition among preschool children individuals and in relation to all children who are malnour- is generally higher in rural than urban areas. Thus, in an ished. When measured in relation to the total population of examination of data from ten countries, Kates and others children, wasting is most prevalent in Africa. This is undoubt- (1988) found that the prevalence of underweight in rural edly a reflection of the deterioration of the African food children was 1.4 to 2 times the prevalence in their urban situation during the late 1970s and the 1980s. Comparisons counterparts. This finding was confirmed by Carlson and between the above figures and estimates reported by the FAO Wardlaw (1990), who found that the prevalence of under- (1987) in the Fifth World Food Survey support the notion of weight in preschool children was higher in rural than in urban deterioration in the nutritional situation in Africa. More re- areas of all thirty-one countries studied. On the average, the cent developments including widespread famine indicate fur- rural prevalence was 1.6 times the urban prevalence. ther deteriorations. Thus, based on a direct comparison of The average rural prevalence of stunting was 1.5 times the urban prevalence, and all but one country shows a lower prevalence in urban than in rural areas. Wasting showed a Table 18-3. Prevalence of Severe and Moderate different pattem, with almost one-third of the thirty-one coun- Malnutrition tries having a higher prevalence in urban areas. On the aver- (percent) age, the rural prevalence exceeded the urban prevalence but Severe only by a factor of 1.2. Although a firm interpretation of this malnutrition finding must await further research, it may be hypothesized as percentage that the smaller difference between urban and rural areas with Indicatorof5 b of severe and respect to wasting, which indicates more recent nutrition ma-nuri-on-Sevee Modrate~ Tocal moderate insults, than with respect to stunting is an indication of a shift Underweight 5.6 30.7 36.3 15.4 of the nutrition problem from rural to urban areas. The shift is Stunting 14.5 20.9 35.4 41.0 due to rural-to-urban migration and deterioration in urban Wasting 0.8 5.4 6.2 13.4 living standards, including reduced real wages, higher food a. Underweight refers to low weight-for-age. Stunting refers to low prices, reduced government expenditure on primary health height-for-age. Wasting refers to low weight-for-height. care, and increased unemployment, resulting from the eco- b. More than three standard deviations below the median for the refer- nomic crises and associated macroeconomic policy reforms ence population. c. Two to three standard deviations below the median for che reference currently undertaken in most developing countries. population. Children are generally exposed to the highest risk of PFM and Source: Carlson and Wardlaw 1990. associated mortality during and immediately following the 394 Per Pinst rup-Andersen, Susan Burger, Jean-Pierre Habicht, and Karen Peterson weaning period, that is, between six months and twenty-four MORTALITY AND MORBIDITY. According to estimates made by months of age. It is during this period that the prevalence of INICEF (1988), PEIM is a contributing cause of the deaths of underweight and wasting is the highest. Beyond the age of two about one-third of the 15 million children that die annually. years, the prevalence of wasting usually drops off sharply and In view of the complex interactions between PEM and diseases the prevalence of stunting is maintained, reflecting a lack of that play a role in child mortality, the effect of PEM is difficult catch-up growth and a reduced risk of acute malnutrition to isolate. The relation between low birth weight and child beyond the age of two years. These relationships are illustrated mortality is well documented (Ashworth and Feachem 1985; in table 18-4. Herrera 1985; Rasmussen and others 1985; Kramer 1987). The The prevalence of low birth weight is an indicator of mal- pattern of risk is associated with the length of gestation and nutrition among women during pregnancy and before. Chen size at birth. Mortality in the first year of life appears to be (1990) estimates that about 24 million infants are born under- greater for premature infants than for those who are full-term weight (defined as weighing less than 2.5 kilograms) every but suffer intrauterine growth retardation (LOGR). The mortal- year.Thisfigurecorrespondsto 16percentofallbirthsglobally. ity risk for children age one to four years, however, is greater The problem is most prevalent in Asia, where an estimated 29 for infants with IUGR (Ashworth and Feachem 1985). In devel- percent of all infants are born underweight as compared with oping countries, as much as 80 percent of low birth weights are 16 percent in Sub-Saharan Africa, 1 1 percent in Latin Amer- due to chronic ILGR, rather than premature birth (Villar and ica and the Caribbean, and 7 percent in the United States Belizan 1982; Villar and others 1986; WHO 1986; Puffer and (McGuire 1988). Serrano 1987). The growth potential of infants with chronic IUGR is restricted because of the severity and length of in utero Functional Consequences deficits (Villar and others 1984, 1986). Among low-birth- weight infants who are premature or suffered acute IUCR, catch- The large magnitude of malnutrition in developing countries up growth is rapid in the first six to eight months of life (Davies has serious functional consequences for the affected individu- and others 1979; Villar and others 1984; Peterson and Frank als and their families as well as economic consequences for 1987) but only if nutritional supplies are adequate. individuals, households, and nations. Although the ultimate The attainable birth weight of infants of women who are consequence of malnutrition is death, increased mortality is stunted as a result of childhood malnutrition is constrained, only one of many reasons why malnutrition is undesirable and and thus the lifelong cycle of chronic growth deficits is perpet- costly. Protein-energy malnutrition in children inhibits their uated (Herrera 1985). In developing countries, maternal growth, increases their risk of morbidity, affects their cognitive height makes a significant independent contribution to birth development, and reduces their subsequent school perfor- weight, an effect mediated through IUGR, not prematLrity mance and labor productivity. In women during pregnancy (Kramer 1987). The coefficient for maternal height indepen- and before, PEM contributes to morbidity and mortality and to dent of weight on birth weight is approximately 7.8 to 10 grams low birth weight of their infants, which in turn increases the per centimeter of mother's height (Thompson, Billewicz, and risk of malnutrition and mortality in infants. Such increased Hytten 1968; Tanner and Thompson 1970; Kramer 1987). In risk is also associated with PEM in women during lactation. longitudinal studies of children in developing countries, ma- Work capacity and labor productivity in adults are negatively temal height predicted both height-for-age and weight-for-age affected by PEM during childhood and adulthood, and the in the second and third years of life, independent of other economic consequences of PEM are the sum of these productiv- biological and social risk factors (Mata 1978; Balderston and ity losses, additional costs of health and other social programs others 1981; Kielmann and others 1983). necessitated by malnutrition, lower school performance, In epidemiological studies of the short-term mortality risk higher educational costs, and lost productivity of care givers among hospitalized children, the proportion of deaths was because of child malnutrition. significantly greater among severely malnourished children than among those who were moderately malnourished (Gomez and others 1956; McLaren and others 1969). Prospective com- Table 18-4. Prevalence of Malnutrition by Age munity-based studies (Kielmann and McCord 1978; Chen and (percent) others 1980; Kasongo Project Team 1983; Bairagi and others Age (years) Underweight Stunting Wasting 1985; Lindskog and others 1988; Yambi 1988) have generally Less than one 14 7 18.Q 4.8 demonstrated an increase in risk of death at weights-for-age One 30.4 35.3 9l4 iess than 70 to 80 percent of the reference median (Kielmann Two 26.6 33.5 5.1 and McCord 1978) and a sharp increase in mortality risk at Three 24.1 34.5 3.4 weights-for-age less than 60 percent of the median, weights- Four 23.2 35.3 3.6 for-height less than 70 percent of the median, and heights-for- ------- --- age less than 85 percent of the median (Kielmann and McCord Note: For thirty-nine countries. Underweight refers to low weight-for- age; stuniting refers to low heighr-for-age; wastinig refers toi low weight-for- 1978; Chen and others 1980; Lindskog and others 1988; Yambi height. 1988). Children who were both severely wasted and stunted Source: Carlson and Wardlaw 1990. were at the greatest risk of death (Chen and others 1980). Protein-Energy Malnutrition 395 Growth faltering(Bairagi and others 1985) and arm circum- preschool children who were adopted into American homes, ferences less than 75 percent of median (Sommer and the mean IQ score at school age exceeded the average score for Lowenstein 1975) also predicted increased mortality risk. Mor- American children and was forty points higher than that tality rates and the predictive ability of nutrition indicators reported for children from similar populations who were re- varied by age (Kielmann and McCord 1978; Kasongo Project turned to their home environment(Winick and others 1975). Team 1983) and were greater among children of lower socio- Several studies have shown that children in industrial coun- economic status and born toshorter and lighter mothers (Chen tries who suffered early malnutrition secondary to organic and others 1980; Chowdhury 1988). A recent synthesis of illness, independent of socioeconomic deprivation, did not studies of the relation between child anthropometry and mor- exhibit low developmental or intelligence test scores at later tality indicates that the relationship holds even for mild to ages (Pollitt 1987). Interventions that provided food alone to moderate malnutrition (Pelletier 1991a). children at risk of PEMi resulted in mild to insignificant im- Numerous studies have shown an association between mal- provements on cognitive development. In contrast, those in- nutrition and infectious disease, especially diarrhea, among terventions that combined health care and educational or children in developing countries. Only a limited number of the psychosocial stimulation with nutritional supplementation ap- researchers, however, used longitudinal analyses to examine peared to have a significant effect on cognitive development nutritional status as a predictor ofdiarrhea and infections; even (McKay and Sinisterra 1974; Cremer and others 1977; fewer controlled for other factors associated with PEM that may McKay and others 1978; Grantham-McGregor and others have contributed to infections. 1979; Grantham-McGregor, Stewart, and Schofield 1980; Among studies in which measurements of diarrhea inci- Grantham-McGregor, Schofield, and Harris 1983). dence and duration over intervals of two to three months were included, malnutrition predicted either increased duration SCHOOL PERFORMANCE. Sensory deprivation because of PEM (Tompkins 1981; Black, Brown, and Becker 1984) or inci- in preschool children results in impaired leaming abilities, dence (El Samani and others 1988; Sep6lveda, Willett, and adverse behavior, poor school attendance, and grade repeti- Mufioz 1988) ofdiarrhea. This relationship remained constant tion. On the basis of a review of more than fifteen studies when previous diarrhea and sociodemographic variables were undertaken in various developing countries and the United controlled in the analysis (Ghai and Jaiswal 1970; Sepulveda, States, McGuire and Austin (1987) conclude that "better Willett, and Mufioz 1988). Anthropometric deficits did not growth is associated with better preschool and school-age I.Q. predict incidence of diarrhea among preschool children in two It is also associated with learning-relevant behavior, early studies in which incidence over intervals of nine months or enrollment in school and better school achievement, all of longer were measured (Chen, lIluq, and Huffman 1981; Peter- wlhich enhance the educational efficiency of and economic son and others 1988b). Thus, the evidence suggests that mal- return on primary schools." Similarly, the ACC/SCN concludes nutrition in preschool children in developing countries that "Malnutrition and infection during the preschool period, predicts short-term risk of increased incidence and duration of interacting with environmental factors related to poverty, are diarrhea, but prediction of long-term risk has not yet been critical determinants of later school performance" (United demonstrated. In addition, the evidence for a causal role of Nations 1990). child malnutritioni in other infectious diseases, including mea- sles, malaria, and acute respiratory infections merits longitudi- LABOR PRODLCTIVITY. Evidence from small controlled exper- nal analysis. iments show a clear causal relation between body size, aerobic capacity, and physical work capacity (Spurr, and others 1977; COGNITIVE DEVELOPMENT. Growth retardation affects the Spurr 1983, 1984). Because the strenuous work undertaken by development of motor and mental functions. Severe PEM affects a large share of the labor force in developing countries requires brain growth, attention span, and short-term memory as well a great deal of physical energy, it seems reasonable to assume as activity level, which in turn affects interaction with the care that the link between body size and work capacity would be giver and the environment (McGuire and Austin 1987). Be- reflected in a strong link between body size and labor produc- cause of the interaction between sensory environment and tivity. Similarly, a strong link between energy intake of adults PEMI-induced growth failure, Pollitt (1987) suggests thatcogni- and their labor productivity would be expected. Evidence of tive development is best understood as the outcome of the both links have been found in a number of studies as reviewed social environment and health history and the interaction of by Martorell and Arroyave (1984), Larham (1985), Strauss these variables with the nutritional status of children. It ap- (1985), Agarwal and others (1987), McGuire and Austin pears that psychosocial stimulation often can override or min- (1987), and Berg (1987). Methodological flaws in the results imize the effects on the cognitive potential resulting from PENM. of many of the studies reviewed, however, make the interpre- In developing countries, early malnutrition has been associ- tation of the findings difficult (Strauss 1985). ated with later developmental delays among children (Pollitt Four recent studies appear not to suffer from serious meth- 1987); but most of the variability in intelligence quotients odological flaws. Strauss (1986), in a study of a sample of (IQs) was explained by differences in socioeconomic status farmers in Sierre Leone, and Sahn and Alderman (1988), in a (Richardson 1980). Among severely malnourished Korean study of Sri Lankan households, found a significant relation 390 Per PiiornpL)-AiL-rseii Szu.vam IrgBr, aemu-lPwne Habichi, aitrl) Krei lPetersin between household-level energy conisumption and labor pro- REM through higher costs of or lower benefits from edlucation, ductivity. Neither of the two Studies had diti available on the higher health costs, higher costs of social programs, and in- height and weight-f6r-height ofthe workersand their inlivid- come iosses as a result of care giving in households with uial energ coLnsullptioll. Thus, the effects of nutritional status malnourished members. No quantitative estimates are avail- in childhood, past energy consumUption during adulthood, an:d able for these costs. cLirreit energy consumption cannot be separated. Such sepa- ration was done by Deolalikar ( 1 988) in a study7 of South Causes of PEM Indian households. He foutndl that labor productivity as m'1ea- sured by attained wages anid farmi output was highly responsive Protein-energy mn111lutrition is caused hy InfectiLous diseases to weight-for-height of the worker. No such responsie was an1d isufticient intake of energy and proteini to meet require- fOUnd with respect to heiglt and contemp rary energy intake. inents (figure 1 8-1 ). The effect of each of the two is a functioni Thus, although m)ore research is needed, it may he hyp bthe- of the sta,te of the other, and. mt)st intdividuals who suffer from sized that weight changes providle a buffer between changes in tEN do sn because of exposure to both. The prevalence and entrgy intake andl labor productivity. The imimedia;te effect of severity of infectious diseases are influenced by sanitary con- reduced energy intake is weight b ss which if con tinued will ditions., quality anl quanitity of water available, access to subsequen1tlV inltuenice labor productiVitvy primalrv health care, behavior of hoLiseholds and individuals, Haddad and Bouts ( 1990) found a strong effect of childlioodi energy and protei n intake, and, in the case of childreni, by child nutritional status-as represented by height-on the labor cTare, breastfeeding, and weaninig practices. Energy and proteill productivity of Plhilippin-e agricultural workers. The elasticity intake i. influeticed by household access to food, food acquisi- of heighit on prod( uctiVity-as measured by wage-w:IS esti- tion and allocation behavior, iiitectious diseases, and the three miated to he 1 .8, implying that a difference of 1 percenit in the child-related factors mentioned above. Access to food, sanita- heigtht of adult workers is associated with a 1 .38 lercent ditfterence in their wages. The fintdings from the tour studiles are powerful because they Figure 18 1. Principal Factors Affecting PEM perii t an estimation of tlhe economnic losses associated witth malnutrition. Taken with estimnates of cost-effectivelness of Household resources various programs, they make it possible to estimate the net (assets, income, time), econiomric gains associated wirtl efforts to alleviate malnutti-1 food and non-food tion. A preliminary estimazte of this nature is made in the next| prices and availability sectioll. Still, more research is needed to verifv these findings and to develop empirical evidence tor various settings to\ strengtilen the validity of suchi estimates. Household Access to sanitation, strengthen ~~~~~~~~~~~~~~~~access to food water, and primary EtONOMIv ((1TS. (O1n the basis of' the Haddad and BoUiis I I I estimates discussed earlier, a rough preliminiary estimate can be maide of the economic losses resulting fron) forgone la,b r productivity dlue to I'EM. Accorditng to tables 18-2 and 18-3, Knowledge and behavior stunting affects 163 million preschool children, ot which 41 percent are severely stutnted. After the age ofabout twenty-two monIths, severely stunted children are 1O centimeters shiorter l than the median, whereas moderately stunted childreni are 7 centiriimeters shlorter (wi ii 1979). Stuntirng during chilidho od Energy and protein 1 Infectious diseases appears to translate to equal height deficiencies in adulthood (Martorell, in press). Assumiiiing an average daily wageof $2.50 and 300 working days per year for the workers exposed to Stunatitng in childhood and an average height of 160 centitme- ters (the average height of the Philippinie sample on the basis Child care, of which Haddad and Bouis estimazted the above-mentionted w weaning practices relation between height and prOductivity), we find that the total economiiic loss due to stunting is $8.7 hillion annually, tr about one-fourtl oftthe total health expenidIture of developing countries (see appendix 1SA for the methodology used in the calcUlation). In additioni to the economic costs associated with nutritioni- related mnortality and productivity losses, household incomes and national economtnic growth are negatively influenced by Source Authors. Protein-Energy Malnutrition 397 tion, water, primary health care, and knowledge is in turn incidence. Caloric intake (Balderston and others 1981) and influenced by household resources, that is, assets, income, and weaning practices (Balderston and others 1981; Mata 1978) time, as well as the prices and availability of food and nonfood were as important as or more important than diarrhea inci- goods and services, including health care services at the com- dence in multivariate models that explained variance in munity level. weights and heights of children from twelve months to thirty- six months of age. New findings from data collected in the Infectious Diseases 1970s in Guatemala and Bogota indicate that the pemicious effect of diarrhea on growth is not seen if children are better fed The effect of infection, especially diarrhea, on the nutritional during periods of diarrhea (Martorell, Rivera, and Lutter 1988; status of poor children in developing countries has been well Lutter and others 1989; Rivera, Martorell, and Lutter 1989). documented (Beisel 1977). Mechanisms by which diarrhea and other infections cause malnutrition include decreased food Energy and Protein Intake intake resulting from anorexia or food withholding, decreased nutrient absorption, increased metabolic requirements, and Insufficient energy and protein consumption to meet desired direct nutrient loss (Chen 1983;Keusch and Scrimshaw 1986). expenditure results in reduced rate of growth, weight loss, Epidemiological studies have demonstrated a negative rela- reduced level of activity, or a combination of these. In chil- tionship between infections-above all, measles, diarrhea, and dren, the rate of growth is sensitive to both deficiency in intake malaria-and concurrent child growth in Latin America and infection. Because of the synergistic relationship between (Mataand others 1972,1977; Martorell and others 1975; Mata food intake and infectious diseases, the expected growth re- 1978), Africa (Rowland and others 1977, 1988; White- sponse to improvements in one ofthe two may not be observed. head 1977) and Asia (Koster and others 1981; Kielmann and Recent multivariate analyses have separated the effects of the others 1983). two and shown strong growth responses to one or the other or Evidence for the effect of respiratory infections on nutri- both, depending on the context (Kennedy 1989; Von Braun, tional status is equivocal. No association has been reported by Puetz, and Webb 1989). several authors (Rowland and others 1977; Whitehead 1977; Martorell and others 1975, 1983), although respiratory illness Child Care, Breastfeeding, and Weaning Practices had as strong an effect as diarrhea on weight loss in Brazilian children (Leslie 1982). A recent longitudinal study (Rowland On the basis of a review of available literature, Huffman and and others 1988) showed that lower respiratory tract infections Steel (in press) conclude that "few interventions have been were significantly associated with rate of weight gain in Gam- shown to be as effective in preventing diarrhea among infants bian infants. After adj usting for the prevalence of d isease, such as breastfeeding." In view of the strong relation between diar- infections accounted for one-quarter and diarrhea for one- rhea and PEM, this conclusion has obvious nutritional im- half of the shortfall in attained weight compared with refer- plications. Community- and hospital-based studies in a ence standards. number of countries, including Brazil (Victora and others Intervention trials reviewed by Stephenson (1987) have 1987), Peru (Brown and others 1988), Indonesia (Lambert pointed to the contribution by anumberofparasitic infections 1988), India (Anand 1981), Costa Rica (Mata 1983), and the to childhood growth retardation. With chemotherapy, chil- Philippines (Clavano 1982), all show that breastfeeding re- dren with parasitic infections showed small improvements in duces the occurrence of diarrhea. rate of weight gain, weight-for-height, and subcutaneous fat As discussed earlier, children are exposed to the highest risk stores (Stephenson 1987). Other infections have not been of malnutrition and growth faltering during the weaning pe- shown to affect growth significantly, including infectious fe- riod from the age of six months to twenty-four months. Wean- vers (pertussis, chicken pox), skin and eye infections, and deep ing foods commonly used are inadequate supplements to infectionsrequiringantibiotic treatment (Martorellandothers nutrients in breast milk and a source of contamination, con- 1975; Rowland and others 1977; Whitehead 1977). Sample tributing to an increased incidence of infectious disease sizes for some of these studies were too small to pick up an (Scrimshaw and others 1968; Wyon and Gordon 1971; Barrell effect, however, except for skin and eye infections. and Rowland 1979; Black and others 1982). Thus, where fecal The relative effect of infections and inadequate energy and contamination of the environment is widespread, prolonged protein intake on the growth of children in developing coun- full and partial breastfeeding is particularly important tries varies with the prevalence of infection (Martorell and (Habicht and others 1988). In Brazil, delaying the introduc- Yarbrough 1983; Keusch and Scrimshaw 1986). In prospective tion of solids promoted better growth in the first five months studies, diarrhea incidence predicted height-for-age during among the urban poor who had limited access to flush toilets, periods of peak incidence, in the second and third years of life piped water, and refrigeration, presumably through reduced (Mata 1978; Balderston and others 1981; Kielmann and others incidence of infectious disease. A recent study of infant feeding 1983; Lutter and others 1989), but it did not explain variance practices in a region of the Republic of Yemen that had a high in the weight or height of children studied before (Peterson prevalence of both acute and chronic PEM showed that and others 1988a) or after (Leslie 1982) the period of peak breastfeeding had the strongest beneficial effect on weight-for- 398 Per Pinstrup-Andersen, Susan Burger, Jean-PzeTTe Habicht, and Karen Peterson length and weight-for-age of infants age three months to six affect PEM by reducing the time needed to acquire the water- months (Jumaan and others 1989). The introduction of other usually a task performed by women and children-and by foods was positively associated with weight-for-length only making more water available for household production of food among children twelve to twenty-three months of age. (Burger and Esrey, in press). Child care is also of paramount importance during this Recent reviews (Esrey, Feachem, and Hughes 1985; Esrey period and beyond. Matemal child-rearing behaviors and and Habicht 1985, 1986; Esrey and others 1990; and Burger attitudes, mother-child interaction, family social relation- and Esrey, in press) found that improved water and sanitation ships, and stress have been associated with poor nutritional are associated with decreased diarrheal diseases, improved status in both industrial and developing countries in theoreti- nutritional status, and lower childhood mortality. The effect cal models (Williams 1962; Klein and others 1972; Caldwell of improved water and sanitation on child morbidity and 1974; Mata 1978; Herrera and others 1980; Rathbun and growth depends on the existing environmental conditions and Peterson 1987). Maternal interaction with both family- and the presence or absence of other related factors and programs community-based networks has been associated with child that influence exposure to pathogens (Burger and Esrey, in growth outcomes in cross-sectional and longitudinal (Peterson press). For example, in Malaysia the introduction of flush and others 1988a) studies, suggesting that social support for the toilets and piped water was found to have a greater effect on mother may be an important aspect of caretaking capacity and the mortality of nonbreastfed than breastfed infants, because resourcefulness. These and related issues are further analyzed breastfeeding reduced the exposure to pathogens (Habicht and by Engle (in press). others 1988). Similar interactions were found between im- proved water and sanitation and income and educational Knowledge and Behavior levels in several countries (Burger and Esrey, in press). On the basis of a review of past interventions, Esrey and Habicht There is a strong association between maternal education and (1986) conclude that water quantity has a greater effect on child nutrition and mortality, although the specific mecha- child morbidity than water quality in contaminated environ- nisms through which the association operates is not fully ments and that improved water quality may have little ef- understood (Leslie, in press). On the basis of a comprehensive fect unless most other important routes of contamination are review of available evidence, McGuire and Popkin (1990) eliminated. conclude that the nutrition effects of maternal education are Evidence from several countries indicates that improved mediated through better management of household resources, access to water increases the amount of time women spend on greater use of available health care services, health behavior food production, processing, and preparation (Burger and that compensates for a lack of such services, lower fertility, and Esrey, in press). Time saved in water collection may also be more child-centered care-giving behavior. used in improved child care, income-generating activities, and other nutrition-related activities. Maternal Health and Reproductive History Household Access to Food Inadequate food availability to meet energy demands may adversely affect the health and nutritional status of mothers as Household access to food is closely related to food consump- well as that of their offspring. Matemal stunting due to early tion by individual household members, but intrahousehold malnutrition and concurrent energy deficits can limit maternal food allocation does not necessarily parallel relative needs. work productivity and birth length and weight of the infant Pattems of intrahousehold food distribution have been re- (Kramer 1987). Evidence suggests that the repeated demands viewed by several authors (Den Hartog 1972; Van Esterik of many pregnancies and close child spacing may deplete 1984; Haaga and Mason 1987). Findings of preferential distri- matemal nutritional status (Hamilton and others 1984; Mer- bution of family food by age are equivocal and difficult to chant and others 1988), although further studies are needed interpret because of differences in the outcome measures and on this topic. A large number of young children in the home the recommended dietary allowance schedule used (Haaga and may also adversely affect the nutritional status of the individ- Mason 1987). ual child (Kielmann and others 1983), independently of The authors of some studies have shown that the adequacy family socioeconomic status and maternal health and of energy intake was greater in adults than for preschool reproductivity history (Peterson and others 1988a), presum- children (Aligaen and Florencio 1980; Nutrition Economics ably because maternal time for child care and feeding must be Group 1982; Pinstrup-Andersen and Garcia 1990), whereas shared by many. others demonstrated that children received a greater propor- tion of their recommended dietary allowance than adults (Nu- Access to Sanitation and Water trition Economics Group 1982; Van Esterik 1984), especially if total intake from both meals and snacks was computed Improved water and sanitation may affect PEM through a reduc- (Harbert and Scandizzo 1982). The adequacy of the energy tion in the transmission of pathogens, which in turn reduces intake of children in relation to that of adults is influenced by diarrheal diseases. In addition, improved access to water may family income and seasonal effects, including food shortages, Protein-Energy Malnutrition 399 lack of maternal time for child care and feeding, and the need influence intrahousehold allocation of food (Van Esterik for food for agricultural workers (Van Esterik 1984). Other 1984). feeding practices that may influence PFM in children include The effect of higher incomes on nutrition may be insignifi- withholding solids from them during diarrhea (Taylor and cant in the short run in areas in which poor sanitation and lack Taylor 1976) and feeding them staples that have low energy of knowledge and primary health care are the most limiting density. constraints to good nutrition. In the longer run, however, A sex bias in food distribution that favors males has been higher incomes are likely to alleviate these constraints. documented in the Middle East and South Asia (Van Esterik 1984; FAO 1987; Haaga and Mason 1987), whereas nosuch bias Interventions for Control of PEM was found in Africa (Svedberg 1989; Von Braun, Puetz, and Webb 1989). The sex bias in South Asia is particularly impor- The factors that influence the two basic causes of PEM (inade- tant and deserves serious attention because of the high rates of quate intake of energy and protein and infectious diseases) may malnutrition in that region. Differences in weaning and access be manipulated by various interventions. In table 18-5 we list to food (Van Esterik 1984) may be reflected as well in patterns the most important factors, interventions, and related activi- of malnutrition and mortality (DeSweemer 1974; Chen and ties. Only the interventions related to intake will be discussed others 1980; Chen, Huq, and D'Souza 1981). A sex bias in food in this chapter. distribution is usually related to other factors, including age, lactation and pregnancy, birth order and sex of child in rela- Single Interventions to Improve Dietary Intake tion to siblings, matemal education, and family income (Ab- dullah 1983; Van Esterik 1984; Haaga and Mason 1987). Single interventions reviewed here include food supplementa- Patterns of sex bias in food distribution and in nutritional and tion of children and women as well as food price subsidies, health status and survival may also be related to regional income transfers, food fortification, and other broad programs ecology, which determines the agricultural economy and the and policies. demand and perceived value of male as opposed to female labor (MacCormack 1988). Societies with an ideology of matrilineal FOOD SUPPLEMENTATION. Food supplementation schemes are descent (and greater investments in health and nutrition of usually not designed in such a way as to permit credible women) seem to be found in high rainfall areas, which require evaluation of effect. Therefore, results from evaluation at- labor-intensive hoe cultivation, whereas patrilineal descent is tempts such as those reported by Beaton and Ghassemi (1982) traditionally seen in dry-land plow regions of continents are difficult to interpret and do not permit inference regarding (MacCormack 1988). the biological effects per unit of food supplement. The synthe- sis undertaken here is limited to food-supplementation Household Resources and Prices schemes designed in such a way as to permit a reliable estima- tion of effect. Four food-supplementation schemes for preg- Poverty is the most important determinant of PENM because it nant women in Colombia, the Gambia, Guatemala, and constrainshousehold access tofood, knowledge, sanitary living Taiwan (China) permitted such estimation (Lechtig and oth- conditions, safe water, and appropriate health care (figure ers 1975; Mora, Clement, Christiansen, Suescun, Wagner, and 18-1). Increasing household income among the poor results in Herrera 1978; Mora, de Navarro, Clement, Wagner, de expanded energy and protein consumption, stronger demand Paredes, and Herrera 1978; Lechtig and Klein 1979; Mora and for education and health care, and improved living conditions others 1979; Herrera and others 1980; McDonald and others (Pinstrup-Andersen 1985; Garcia 1988), although the effect 1981; Delgado, Martorell, Brineman, and Klein 1982; Del- of increases of income on energy consumption may be less than gado, Valverde, Martorell, and Klein 1982; Overholt and previously expected (Behrman 1988). Similarly, households others 1982; Mora 1983; Adair, Pollitt, and Mueller 1983, respond to food price changes by adjusting household food 1984; Prentice and others 1983; Adair and Pollitt 1985; Pren- consumption (Pinstrup-Andersen 1985). tice and others 1987). The effect on dietary intake of supple- As discussed above, changes in household food consump- mentary food during pregnancy (detailed in table 18-6) was tion influence that of high-risk household members, which in studied in relation to maternal weight gain, maternal activity, turn influences their nutritional status. Thus, evidence of maternal body mass, and birth weight! causal effects for each of these steps implies a causal effect Recent studies (Merchant, Martorell, and Haas 1990a, between income and nutritional status. Several researchers, 1990b) of the nutritional stress of simultaneous pregnancy and including Garcia (1988) in the Philippines, Alderman (1990) lactation in poorly fed mothers show that mothers appear to in Ghana, and Sahn (1990) in C6te d'lvoire, found a strong absorb much of the deficit in energy by mobilizing their fat association between income changes and the nutritional status stores, and birth weight is little affected. Thus one might of preschool children. In addition to the importance of the expect mothers more than their fetuses to benefit from the food absolute level of incomes and prices, fluctuations in incomes supplementation. This was not the case. These studies showed and prices, including seasonality, are important determinants little effect of supplemental food on matemal weight gain but of PEM in many rural areas (Payne 1985; Sahn 1989) and may did show a substantial improvement in mean birth weight, 400 Per Pinstrup-Andersen, Susan Burger, Jean-Pierre Habicht, and Karen Peterson Table 18-5. Causes and Interventions for Malnutrition Causes Contributingfactors Interenntions Activities Inadequate intake of calories, Food (quantity and quality) Resource transfer: supplementary Food delivery to target individual protein, amino acids foods or to family Feeding programs Nutrition rehabilitation Subsidies, transfer programs Price subsidies Food stamps New foods Weaning food preparation: processed and home preparation Fortification Processing, marketing, and distribution Incentives (pricing) Agricultural production Prices, incentives Income Income generation Income generation projects Knowledge Nutrition education: information Face-to-face (individuals and transfer, behavior change groups) Mass media Infant feeding Promotion of breast-feeding Training of health care personnel and mothers Restrictive legislation Time Provision of child care Support groups Day care centers Accessible cooking and water Better stoves facilities Access to fuel Water projects Nutrient losses, metabolic Knowledge Hygiene education Face-to-face increases, and anorexia Mass media Water Improved quality Water projects, wells, pumps Improved quantity Feces Latrines Feces disposal Drugs Drug administration Distribution Prices Drug education Training of health care personnel, individuals Vaccines Immunization Injections Distribution of vaccines Dehydration Rehydration therapy: package Intravenous, nasogastric, oral or home remedy (ORT) Medical care Medical care Medical care Source: Authors. which was greater the more malnourished the pregnant woman intake ranged from a high of 34 grams per 10,000 kilocalories (table 18-7). Neither weight gain (Adair, Pollitt, and Mueller in Colombia to a low of 14 grams and 8 grams per 10,000 1983; Adair and Pollitt 1985; Prentice and others 1987) nor kilocalories, for males and females, respectively, in Taiwan measured activity pattems during pregnancy (Roberts 1982; (China). Lawrence 1988) changed markedly in response to supplemen- Because of the relation between infant mortality and birth tation. Standardizing for the increase in total net caloric intake weight, improving the caloric intake of malnourished pregnant attributable to the supplement over the course of pregnancy women through food supplementation can result in a lower (derived from values in table 18-6), we find that the im- infant mortality rate. The difference in incidence of low birth provements in birth weight per unit of increase in energy weight attributable to the increase in energy intake reported Table 18-6. Supplementation during Pregnancy Maternal anthropometrr Materal intae Supplement distributed Net intake Period of Weight Height Energ'Q Protein Energy Protein EnergY Protein Location supplementation Treatment group (kg) (cm) (kcal/daY) (g/day) (kcal/day) (glday) (khal/day) (gIday) BogotA, Colombia Third trimester Supplement - 150 5 1,646 + 630 37 23a 856 38 133 20 No supplement - 150 5 1,606 665a 37 20a 0 0 33 2 Low maternal weight-for-height ad id b, Supplement - - 1,589 632'd 35 21' 856 38 334b 295 Nosupplement - - 1,589632' 3521'd 0 0 29 1,e Ibj Full-term Supplement - - 18623+635 36t 19 o56 38 150 9h No supplemient - - 1,621 655 37 23a 0 0 48 1I Taiwan (China) PTevious lactation Supplement A 50 5 155 + 5 1,121 37 800 40 1,650h 64h andpregnancy SupplementB 495 1545 1,197' 37Q 80 6 1,206 37 Guatemala Entire pregnancy Atol6 48d 149+5 1,473 467g 4213' 163' 11.5' 1,580 48 Fresco 48d 149 5 1,411 467' 42 13g 59' 0 1,492 42 41 ~ ~ ~~~~~~~~~~~~~~~~~~~~~~~~d d 2 Keneha, The Gambia From sixteenth week Postsupplement 50d 158d - - 1,30( 49 1,898 - Presupplement 50 1558 1,467 - 0 0 1,467 Wet season Postsupplement 50d 1581 - 1,50d 56 1,838 Presupplement 50 158 1,418 - 0 0 1,418 -Data not available. a. In sixth month of pregnancy. b. Difference between post- and presuippleinentation periods. c. Less than 0.36 kg/cm. d. For combiised groiups. e. Sigisificant difference betweeni groups; P < 0.005. f Significant difference between grouIps; P < 0.001. g. Home dietarv intake; no iniitial dietary intake values reported. h. Significant difference between groups; P < 0.01. i. Content per 180 ml; offered ad libitum. j. Maximum. Source: Lechtig and others 1975; Mora, Clement, Christiansen, SuesCin, Wagner, and Herrera 1978; Mcira, de Navarro. Clement, Wagner, de Paredes, and Herrera 1978; Lechtig and Klein 1979; Mora and others 1979; Herrera and others 1980; McDonald and others 1981; Delgado, Martorell, Brinemilan, and Klein 1982; Delgado, Valverde, Martorell, Brineman, and Klein 1982; Overholt and others 1982; Mora 1983; Adair, Pollirr, and Mueller 1983, 1984; Prentice and orhers 1983; Adair and Pollitt 1985; Pretitce and others 1987. 402 Per Pmnstup-Andersen. Suisan Burger, jean-Pierre Habicht, and Karen Peterson Table 18-7. Supplementation during Pregnancy: Outcome Mean birth weight Incidence of LBW Maternal weight Mean birth per intake Location Treatment group (percent) gain (kg) weight (gl l04 kcal)b Bogota, Colombia Supplement 8.7 - 2,978 377 34.1 No supplement 11.0 - 2,927 392 n.a. Low matemal weight-for-height Supplement - - 3,014 379C 55.4 No supplement - - 2,833 412c n.a. Full-term d Supplement - 4.2 1.8d 3.003 354f 35 4 d,e No supplement - 3.5 1.7 2,940 318 n.a. Taiwan (China) Males Supplement A - 7.5 2.8' 3,216 348 13.8 Supplement B - 7.8 2.5g 3,161 364 n.a. Females Supplement A - 7.5 2.8g 3,013 411 8.0 Supplement B - 7.8 2.5g 2,981+ 321 n.a. Guatemala Aroeh - - 3,077 334 21.0 Fresco - - 3,027 461 n.a. Keneba, The Gambia Postsupplement 8 O - 2,891 271,k 16.0 Presupplement 18.4 - 2,882 37, n.a. Wet season Postsupplement 7.5 - 3,031 32 28.8 Presupplement 23.7 2,842 48e, n.a. - Data not available. n.a. Not applicable. a. Birth weight less than 2,500 g. b. Calcuilated from the net intake per day times an estimnate of the total number of days supplemented during pregnancy. In Guatemala, women did not start receiving supplement until they recognized that thev were pregnant, so birth weight per input is likely to be a slight underestimate. c. Significant difference between groups; P < 0.01. d. Third trimester. e. Significant difference between groups: P < o.Xs5. f. Significant difference between groups; P < 0.001. g. Total groLIp. h. Atole is a food supplement high in energy and protein. i Fresco, which has a low content of energy and no protein, was used as a control. . Adjusted for sex, parity, and gestational age. k. Significant difference between groups; P < 0.05. I. Significant difference between groups; P < 0.002. Source; Mora and others 1979; Herrera and others 1980; McDonald and others 1981; Delgado, Martorell, Brineman, and Klein 1982; Mora 1983; Adair, Pollitt, and Mueller 1984; Prentice and others 1983; Adair anid Pollitt 1985; Prentice and others 1987. in table 18-6 varied from 2.3 percentage points in Colombia, height was low and whose diet was supplemented was 181 where the incidence of low birth weight in the unsupplement- grams more than their counterparts whose mothers received ed group was 8.7 percent, to 10 percentage points in the no supplements (Mora 1983); in the total group, the difference Gambia, where the presupplementatiotn incidence (averaged in birth weight between the infants of those whose diets were over all seasons) had been 18 percent. supplenmented and those whose diets were unsupplemented was The selection of subgroups for targeted interventions on the only 51 grams. In Taiwan (China), "tall and thin" women who basis of characteristics of those who are known to respond were below the median weight and above the median height rather than the selection of subgroups oni the basis of charac- had larger birth weight in response to supplementation than teristics associated with malnutrition is likely to be much more "short and thin" subjects (Adair and Pollitt 1985). The better effective. The response in birth weight to maternal food sup- response in tall, thin women to supplementation emphasizes plementation is greater among chronically and acutely mal- an irreversible detrimental effect of stunting in women and nourished women, as well as those who smoke. In Colombia, points to an additional advantage of preventing stunting dur- the mean birth weight of infants of women whose weight-for- ing childhood. Women may also be more responsive to food Protein-Energy Malnumtion 403 supplementation during periods of seasonal food shortage, as Klein 1982) are reviewed here. Improved dietary intake among illustrated by results from the Gambia, where the gain in birth children as a result of the provision of supplementary food weight per unit of supplementation was higher during the wet (detailed in table 18-8) has the potential to be partitioned into season, when food availability is low (table 18-7). different biological responses, such as growth, activity, and In the studies conducted in relatively well nourished popu- compensation for energy lost during illness, although most lations, supplementation of the diets of smokers had a greater studies reported only anthropometric indicators of growth. effect on birth weight than the supplementation of those of Behavioral changes resultingfrom provision ofenergy (Chavez nonsmokers (Rush, Stein, and Susser 1980a, 1980b; Kennedy and Martinez 1975; Chavez, Martinez, and Yaschine 1975) and Kotelchuck 1984). This effect may be of importance in may also improve cognitive development, but that will be developing countries as smoking among women increases. If discussed in the section on integrated interventions, because the effect of smoking on birth weight is a result of carbon well-designed studies which have measured this effect have monoxide and not nicotine or cyanide compounds (Kramer also provided behavior-modifying interventions to the child. 1987), it may also be relevant among women who are exposed Supplementation of children's diets improved both their to smoke from wood stoves in poorty ventilated houses. weight and height (table 18-9). The response in growth de- Studies of food supplementation in lactating women have pended on the age of the child, which determined both the shown mixed results. A well-implemented trial showing no magnitude and kind of deficiency, whether in protein or calo- effect (Prentice and others 1980; Prentice, Roberts, and Pren- ries. Provision of supplemental food remedied these deficien- tice 1983) may still be confounded by secular trends. The only cies to a marked but still imperfect degree. Taking into account randomized double blind trial conducted to date (Gonzalez and the duration of supplementation (derived from table 18-8), we others 1991) of food supplementation in lactating women who estimate the height difference to vary from a high of 5.0 had no supplementation during pregnancy showed a clear-cut centimeters per 100,000 kilocalories ingested by Guatemalan increase in breast milk in a group of chronically malnourished females age thirty-six months to a low of 0.8 centimeters per women. The increase was larger among the initially most 100,000 kilocalories ingested by Indian children age forty- malnourished. This study also showed that women whose diet eight months to sixty months. The weight difference per was supplemented prolonged full breastfeeding. These findings supplement ingested ranged from a high of 800 grams extra per have implications not only for maternal and infant nutrition 100,000 kilocalories, attained by male Guatemalan children at but also for birth spacing, because full breastfeeding postpones thirty-six months of age, to a low of 40 grams extra per 100,000 the return of fecundability much more than does matemal kilocalories, ingested by Colombian children during the inter- malnutrition (Kurz and others 1990). val between eighteen months and thirty-six months of age.' Results of a recent observational study in Brazil point to a The variability in response to food supplementation in these "weanling dilemma" for infants of poorly nourished mothers in studies, particularly in weight, may be because of the differ- contaminated environments (Martines and others 1988). ences in age groups supplemented, infant and child feeding Growth faltered at about three to four months of age practices, incidence of infectious diseases, and the duration of among solely breastfed infants of poorly nourished mothers, supplementation. The studies in Bogota and Guatemala in- whereas solely breastfed infants of well-nourished moth- cluded the most complete data for examining the effects of ers maintained normal growth at this age (Martines and others supplementation on incremental growth at different ages. The 1988). The provision of solid food to the infants of poorly differences in the resulting relative rates of growth in length nourished mothers did not appear to be a solution, because and in weight between the supplemented and unsupplemented diarrhea and other infections resulting from contamination of children were greatest between nine months and twelve the foods had an even worse effect on growth than the inade- months, during peak incidence and duration of diarrhea, fol- quacy of maternal milk production (Martines and others lowed by the weaning period from three to six months (Lutter 1988). Thus, whereenvironmental sanitation ispoor,postpon- and others 1990). Food supplementation also modified the ing the introduction of weaning foods beyond the optimal negative effect of diarrhea on growth. Diarrhea was negatively period for infants in sanitary environments is less detrimental related to growth in length in the unsupplemented group, but to infant growth than is the introduction of contaminated had no effect on length in the supplemented group (Lutter and foods at that time. The "weanling dilemma" might be solved others 1989). The effect of supplementation depended on the by improving the nutrition of the malnourished mothers so level of diarrheal disease, having a greater effect on length in that growth of their infants does not falter, or by providing those with the most diarrhea and no effect on children with uncontaminated foods to infants. Empirical studies, however, low levels of diarrhea (Lutter and others 1989). Similarly, in have as yet not shown that either of these strategies improve Guatemala, in the villages provided with a protein and energy- infant growth. enriched supplement, the effect of diarrhea on growth was Several studies that were sufficiently well designed to make attenuated (Martorell, Rivera, and Lutter 1988; Rivera, estimations of the effect of food supplements on the growth of Martorell, and Lutter 1989). children (Gopalan and others 1973; Martorell, Klein, and Finally, children also responded differently tosupplementa- Delgado 1980; Mora and others 1981; Martorell, Habicht, and tion at different ages according to their body composition. In Table 18-8. Supplementation to the Child Home dietary intakea Total supplement distributed Net intake Duration of Age of child Energy Protein Energy Protein Energy Protein Location supplementation Treatment group (montks) (kcal/day) (giday) (cal/dav) (g/day) (kcal/day) (giday) India Fourteen months Supplement 12-23h 700 18b 310 3 310 3 No supplement 12-23 700 I 8 0 0 0 0 Bogota, Colombia From mother's last Supplement 18 1,020 706 24 38 623 30.0 1,478711 5840 trimestertothreeyears Nosupplement 18 1,310445 3618 0 0 1,310445 3618 of age Supplement 36 1,017 373 24 16 623 30.0 1,380 453 48 24 JP No supplement 36 1,167 476 27 16 0 0 1,167 476 27 16 42 Guatemala Fromn mother's pregnancy Male to three years of age Atole 15-36 785 + 213 206 163 11.5 941 226 31 8 Fresco 15-36 814+213 196 59 0 C 840226 228 Female Atol6 15-36 718+213 206 163 11.5 868226 21 8 Fresco 15-36 756+213 19+6 59 0 779226 88 a. Home dietary intake valLes are likely ro be an underestimiiate of the intake the population wotild normally consume, since some of the home diet is probably displaced by rhe supplement. b. For entire population. c. Per 180 ml, fed ad libitum. Source: Gopalan and others 1973; Martorell, Klein, and Delgado 1980; Mora and others 1981; Martorell, Habicht, and Klein 1982. Table 18-9. Supplementation to the Child: Outcome Initial measurements Final measurements Increment Outcome + Input h Age of child Staturea Weight Stature' Weight Stature' Weight Stature' Weifht Location Treatment group (months) (cm) (kg) (cm) (kg) (cm) (kg) (cm/I05 kcal) (kg/10 kcal) India Supplement 12-24 - - - - 9.3 2.35' 2.1 0.49 No supplement 12-24 - - - - 6.5c 1.71I n.a. n.a. Supplement 24-36 - - - - 9.5 2.34c 1.3 0.48 No supplement 24-36 - - - - 7.8c 1.7 1 n.a. n.a. Supplemnent 36-48 - - - - 9.cI 2.04d 1.3 0.35 No supplement 36-48 - - - - 74 1.58d n.a. n.a. Supplement 48-60 - - - - 8.4e 1.86a 0.8 0.37 No supplement 48-60 - - - - 7.3e 1.38 n.a. n.a. Bogota,Colombia Supplement 18-36 75.1 2.7 9.48 1.09 87.5 3.4. 12.35 1.40e 12.4 2.87 1.5 0.05 No supplement 18-36 73.5 3.1 9.06 1.04 85.3 4.4f 11.88 1.19e 11.8 2.82 n.a. n.a. Guatemala Male Atole 36 84.3 + 4 11.6 1.3g 86.9 3.9' 12.4 1.3 - - 2.8 0.79 Fresco 36 84.3 4 11.6 1.3 85.1 3.9' 11.9 1.3 -- - n.a. n.a. Female AtolI 36 84.34 11.61.3 85.93.9c 1l.93c _ _ 5.0 0.70 Fresco 36 84.34g 11.61.3g 83.1 3.99 10.8 1.3 - - n.a. n.a. -Data nor available. n.a. Nor appiLcable. a. Some studies reported recumbent length, and others reported standing height. The use of length or height makes no difference to the net outcome, because it is detetmined by the difference between measurements. b. Inptit is calculated as the kilocalories of supplement ingesred per day dimes the duration of supplementation. In Colombia, the intake was the average of the intakes at eighteen and thirry.six months. In Guatemala, where infants also receive breast milk, the amount of supplement is likely to be less; the duration of supplementation therefore was considered to be the period between twelve and thirty-six months. c. Significant difference between groups; P < 0.001. d. Significant difference between groups; P < 0.01. e. Significant difference between groups; P < 0.05. f. Significant difference berween groups; P < 0.005. g. Total population. Sources: Gopalan and others 1973; Martorell, Klein, and Delgado 1980; Mora and others 1981. 406 Per Pinstrup-Andersen, Susan Burger, Jean-Pierre Habicht, and Karen Peterson Guatemala, children of high weight-for-length at six months prove weaning practices frequently involve the promotion of (Rothe, Rasmussen, and Habicht 1988) had higher weight improved weaning food, whether made from local foods or gain, and at eighteen months of age (Marks 1989) had better commercially premixed (Gibbons and Griffiths 1984). Re- linear growth in response to supplementation than children of searchers who studied projects promoting weaning foods in low weight-for-length. At thirty months of age, this response Haiti (Berggren 1981) and Burkina Faso (Zeitlin and was reversed; children of low weight-for-length responded Formaci6n 1981) found strong positive effects on the nutri- with better linear growth (Marks 1989). The better linear tional status of children. The affordability of weaning foods is growth of fatter children at younger ages in response to supple- an important consideration, and emphasis should be placed on mentation could be explained by two potential mechanisms: locally available foods. Successes from recent integrated inter- (a) protein is more limiting in younger children, or (b) younger ventions may reflect a positive effect of improved weaning children need to build up fat reserves before they can respond foods. Attempts to isolate the nutrition effect of each of a with better linear growth. Further research is needed to test number of interventions, such as weaning foods, that form part either of these hypotheses. of an integrated program may not be warranted because the effect of one is heavily influenced by the presence of others. FOOD SUBSIDIES AND TRANSFERS. Food price policy, food sub- sidies, and food-related transfer programs influence the ability FORTIFICATION. Although fortification may be effective in of households to acquire food (Pinstrup-Andersen 1985). Ev- alleviating deficiencies in micronutrients, fortification to alle- idence from studies in more than a dozen countries shows that viate PEM has not generally been successful. Amino-acid forti- food price subsidies and food stamp programs have increased fication has been tried in a number of populations. In many incomes and improved food consumption among the poor, cases, however, little or no effect was detected, partly because particularly but not exclusively in urban areas (Pinstrup- the programs were not designed in such a way that they could Andersen 1988). In many of the countries studied, the transfers be evaluated and partly because the staple foods on which from such programs contributed 15 to 25 percent of total fortification was tested-for example, wheat in Tunisia and income of the poorest 10 to 20 percent of the population rice in Thailand (N.AS 1988)-were less likely to lead to a reached. In general, each I percent increase in the income of critical imbalance in amino acids than other staples, such as the poor results in a 0.2 to 0.4 percent increase in household corn. cassava, or sweet potatoes. Thus, the additional protein energy consumption and a somewhat higher percentage in- provided was most likely converted to energy. crease in the consumptionofprotein. Thus, although the effect varies among countries and population groups, food subsidy AGRICULTURAL PRODUCTION. Agricultural programs and pol- programs that add 20 percent to the purchasing power of the icies may have important nutrition implications through their poor can be expected to increase their energy consumption by effect on incomes of the rural poor, food prices to be paid by 4 to 8 percent. The effect will be higher among the poorest and poor consumers, time allocation by members of poor house- most destitute. holds, energy and protein expenditure, infectious diseases, and The extent to which increases in household food consump- food consumption by high-risk individuals (Mebrahtu, Pelle- tion lead to improved nutritional status depends on the degree tier, and Pinstrup-Andersen, in press). Positive nutrition ef- to which malnourished household members share in the in- fects may be increased and negative ones alleviated by crease and on the relative importance of food deficiency and including in the design of such programs and policies consid- infectious diseases. If food deficiency is not the only limiting eration of how each of these six factors will be affected. constraint to improved nutrition, increased food consumption Agricultural research and technology have contributed to mayhave little orno effect. In Egypt, forexample, existingfood large increases in food production in many developing coun- subsidies contributed to high levels of food consumption even tries during the last twenty-five years, most notably in Asia and among the poor. Malnutrition was still significant, however, Latin America (Pinstrup-Andersen 1982). These increases because the effect of poor sanitation and diarrhea on growth have resulted in higher incomes for farmers and landless work- were not effectively addressed (Alderman and Von Braun ers as well as lower food prices for consumers. Low-income farm 1984; Alderman, Von Braun, and Sakr 1982). families with members at risk of rEM have participated in these Empirical evidence on the effect of food subsidies and price income gains, and the lower food prices have been particularly policies on the nutritional status of preschool children is important for poor consumers, becausL they spend a large share limited. Kumar (1979) found that the weight-for-age of chil- oftheir income on food (Pinstrup-Andersen and Hazell 1985). dren in Kerala, India, would fall by 8 percent if the existing Although evidence from other sources shows that incomes and subsidized food ration sclheme was discontinued. Similarly, food prices are closely associated with the nutritional status of food subsidies contributed significantly to increases in energy household members, implying a positive nutrition effect of and protein intake and weight-for-age of preschool children in agricultural research and technology, no studies have been the Philippines (Garcia and Pinstrup-Andersen 1987). done to establish the link directly. The authors of recent studies of projects to increase the WEANING FOODS. The risk ofPEM and PEM-related mortality is commercial agricultural production by small farmers in the particularly high during the weaning period. Programs to im- Gambia, Guatemala, India, Kenya, the Philippines, and Protein-Energy Malnutrition 407 Rwanda (Alderman 1987; Kennedy and Cogill 1987; Von relatively low cost. in low-income households, such programs Braun, Hotchkiss, and Immink 1989; Von Braun, Puetz, and are most likely to be successful when the behavioral changes Webb 1989; Bouis and Haddad 1990; and Von Braun, de Haen, can be accomplished without additional resources (Homik and Blanken, 1991) found strong positive effects on farm 1985). The interaction between resource availability at the incomes and family food consumption. The effect on the household level and the opportunity for nutritional improve- nutritional status of preschool children was also positive but ment through behavioral changes is an important consider- small, in some cases statistically insignificant. The results of ation in the design of nutrition intervention programs. In very these studies demonstrate that agricultural projects may be poor households, constraints on resources may prohibit behav- powerful in alleviating the household resource constraints to ioral changes and thereby limit the effectiveness of nutritional good nutrition but may have little effect on nutrition onless education. other constraints, such as infectious diseases and adverse Cerqueira and Olson (in press) argue that lack of success in household behavior, are removed at the same time. many past nutrition education programs is due in large part to the influence of the traditional medical model, that is, curative INCOME GENERATION. Protein-energy malnutrition is closely rather than preventive. Recent participatory models with or associated with poverty, and increased household incomes are without the use of mass media have contributed to significant important for improved nltrition. Still, as illustrated above, nutrition improvements in several countries, including Indo- additional income is not always sufficient for its elimination. nesia (Zeitlin and others 1984), the Dominican Republic There are several reasons for this. In the short run, additional (USAID 1988), Tanzania (UNICEF 1988), and India (Shekar income is likely to increase access to food but may have little 1991). In most of these cases growth monitoring played an effect on general sanitary conditions, knowledge, and access to important role as a source of information for the education and use of health care. In the absence of good sanitation and effort, and in some but not all of these programs nutrition health care, additional food may have little nutritional effect. education was closely integrated with other interventions In the longer rin, increased income is likely to improve living (Cerqueira and Olson, in press). Thus, nutrition education conditions, reduce health risks, and improve nutrition, partic- offers great promise for nutritional improvements through ularly if accompanied by better information and behavioral behavioral changes when (a) the design and implementation changes. In the meantime, increased access to food may have is based on a thorough understanding of the environment a significant effect on nutrition only if primary health care and within which the intended beneficiaries live and of the con- programs to modify household behavior and improve sanitary straints they face, including those that limit behavioral conditions and access to more and cleaner water are intro- changes; (b) the target community is intimately involved in duced as well. the design and implementation of the complete set of inter- Another reason why income increases may not be as effec- ventions; and (c) household resource constraints are alleviated. tive as expected is that households may be unaware that a nutrition problem exists, they may lack knowledge and infor- PROMOTION OF BREASTFEEDINCG. Breastfeedingcontributessig- mation about how best to use additional income for nutritional nificantly to reduced infections and malnutrition in infants improvements, or they may be faced with extreme scarcity of and children. Promotion of breastfeeding has been successful other basic necessities which compete with nutrition for in increasing the rate ofbreastfeeding (Feachem and Koblinsky household resources. In such cases, growth monitoring or 1984). Maternal behavioral response to breastfeeding promo- nutrition education may be needed along with efforts to assist tion varies according to socioeconomic characteristics (Hardy households in meeting other basic needs. and others 1982). Training and education of health profession- The nutrition effect of income increases may also be less als in the advantages of breastfeeding and in techniques to help than expected because women allocate more rime to income mothers breastfeed have been successful in a number of coun- generation and less to child care, cooking, and other nutrition- tries (Huffman and Steel, in press), including Indonesia, Thai- related activities. In such cases, programs to increase the land, Panama, and the United States. Appropriate hospital productivity of women's time within or outside the household delivery practices, community support groups, and arrange- are needed. Finally, intrahousehold income control may be in ments permitting mother and infant to stay together while in the hands of household members who place low priority on the hospital have proved effective in increasing breastfeeding nutrition improvement. rates and the ways breastfeeding is practiced in several coun- tries (Huffman and Steel, in press). NUTRITION EDUCATION. Several researchers have completed reviews of the effect of nutrition education programs during Integrated Interventions to Improve Dietary Intake the last few years, including Whitehead (1973), Zeitlin and Formaci6n (1981), Hornik (1985), Johnson and Johnson Synergism among interventions may result in integrated (1985), and Cerqueira and Olson (in press). The general strategies' having effects different from the sum of the effect of conclusion from these reviews is that well-designed and well- each of the individual interventions. Substitution and comple- implemented nutrition education programs can bring about mentarity between and among interventions often play an behavioral changes that contribute to improved nutrition at important role in determining the total effect of a set of 408 Per Pinsmrp-Andersen, Susan Burger, Jean-Pierre Habichr, and Karen Peterson interventions. On the basis of a review of nutrition education need not be among the criteria for receiving treatment. Nutri- programs in developing countries, Hornik (1985) concludes tional status may be used to target interventions such as food that nutrition education generally has been most successful supplementation, but the goal of the selection procedures is to when combined with increased household resources. Comple- prevent severe PEM and its consequences rather than to provide mentarity between nutrition education and food supplemen- rehabilitation once it occurs. Severely malnourished children tation has been demonstrated by Gilmore and others (1980) are commonly treated in two types of facilities: hospitals and in Morocco, and Garcia and Pinstrup-Andersen (1987) found nutrition rehabilitation (or mothercraft) centers. complementarity between nutrition education and food price In hospitals generally, the case-fatality rate of severe malnu- subsidies in the Philippines. Developmental education to in- trition has been reported to be approximately 25 percent fluence maternal knowledge, behavior, and time spent with (McLaren and others 1969; Cook 1971). Most studies of hos- the child appears to increase the effect of food on growth and pital-based case management of severe PENi on which these cognitive development. Home visits by fieldworkers to stitnu- rates are based were conducted more than twenty years ago. In late learning and development and positive caretaker-child the interim, new methods of treatment and feeding schedules interaction among families participating in the Bogota Nutri- with better potential to reduce this high case-fatality rate have tion and Child Development Project (Herrera and others been developed (Waterlow and others 1978; WHO 1981). The 1980) resulted in a growth in height of 1.3 centimeters more in effectiveness of these innovations will depend on adequate children in the supplemented group who had received stimula- training and sufficient time and personnel to administer the tion than in those who had not (Lutter 1987). This effect on frequent feedings required for rehabilitation. Follow-up upon height-for-age persisted at six years of age (Super and others 1989). release from the hospital is necessary to avoid relapse. Re- The main effect of stimulation was stronger than the main ported case-fatality rates ranged from 15 to 37 percent, hut effect of supplementation in the Bogota study for almost all the they dropped to nearly zero in the twelve-month period if indexes of cognitive development tested (Cremer and others "conscientious efforts were made for follow-up" (Cutting 1983, 1977). The highly significant interaction of supplementation p. 121).4 and stimulation (Cremer and others 1977), points to the Nutrition rehabilitation centers vary from those incorpo- synergism of food and behavioral stitnulation interventions. rated into a highly technical medical infrastructure to those Similarly, behavioral stimulation had a much larger effect on completely separate from medical facilities (Beaudry-Darisme cognitive development than dietary supplementation in chil- and Latham 1973). The case-fatality rates in a four-month dren in Cali, Colombia (McKay and Sinisterra 1974, McKay period ranged from 0 to 6 percent (Beghin and Viteri 1973). and others 1978), although these results may be biased because The effect of length of attendance at rehabilitation centers participants were explicitly told the nature and the expected on nutritional status was retrospectively examined in two outcome of the study. Stimulation of severely malnourished different countries, with equivocal findings (Beaudry-Darisme children in a hospital also had similar effects (Grantham- and Latham 1973). In Haiti, improvement in the median McGregor and others 1979; Grantham-McGregor, Stewart, weight-for-age by the time of discharge and ability to maintain and Schofield 1980; Grantham-McGregor, Schofield, and it one year after discharge were significantly greater in those Harris 1983). who attended the centers for two months than for those who Integration of efforts to improve nutrition and health in attended for one month. The mean percentage of Guatemalan small-scale projects has positively affected nutrition and mor- children who attained and maintained the median weight-for- tality (Gwatkin, Wilcox, and Wray 1980; Lamptey and Sai age was the same regardless of length of attendance. 1985). Two recent integrated projects-the Tamil Nadu Inte- Still, the above outcomes are all the result of poor planning grated Nutrition Program in India and the Iringa Project in or management rather than intrinsic biological constraints on Tanzania-have been particularly successful in reducing child recovery. In a well-managed trial in Guatemala, all malnour- malnutrition. Thus, TINP is estimated to have reduced the ished children recovered when they received 10 percent or prevalence of severe malnutrition among children six months more of their recommended energy intake from food supple- to thirty-six months old by one-third to one-half the preva- mentation (Rivera, Habicht, and Robson, in press). lence that existed at the beginning of the project. The contin- Questions yet to be resolved are what referral criteria would ued effect of TINP and another nutrition project in the region be most appropriate and under what conditions are the two is estimated to be a 50 percent reduction of the prevalence of types of facilities most useful to rehabilitate severely malnour- malnutrition among children less than five years old (World ished children when preventive efforts fail. A comparison of Bank 1990). An equally impressive result was obtained by the the relative effectiveness of hospitals and nutritional rehabil- Iringa Project, which is estimated to have reduced severe and itation centers indicated that nutrition rehabilitation centers moderate malnutrition in the project area by 70 and 32 per- were more effective in curing malnutrition, although the out- cent, respectively (UNICEF 1989; Yambi and others 1989). come measure was not specified (Cook 1971). The criteria for hospital admission and prognosis may have been different for Case Management of Severe Malnutrition the two types of facilities, however, because hospitals usually deal with severe malnutrition, whereas rehabilitation centers Most of the interventions previously described can be consid- handle less severe cases. This type of comparison is not useful ered preventive because a diagnosis of severe malnutrition in determining clinical criteria for referral of children from Protein-Energy Malnutrition 409 nutritional rehabilitation centers to hospitals. Perhaps because dren and their rate of mortality estimated for five countries and the costs are so high, the further study of criteria for referral synthesized by Pelletier (1991). The result from the assess- has been grossly neglected. ment, which should be interpreted as rough magnitudes rather than exact figures because of the assumptions made, is that the Cost, Cost-Effectiveness, and Benefit-Cost cost of averting a death is $1,942, which converts to a cost of $62.65 per disability-adjusted life-year (DALY'). The second While direct costs of nutrition intervention are frequently refers to food supplementation of pregnant women and is based reported, total costs and benefits are difficult to estimate. This on the relation between food supplementation and birth section provides rough estimates only. weight shown in table 18-7 and the relation between birth weight and infant mortality shown by Walsh and others (in FOOD SUPPLEMENTATION. As shown in table 18-10, the cost press). If we assume a gain of 300 grams in the birth weight per of food supplementation varies among programs. Such varia- 100,000 calories transferred to pregnant women (table 18-7) tion provides little indication of cost-effectiveness, however, and a decrease of 9 per 1,000 in the child mortality rate for because the programs vary in nature and presumably in nutri- each 100-gram increase in the birth weight (approximated tion effect. As expected, the cost is lowest for food subsidy from Walsh and others, in press), the cost per child death schemes that depend on the market for food distribution and averted is estimated at $733.00, or $23.65 per DALY. The include no other program components, such as those in the assumptions made and the methodology used in this assess- Philippines, Sri Lanka, and Egypt, and highest for integrated ment are explained in appendix 18A. programs implemented by the public sector, such as the Tamil The benefit-cost analysis is based on the elasticity of labor Nadu and the Colombian programs. productivity of 1.38 with respect to height of the worker, These cost estimates provide an input into assessments of reported by Haddad and Bouis (1990) and discussed in an cost-effectiveness. On the basis of the estimates presented in earlier section of this chapter, and the estimated relation table 18-10, it seems reasonable to assume a cost of $0.20 per between food supplementation and height of preschool chil- 1,000 calories transferred in programs that provide no other dren (table 18-9). Estimates of the latter vary from 0.8 to 5.0 services. This rough overall estimate is used in an assessment centimeters per 100,000 calories transferred. For this analysis, of cost-effectiveness and a benefit-cost analysis presented an estimate of 2.0 is used. This corresponds to the finding for below. Indian children age one to two years (table 18-9). The sensi- Two cost-effectiveness analyses are reported here. Both tivity of the result to the choice of estimate is tested by using assess the cost of averting a child death by means of food estimates of 1.0 and 3.0 centimeters per 100,000 calories supplementation. The first, which relates to food supplemen- transferred. As above, a cost of $0.20 per 1,000 calories trans- tation for preschool children, is based on the relation between ferred is assumed. The cost associated with an increase in food supplementation and child growth, shown in table 18-9, height of 1.0 centimeter is then the cost of 50,000 calories, or and the relation between the weight-for-age of preschool chil- $10.00. Under the assumptions of a daily current wage of $2.50 Table 18-10. Cost of Food Transfers in Various Types of Programs Cost per 1 ,000 calories delivered Progiram type Country Year (U.S. dollars) Source Food subsidy Brazil 1980 0.30 Berg 1987 Food subsidy Colombia 1981 0.79 Berg 1987 Food subsidy Egypt 1982 0.18 Alderman and Von Braun 1984; Kennedy and Alderman 1987 Food subsidy Mexico 1982 0.42 Kennedy, Overholt, and Haddral 1984 Food subsidy Philippines 1984 0.09 Garcia 1988 Food subsidy Sri Lanka 1982 0.10 Edirisinghe 1987 MCH and feeding Bolivia 1988 0.12 World Bank 1989 MCH and feeding Brazil 1980 0.53 Berg 1987 MCH and feeding Colombia 1982 0.38 Anderson and others 1981; Kennedy and Alderman 1987 MCH and feeding Costa Rica 1982 0.60 Anderson and others 1981; Kennedy and Alderman 1987 MCH and feeding Dominican Republic 1982 0.20 Anderson and others 1981; Kennedy and Alderman 1987 MCH and feeding Ecuador 1988 0.21 World Bank 1989 MCH and feeding Guatemala 1988 0.33 World Bank 1989 MCH and feeding Honduras 1988 0.36 World Bank 1989 MCH and feeding India (Tamil Nadu) 1985 0.69 Berg 1987 MCH and feeding India 1982 0.20 Anderson and others 1981; Kennedy and Alderman 1987 MCIIH and feeding Pakistan 1982 C.38 Anderson and others 1981; Kennedy and Alderman 1987 MCH and feeding Paraguay 1988 0.90 World Bank 1989 MCH and feeding Peru 1988 0.32 World Bank 1989 MCH and feeding Uruguay 1988 0.24 World Bank 1989 Souirce: See l ast column. 410 Per Pinstrup-Andersen, Susan Burger, Jean-Pierre Habicht, and Karen Peterson and 300 working days per year, the annual current benefit is relatively high cost estimate is that, although severely mal- estimated at $6.45. Still, these benefits will begin only after nourished children were most closely monitored, program the child enters the labor market, for example, at the age of resources were provided to all children, of which only 12 eighteen years. Assuming a 2 percent annual rate of increase percent were classified as malnourished at the start of the in the real wage, we find that the wage rate will have increased program. If the program could be targeted to the malnourished to $3.43 per day, or $1,030 per year, for the year the person only, the cost would be reduced considerably without a reduc- enters the labor market. Thus, an investment of $10.00 in a tion in the effect as measured. Alternatively, the cost- child age six months to eighteen months will generate an effectiveness is likely to be higher in populations with a higher income stream beginning sixteen years later and continuing prevalence of malnutrition. Furthermore, only benefits associ- for the duration of the productive life of the worker, say, until ated with the movement of children from below to above the age fifty-five. This annual income stream will begin with $8.84 cutoff of 75 percent of standard were considered, and nutri- at age eighteen and end with $18.06 at age fifty-five. At a real tional improvements occurring within each of the two groups discount rate of 3 percent, the present value of such an income were ignored. stream is estimated to he $174.00, which, compared with the initial investment of$10.00 yields a benefit-cost ratic) of 17.4. INTEGRATEL) PROGRAMS. Several recent integrated health This is above most other investments made by the public and nutrition programs, including the Tamil Nadu Integrated sector. Reducing the effect of food supplementation from 2.0 Nutrition Program and the joint WHO/UNICEF Nutrition Sup- centimeters to 1.0 centimeter per 1 00,000 calories resulted in port Program in Iringa, Tanzania, have been successful in a benefit-cost ratio of 8.7, whereas an increase to 3.0 was significantly reducing PEM. The cost of removing a child from associated with a benefit-cost ratio of 26. Thus, even at the moderate and severe malnutrition was estimated to be $33 in lowerestimateoftheeffectoffoodsupplementationonheight, Tamil Nadu (Ho 1985) and $46 in Iringa (estimated on the the return on investment in the growth of preschool children basis of data in UNICEF 1988). Using the methodology reported through food supplementation is high. The methodology used in the appendix, we estimated the cost per averted child death is presented in appendix 18A. in Iringa to be $2,560, corresponding to a cost per DALY of $82. Estimates by Selowsky and Taylor (1973) for Chile and by Selowsky (1981) for Colombia also suggest high ecolnomic COSTSOF CASE MANAGEMENT.The costs of case management returns to government programs to improve child nutrition. In of moderate to severe protein-energy malnutrition are consid- another studv in Chile, Torche (1981) estimated that the erably higher than the costs of prevention. Beaudry-Darisme economic return from a food supplementation program ex- and Latham (1973) compared the costs of rehabilitation cen- ceeded the discount rate of 17 percent usually applied to ters in two different countries using standardized food and determine the economic viability of investnent projects in the salary costs. The mean cost per child with a positive change in government sector. Additional studies of this nature are ur- weight-for-age above the control was $605 in Haiti and $2,672 gently needed toprovide evidence of the economic return from in Guatemala and was even higher for the decreased incidence nutritional improvements compared with the economic return of severe PEM above the control group at $3,627 in Haiti and from alternative investment and to increase the understanding $5,344 in Guatemala. In a comparative review of rehabilita- of how to design nutrition interventions with high economic tion centers and hospitals, costs for the period of recuperation payoff. varied from $46 to $54 in Guatemala to $120 in Costa Rica in rehabilitation centers (Beghin and Viteri 1973). In urban NUTRITION EDUCATION. There is little reliable quantitative Uganda the cost of recuperation was $78 in the recuperation evidence on the cost-effectiveness of nutrition education. It center and $120.00 in the hospital (Beghin and Viteri 1973). may be difficult nr impossible to isolate the effect of nutrition In a study of six Latin American countries, the cost perbed-day education when such education is an integral part of an inter- at a hospital ranged from 4.5 to 18 times the cost per child per vention program. In such cases it may make more sense to eval- day at a rehabilitation center (Beghin 1970). Still, the severity uate the complete program, as discussed in the next section. of the PEM, the criteria employed for referral, and the duration Two recent nutrition education programs provide informa- of recovery, among other factors, could account for some of the tion on costs and cost-effectiveness. Both are based on the difference in cost. Costs varied from $117.60 for inpatient social marketing strategy and both have been effective in hospital treatment for kwashiorkor for two weeks to $77.00 for improving nutrition. In Indonesia, the cost was $3.90 for each a six-week inpatient stay that included educating the mother. participant and $9.80 for each child whose nutrition was improved. In the Dominican Republic, the Applied Nutrition Strategies and Priorities Education Program included about 9,000 children and cost about $23.00 per child annually. The prevalence of malnutri- An effective strategy must identify and combine programs and tion, defined as weight-for-age below 75 percent of standard, policies that are likely to have the greatest sustained effect on decreased from 12.2 to 6.9 percent in a two-year period (1984- the nutritional status of a particular population group or groups 86). The cost per child removed from malnutrition was esti- per unit of resources spent. The design of such a strategy is mated at about $500.00 (USAID 1988). One reason for this difficult because of strong synergisms, such as those between Protein-Energy Malnutrition 411 food intake and infectious diseases and between purchasing growth in an individual child. Most of the successful integrated power and nutrition knowledge. These synergisms imply that programs mentioned earlier include growth monitoring as a the effect of changes in one factor may be slight unless other tool to integrate intervention activities. From the success of factors are changed simultaneously. Furthermore, because the these programs it appears that growth monitoring and promo- nature, causes, and relative rewards of particular interventions tion can be an important instrument in an integrated strategy. vary among population groups, across countries, and over time, Furthermore, the cost of adding growth monitoring to nutri- the strategy must be tailored to a particular set of circum- tion intervention programs is small (Griffiths 1985). stances. No one strategy is likely to be optimal for all popula- In their review of the early attempts at growth monitoring, tion groups. Gopalan and Chatterjee (1984) came to the conclusion that A national strategy to alleviate PEM should consider direct muchofthefailuretoachievethepromisedpotentialofgrowth nutrition interventions, such as food supplementation and monitoring was due to the difficulty of interpreting the growth nutrition education; indirect interventions, such as programs curve and the consequent lack of appropriate follow-up action. to improve primary health care, sanitation, and water avail- The Applied Nutrition Education Program in the Dominican ability; and broader policies and programs, such as those in- Republic amongothershasdemonstrated thattrainingworkers volving price, income, credit, interest rate, and employment to implement specific activities to address specific patterns on policies and those influencing asset ownership and user rights. the growth chart can be successful (ULSAID 1988). This is, Such broader policies are likely to exercise powerful influences however, but one of the constraints to be overcome to make on the nutritional status of the poor, even though they may growth monitoring successful (Ruel, in press). not be explicitly addressed to nutritional concerns. Their Just as growth monitoring can be used to integrate health nutrition effects-positive or negative-should not be ig- and nutritional interventions at the level of the individual, so nored. This is particularly important because poverty, with its nutritional surveillance can be used to plan and implement associated unsanitary living conditions and lack of access to interventions in populations and coordinate and integrate sufficient food, health care, and information, is clearly the these interventions to make them more effective and less overwhelming determinant of PEM. As mentioned earlier, how- costly (Mason and others 1984; Tucker and others 1989). Like ever, policies and programs with the sole goal of increasing growth monitoring, nutritional surveillance does not by itself incomes have not been as effective in alleviating malnutrition prevent or cure malnutrition. Instead it provides the relevant in the short run as expected. information for the choice, design, and correct timing, syn- The choice, design, and implementation of nutrition inter- chronization, and concatenation of effective interventions. vention programs should be made within the context of exist- Because the relation between information and interventions ing policies and expected changes in them, because specific effective for the population level is much less clear than it is policies benefit some groups of the poor and hurt others, between information and growth monitoring at the individual thereby changing the needs and the appropriate target groups level, it is not surprising that no useful evaluations of effect for nutrition programs. The most appropriate program choice, have been done on nutritional surveillance. In the absence of design, and implementation strategy will depend on existing such evaluations no firm conclusions can be drawn regarding economic policies as well as opportunities for policy change. the nutrition effects of nutritional surveillance. The relatively The specific strategy and related programs and policies must narrow form in which nutritional surveillance efforts have be tailored to each location and time periocd. Ideally, the most been operationalized in the past, however, suggests that any appropriate govemment support will be identified through positive effects which have taken place are far short of the effective participation by communities and target households potential (Pelletier, in press). As illustrated by the willingness in problem diagnosis, program implementation, and monitor- of the private sector to pay for market information, ing. With or wvithout such participation, however, the choice, decisionmakers value relevant information. Information plays design. and implementation of nutrition intervention pro- an important role in guiding government decisions related to grams should be preceded by (a) identification of the target food and nutrition (Alderman, in press). Thus, nutrition sur- groups, assessment of the constraints to good nutrition with veillance activities that make relevant and timely information which thev are faced, an assessment of their food acquisition available to decisionmakers is likely to result in better inter- and allocation and health-seeking behavior, and identification ventions. Because the cost of nutritional surveillance is such a of opportunities for nutrition improvement; (b) assessment small fraction of the cost of the interventions, the potential of institutional and administrative capabilities for program cost of wrong decisions, and the cost of maintaining the implementation; and (c) identification of sources of financing. capability of intervening in crises that the generation of the Two vehicles forthe genieration of the relevant information- type of information provided by effective nutrition surveil- growth monitoring and nutritional surveillance-have shown lance should be encouraged. great promise, not only for information generation, but also as The interaction between interventions and the socioeco- mechanisms with which to identify, coordinate, and integrate nomic and cultural environment within which they are intro- interventions and policies into an effective strategy. duced as well as the interaction among types of interventions Growth monitoring by itself is not an intervention; rather and the synergism among factors influencing nutrition are of it is a tool to coordinate appropriate interventions to promote paramount importance in designing the strategy. This implies 412 Per Pinstrup-Andersen, Susan Burger, Jean-Pcrre Habicht, and Karen Peterson that integrated or concurrent interventions are more likely to effective if designed by and for the community target groups be successful than single ones. Still, integrated interventions with outside support. require institutional and administrative capabilities and infra- In view of the above, high priority should be placed ont structure that are frequently in short supply. strengthening the capacity of households, communities, and In countries in which the necessanr capabilities and infra- government agencies to assess the nutrition situation, to diag- structure are available or can be developed as part of the nose the problem, to identify the critical constraints to good program, integrated health and nutrition programs combined nutrition, and to design, implement, and monitor strategies with favorable govemment policies offer great promise. Each most appropriate for a particular situation. Priority should also program should be tailored to the particular circumstances but be placed on making available to program designers and im- a combination of growth monitoring; nutrition education that plementors information and technologies needed to ensure emphasizes breastfeeding, child spacing, and weaning prac- success in such strategies. tices; financial and technical assistance for the production and International assistance to reduce PEM should focus on the distribution of weaning food; distribution of food stamps to strengthening of this capacity through a combination of train- participating households; and overall primary health care may mng, technical assistance, research, technology transfer, and form the core of most programs or part of the menu of options financial support to cover program costs. The nature of each for consideration. Active community and target group partic- of these activities and the correct combination needs to be ipation in various aspects of program design and implementa- defined for each country and community. The overriding tion is important for long-term sustainability, and separate but priority of international aid agencies and national govern- related efforts to assist the target groups in strengthening their ments should be to ensure that each community and each income-generating capacity are needed to reduce the need for country is capable of defining, implementing, and obtaining future outside financial support. financing for the most appropriate strategy and associated In countries in which the institutional and administrative policies and programs. That-and not a specific technology or capabilities and infrastructure are weak, less complex programs intervention-is the magic bullet to be applied across coun- should be pursued. For example, in areas that have health tries and communities. posts, if insufficient food intake is a constraint to good nutri- In support of these priorities for action there is a need for tion, the distribution of food stamps to low-income mothers operationally relevant research on a number of topics. First, a who bring their preschool children to the posts should be better understanding is urgently needed of how to make more considered. Such a strategy was successful in urban and rural effective the generation of useful information to support deci- areas of Colombia and in both remote and less remote rural sions by mothers, communities, and government agencies. For areas ofthe Philippines (Garcia and Pinstrup-Andersen 1987). this purpose, research is needed to determine (a) how best to Using food stamps instead of the more traditional food supple- generate such information, for example, various formulations mentation relieves the health system of physically distri- of growth monitoring or alternative sources of information for buting the food, a job for which it is not well equipped, mothers and cost-effectiveness of various methods of nutrition without removing the food-related incentive for coming to the surveillance for communities and govemment agencies; (b) health post. how best to integrate information generation with the design Rather than repeating the many other examples of appro- and implementation of interventions and broader policies and priate interventions already discussed, we will only suggest that programs; and (c) how best to analyze, interpret, and make the introduction of any such intervention in a particular available information, including the most appropriate institu- community should be preceded by a sound assessment of the tional arrangements. situation and that the intended beneficiaries should partici- Second, there is an urgent need for sound evaluations of the pate in that assessment. nutritional effect of various formulations of integrated nutri- It should be stressed that attempts to set up large-scale tion programs; of how such effect is influenced by the socio- nutrition programs that exceed the institutional and adminis- economic and cultural environments, including existing trative capabilities have failed in the past and will fail if tried nutrition-related policies; and of how information from such again. Instead, emphasis should be on strengthening these evaluation can be generalized for use in the guidance of pro- capabilities, including training at all levels and the building of grams elsewhere. nationwide primary health care systems, which may become As shown earlier, virtually all scientifically sound informa- the conduits for integrated nutrition and health programs. tion about the effect of nutrition interventions refers to single Such linkage will be successful only if nutrition is given a more interventions. Yet, the complex set of interacting determi- prominent role in primary health care. At the same time, nantsandstrongsynergismbetweenandamongvariousfactors nutritional improvements should be pursued through policies influencing the nutritional status lead to the conclusion that and programs that require less administrative and institutional single interventions may not provide the most cost-effective capabilities and infrastructure, such as price, income, and strategy. Evidence from various integrated programs support employment policies; credit and technical assistance to low- this conclusion. Much more solid evidence is needed, however, income people; basic training and education; and small-scale to guide the design and implementation of the most appropri- intervention programs. Such programs may be very cost- ate integrated programs for the future. Protein-Energy Malnutrition 413 Third, current attempts to deal effectively with the nutrition show the potential benefits to the various groups and to society problem are hampered by a lack of a clear understanding about as a whole, including additional analysis of the economic cost the most appropriate role of each group of key actors within a of malnutrition and cost-benefit and cost-effectiveness analy- given socioeconomic and political environment. In some ses of nutrition interventions. cases, agencies of national governments tty to do what would Fourth, operational research is needed on a number of be done more cost-effectively by the mother or the community, interventions to make available better information about what whereas in others, unreasonable expectations are placed on types of programs are likely to be most cost-effective under mothers to solve problems that can only he solved by agencies what conditions and for which population groups. of a national government. Fifth, certain biological and behavioral research is needed Attempts to provide predetermined solutions or magic bul- on the magnitudes and determinants of PEM among adult lets to solve problems that are poorly diagnosed frequently fail. women, including but not limited to maternal depletion and Conversely, universal magic bullets are useful to remove cer- how modifications in household behavior and access to re- tain constraints. The issue is to identify the constraints most sources may influence the nutritional status of women. This effectively alleviated by generalized solutions and those requir- area has been almost totally ignored in the past. Finally, there itig imlore specific solutions and to allocate available resources is an urgent need for research on how best to deal with the accordingly. increase of PEM as a result of economic crises and reforms and To do this, sound knowledge about household behavior socioeconomic transitions, such as urbanization and associated related to food acquisition and intrahousehold allocation as changes in dietary patterns. well as health and nutrition-seeking activities is essential. The above list of priorities for resource allocation and Erroneous assumptions about the response of households toz operational research should not be interpreted as being all- external intluences frequently result in policies and pro- inclusive. Because PEM is influenced by such a complex set of grams that are poorly designed, poorly implemented, and cost- factors, many more priorities could be listed. We have tried to ineffective. Insufficient knowledge about the constraints list those that we feel are most important. We have not listed to good nutrition at the household level adds to the prob- priorities associated with diseases discussed in other chapters lem. Finally, a better understanding of household coping strat- in this collection, although many of them (for example, diar- egies will greatly increase the ability to design appropriate rhea and measles), are of great importance in PEM. strategies. Research is also needed on the appropriate role of commu- nity action in alleviating health and nutrition problems. What Appendix 1 8A. Estimating Nutritional Benefits type of action is likely to be most useful in various socioeco- nomic and political environments, and how can such action This appendix summarizes various ways to estimate nutritional be most appropriately supported from outside the community? benefits. A great deal of lip service has been paid to community partic- ipation in primary health care and nutrition programs. In fact, Estimating Econonmic Losses from Stunting much of such participation has been limited to the im- plementation of externally designed aid conitrolled interven- Carlson and Wardlaw (1990) report that 163 million pre- tions. Community participation in problem diagnosis and the school children are stunted. Of these, 41 percent, or 66.8 choice, design, and implementation of health and nutrition- million, are severely stunted, whereas the rest, 96.2 million, related action that is sustainable without external material are moderately stunted. Assuming a normal distribution and support has not been widespread, and research is needed to the standard deviationi reported by W'HO (1979), the height of improve the understanding of how community action may be severely and moderately stunted children at the age of two most effective. years is I 0.0 and 7.0 centimeters below standard height, respec- With regard to researcl on the role of each of the key actors, tively. Martorell (in press) found that the absolute stunting in there is an urgent need to improve existing knowledge about childhood measured in centimeters translates to an equal institutional and political economy factors and how these reduction in the height of adults. On the basis of a study of a factors, including the behavior of various public and private sample of rural workers in the Philippines, Haddad and Bouis sector agencies and groups, affect nutrition and nutrition in- (1990) estimated the elasticity of labor productivity with re- terventions. The poor and malnourished usually possess very spect to height of the workers to he 1.38; that is, a I percent little political power and, although altruism may play a role, difference in height is positively associated with a difference government action to alleviate poor nutrition effectively is of 1.38 percent in labor productivity as measured by the actual likely to come about only if the nation as a whole or politically wage received. The average height of the sample workers was powerful groups within are perceived to gain. Therefore, the 160.0 centimeters. Thus, 10.0 and 7.0 centimeters correspond poor and malnourished must establish coalitions with more to 6.25 and 4.38 percent of the height, respectively. Multiply- powerful groups. Furthermore, efforts to alleviate malnutrition ing the elasticity and the height reduction due to childhood must be institutionalized if they are to receive the necessary stunting yields estimated losses in labor productivity of 8.63 political support. Finally, information must be available to percent for severely stunted and 6.04 for moderately stunted 414 Per Pistrup-Andersen, Susan Burger, Jean-Pierre Habicht, and Karen Peterson individuals. Assuming a daily current wage of $2.50 and 300 Notes working days annually, we arrive at an annual current wage income of $750.00 per person. The total economic loss as a 1. These levels are not likely to be sufficient to cover energy expenditure result of stunting is then estimated as the total number of assoiciated with work and other activities desired by most people. Thus, they children stunted times the economic gain per individual or a should be considered stiboptimal in most cases. children st, o. Measurable responses may be less than expected by a purely facto- loss of $4.32 billion as a result of severe stunting and a loss of rial addition because physiological adaptations during pregnancy, such as $4.36 billion as a result of moderate stunting. These figures changes in the resting metabolic rate and rhermogenesis, may be altered by yield a total annual economic loss of $8.68 billion. protein-energy malnutrition. Some of che improvement in dietary intake might be channeled into meeting the energy costs of these physiologic changes. Estimating Benefit-Cost Ratio of Preschoolers' Food 3. Weight at ages other than eighteen months and thirty-six months is not Supplement reported for Colombia in table 18-8 because the correspondingdietary intake was nor reported for other ages. 4. The "conscientious etforts" wvere not specifically described. On the basis of results reported in table 18-9 it is assumed that a food transfer of 100,000 calories to malnourished preschool children results in an addition of 2.0 centimeters to the height References of the child and that this height addition is maintained in adulthood.Withreferencetotable 18-10, it isfurtherassumed Abdullah, Mi. 1983. "Dimensions of Incra-household Food and Nutrient that the cost of the food supplementation program is $0.20 per Allocation: A Study of a Bangladesh Village." Ph.D. diss., UniversitV of 1,000 calories transferred. Thus, the cost per centimeter is London. $10.00. Using the above-mentioned estimates from the study Adair, Linda S., and Ernesto Pollitt. 1985. "Outcome ofMaternal Nutritional by Haddad and Bouis (1990), one centimeter is equal to 0.625 Supplementation: A Comprehensive Review of the Bacon Chow Study." . , . ~~~~~~~~~~~~~~AntericanlJourttalof Clincal Numtion 41:948-78. percent of the height of the workers studied, and the corre- sponding increase in wages would be 0.625 x 1.38, or 0.86 Adair, Linda S., Emesto Pollitt, and W. H. Mueller. 1983. "Matemal Anthro- pometric Changes Duiring Pregnancy and Lactation in a Rural Taiwanese percent. Under the above assumptions regarding daily wage Population." Human Biology 55(4):771-87. and number of working days per year, this amounts to $6.45 ---. 1984. "The Bacon Chow Study: Matemal Nutritional Supplementa- per worker annually. Assuming that this wage gain begins at tion and Birth Weight of Offspring." AmencanJournal of Cinical Numtion the age of eighteen, that is, seventeen years after the food 34:21 33-44. supplement was received, and continues to age fifty-five, and Agarwal, D. K., and others. 1987. "Nutritional Status, Physical Work Capac- using a discount rate of 3 percent and an annual increase in ity. and Mental Function in School Children. Scientific Report 6. Nutri- real wages of 2 percent, we arrive at $1 74.00 as the present tion Foundation of India, New Delhi. value of the gain per worker, or a benefit-cost ratio of 1 7.4. For Atderman, Harold. 1987. Cooperative Dairn- Development in Kar-nataka, India value of the gain per worker, or a benefit-cost ratio of 17.4. For An Assesssnent. Research Report 64. Intemational Food Policy Research the purpose of sensitivity analysis, we estimated the benefit- Institute. Washington, D.C. cost ratios associated with height gains of 1.0 and 3.0 centime- . 1990. "Nutritional Stattis in Ghana and Its Determinants." Working ters per 100,000 calories. The corresponding benefit-cost ratios Paperon Social DimensionsofAdiustment in Sub-Saharan Africa 3. World are 8.7 and 26, respectively. Bank, Washington, D.C. -- . In press. 'Information as an Input into Food Policy Formation." In Per Pinstrup-Andersen, David Pelletier, and Harold Alderman, eds., Estimating Cost per Child Death Averted by Supplement Beyolnd Chdid Survival: Enhancing Child Growth and Nutinoi in Developing Counmes. Assuming a normal distribution and the standard deviation Alderman. Harold. and Joachim Von Braun. 1984. The Effecus of the Egyptian reported by WHO (1979), we estimated the prevalence of severe Food Ration and Subsidy System on Income Distibution and Consumption. and moderate malnutrition from the average weight of the Research Report 45. Intemational Food Policy Research Institute, Wash- and mooerate malnutrition from tne average weight of the e ington, D.C. preschool children participating In food supplementation .4Alderman, Harold, Joachim Von Braun, and Sakr Ahmed Sakr. 1982. Egypt's (table 18-9) prior to the supplementation (7.6 kilograms) and Food Subsidy and Rationing S,ystern: A Desciption. Research Report 34. after a hypothetical average weight gain of 0.5 kilograms, intemnational Food Policy Research Institute, Washington, D.C. which according to table 18-9 is brought about by a transfer of Aligaen, M., and C. Florenclo. 1980. "Intra-household Nutrient Distribution 100,000 calories. Based on the different mortality rates re- and Adequacyof Food andNutrient Intakeof FilipinoUrban Households." ported by Pelletier (1991) for severe and moderate under- PhilippiMe.lournalofNumntion 33:11-19. weight and using the prevalence of underweight reported in Anand, R. K. 1981 "The Management of Breastfeeding in a Bombay Hospi- table 18-3, we estimated the effect of a weight gain of 0.5 tal."nAssi Annme,itChaldren5s/56:i67-a0. kilograms on the overall mortality rate to be 1.03 percentageAnn, James Austin Joe D. Wray, Maia F. Zeitin. 19. Numtion Intervention in DevelopingCounmes, Study 1: Supplementary Feed- points. Considering this estimate as the reduction in the prob- ing. Cambridge: Oelgeschlager, Gunn. and Ha. ability of dying, we estimated that the average cost of averting Ashworth, A., and R. G. Feachem. 1985. "Interventions for the Control of a child death would be $1,942, the cost of transferring O 00,000 Diarrhoeal Disease Among Young Children: Prevention of Low Birth calories times 100 divided by 1.03. Weight." Btlletn of the World Health Organization 63:165-84. Protein-Energy Mvalnutrition 415 Bairagi, Radheshyam, Mridul K. Choudhury, Young ]. Kim, and George T. Chavez, A., C. Martinez, and R. Yaschine. 1975. "Nutritional Behavioral Curlin. 1985."AlternativeAnrthTopometriclndicarorsofMortalitv."Amer- Development and Mother-Child Interaction in Young Rural Children." icanJournalofClinicalNu-rition42(2):296-306. Federation of American Societies for Experimental Biology. Federation Balderston, J. B., A. B. Wilson, M. E. Friere, and M. S. Simonen. 1981. Proceedings34(7):1574-82. MalinouishedChildren of the Rural Poor. Boston: Auburn House. Chen, Lincoln C. 1983. "Interactions of Diarrhea and Malnutrition." In Barrell, R. A. F., and M. G. M. Rowland. 1979. "Infant Foods as a Potential Lincoln C. Chen and Nevin S. Scrimshaw, eds., DiarrheaandMalnumnon: Source of Diarrheal Illness in Rural West Africa." Transactions of the Royal Interactionts, ,Mechanisms, and Interventions. New York: Plenum Press. Society of Tropical Medicine and Hygiene 73:85-90. Chen, Lincoln C., A. K. M. A. Chowdhury, and S. L. Huffman. 1980. Beaton, George H., and Hossein Ghassemi. 1982. "Supplementary Feeding "Anthropometric Assessment of Energy-Protein Malnutrition and Subse- Programs for Young Children in Developing Countries." AmencanJournal quent Risk of Mortality Among Preschool-Aged Children." AmericanJour- of Cinical Numntion 35:864-916. nal of Clinical Nutriton 33:1836-45. Beaudry-Darisme, M., and M. C. Latham. 1973. "Nutrition Rehabilitation Chen, Lincoln C., Emdadul Huq, and Stan D'Souza. 1981. "Sex Bias in the Centers: An Evaluation of Their Performance."Journal of Tropical PediatTies Family Allication of Food and Healthcare in Bangladesh." Population and and Environmental Child Health 19:299-332. Development Review 7 ():5 5-70. Beghin, 1. D. 1970. "Nutritional Rehabilitation Centers in Latin America: Chen, Lincoln C., Emdadul Huq, and Sandra L. Huffman. 1981. "A Prospec- Critical Assessment." AmeencanJournalofClimical Nutrition 23:1412-17. tive Study of the Risk of Diarrheal Diseases According to the Nutritional Beghin, I. D., and R. E. Viteri. 1973. "Nutritional Rehabilitation Centres: An Status of Children." American Jourmal of Epidemiology 114:284-92. Evaluation of Their Performance."Journal of Tropical Pediatrics and Environ- Chen, Robert S_. ed. 1990. "The Hunger Report: 1990." Brown University, mieital Child Health. 19:404-16. Alan Shawn Feinstein World Hunger Program, Providence, R.]. Behrnman, Jere. 1988. "Nutrition and Incomes: Tightly Wedded or Looselv Chowdhury, A. K. M. A. 1988. "Child Mortality in Bangladesh: Food Versus Meshed?" Pewv/Comell Lecture Series on Food and Nutrition Policy. Ith- Health Care." Food and Nutrition Bulletin 10(2):3-8. aca, N.Y.: Comell Food and Nutrition Policy Program. Clavano, N. R. 1982. "Mtode of Feeding and Its Effect on Infant Mortality and Beisel, W. R. 1977. "Resume of the Discussion Concerning the Nutritional Morbidity."JournalofTropical Pediat-ics 28:287-93. Consequences of Infection." American Journal of Clinical Nutrition Cook, R. 1971. "Is Hospital the Place for Treatment of Malnourished Chil- 30(8):]294-1300, August. dren!" Journal of Tropical Pediatrics and Environmental Child Health Berg, Alan. 1987. Malnutrition: What Can Be Done? Baltimore: Johns Hopkins 17(1):15-25. Universitv. Cremer, H. D., A. F16rez, L. de Navarro, L. Vuori, and M. Wagner. 1977. "The Berggren, Gretchen. 1981. "Home-Prepared Food Supplements, Mothercraft Influence of Food Supplementation and/or Psychological Stimulation on Centers, and Nutrition in Haiti." Food and Nutrition Bulletn 3(4):29-33. Mental Development." Nutrition and Metabolism 21:358-71. Black, R. E., Kenneth H. Brown, and Stan Becker. 1984. "Malnutrition is a Csitting, W. A. Ni. 1983. "Nuttitional Rehabilitation." In D. S. McLaren, ed., Numnton Inl the Community. New York: John Wiley and Sons. Determining Factor in Diarrheal Duration, But Not Incidence, in a Longi- tudinal Studv in Rural Bangladesh." Amencan Journal of Clinical Nutrition Davies, D. P., P. Platts, J. Ni. Pritchard, and P. W. Wilkinson. 1979. "Nttri- 39(1):87-94. tional Status of Light-for-Date Infants at Birth and its Influence on Early Black, Robert E., Kenneth H. Brown, Stan Becker, A. R. M. Abdul AIim, and Postnaial Growth." Archives of Disease in Childhood 54:703-6. Michael H. Merson. 1982. "Contamination of Weaning Foods and Trans- Delgado, H. L., Reynaldo Martorell, E. Brineman, and R. E. Klein. 1982. mission of Enterotoxigenic Escherichia coli Diarrhoea in Children in Rural "Nutrition and Length of Gestation." Nutriion Research 2:117-26. Bangladesh." Transactiots of the Royal Society of Tropical Medicine and Delgado, H. L., V. E. Valverde, Reynaldo Martorell, and R. E. Klein. 1982. Hygiene 76:259-64. "Relationship of Maternal and Infant Nutrition to Infant Growth." EarlI Bouis, Howarth E., and Lawrence J. Haddad. 1990. Effects of Agricultural Human Development 6:273-86. Commercialization on Land Tenure, Household Resource Allocation, and Nu- Den Hartog, A. P. 1972. "Unequal Distribution ot Food Within the House- tntion in the Philippines. Research Report 79. International Food Policy hold." Numtion Newsletter 10:8-17. Research Institute, Washington, D.C. Deolalikar, A. B. 1988. "Nutrition and Labor Productivity in Agriculture: Brown,KennethHi.,RobertE. Black,GuillermoLopezdeRoniafia,andHilary Estimates for Rural South India." Review' of Economics and Statistics Creed de Kanashiro. 1988. "Infant Feeding Practices and Their Relationship 70(3):406-13. with Diarrheal and Other Diseases m Huascar (Lima), Penu." Pediatrics 83:31-40. De Sweemer, Cecile. 1974. "Growth and Morbidity." Ph.D. diss., Johns Burger, Susan, and S. A. Esrey. In press. "Water and Sanitation: Health Hopkins University. Benefits to Children." In Per Pinstnmp-Andersen, David Pelletier, and Edirisinghe, Neville. 1987. The Food Stamp Scheme in Sn Lanka: Costs, Benefits, Harold Alderman, eds., Beyond Child Survival: Enhancing Child Growth and and Options for Modification. Research Report 58. International Food Policv Nutntion in Developing Countries. Research Institute, Washington, D.C. CaIdwell, B. M. 1974. "The Malnourishing Environment." In]l. Cravioto, L. El Samani, F. Z.. W. C. Willett, and J. H. Ware. 1988. "Association of Hambraeus, and B. Vahlquist, eds., Early Malnutrition and Mental Develop- Malnutrition and Diarrhea in Children Aged Under Five Years." American nient. Uppsala: Swedish Nutrition Foundation. Journal of Epidemiology 128:93-105. Carlson, Beverley A., and Tessa M. Wardlaw. 1990. "A Global, Regional, and Engle, PatrIce. In press. "Child Care and Preschool Nutrition." In Per Pinstrup- Country Assessment of Child Malnutrition." l!NICEF Programme Division. Andersen, David Pelletier, and Harold Alderman, eds., Beyond Child Sur- Staff Working Papers 7. New York. vival: Enhancing Child Grouwth and Nutrition in Developing Countes. Cerqueira, Maria Teresa. and Christine M. Olson. "Nutrition Education in Esrey, SA.., R. G. Feachem, and J. Ml. Hughes. 1985. "Interventions for rhe Developing Countries." In press. In Per Pinstrup-Andersen, David Pelletier, Control of Diarrhoeal Disease Among Young Children: Improving Water and Harold Alderman, eds., Beyond Child Survival: Enhancing Child Growth Supplies and Excreta Disposal Facilities." Bulletin of the World Health Orga- and Nutrition in Developing Countries. nizanon 63:757-72. Chavez, Adolfo, and Celia Martinez. 1975. "Nutrition and Development of Esrev, S. A., and Jean-Pierre Habicht. 1985. "The Inspact of Improved Water Children from Poor Rural Areas v. Nutrition and Behavioral Develop- Supplies and Excreta Disposal Facilities on Diarrheal Morbidity, Growth, ment." Nutrition Reports International 2(6):477-89. and Mortality Among Children." In S. A. Esrey, Jean-Pierre Habicht, and 416 Per Pinstrup-Andersen, Susan Burger, Jean-Pierre Habtcht, and Karen Peterson W. P. Butz, eds., A Methodology to Review Public Health Interventions: Results Haaga, J. G., and J. B. Mason. 1987. "Food Distribution Within the Family: from Nutrition Supplementanon and Water and Sanitation Projects. Cornell Evidence and Implications for Research and Programmes." Food Policy Intemational Nutrition Monograph Series 15. Ithaca, N.Y.: Comell Food 12(May):146-60. and Nutrition Policy Program. Habicht, Jean-Pierre, Julie DaVanzo, and William P. Butz. 1988. "Mother's . 1986. "Epidemiologic Evidence for Health Benefits from Improved Milk and Sewage: Their Interactive Effects on Infant Mortality." Pediamcs Water and Sanitation in Developing Countries." Epidemiology Reviews 81:45641. 8:117-28. Haddad, LawrenceJ., and Howarth E. Bouis. 1990. "The Impact of Nutriional Esrey, Steven A.. J. B. Potash, L. Roberts, and C. Shiff. 1990. Health Benefits Status on Agriculmtral Productivity: Wage Evidence From the Philippines." from Iinpro'ements in WVater Supply and Sanitation: Survey and Analysis of the Oxford Bulletin of Economic Statistics 53()1:45-68. Literature on Selected Diseases. Technical Report 66. Water and Sanitation Hamilton, Sahni, Barry Popkin, and Deborah Spicer. 1984. Women and for Health Project, Arlington, Va. Nutrition in Third WXorld Countries. New York: Praeger Press. FAQ (Food and Agriculture Organization). 1987. Fifth Annual World Food Harbert, Lloyd, and Pasquale Scandizzo. 1982. "Food Distribution and Nutri- Survey. Rome. tion Intervention: The Case of Chile." Working Paper 512. World Bank, Feachem, R. G., and NI. A. Koblinsky. 1984. "Interventions for the Control Washington, D.C. of Diarrhoeal Diseases Among Young Children: Promotion of Breastfeed- Hardy, Ellen E., Ana M. Vichl, Regina C. Sarmento, Lticila E. Moreira, and ing." Bulletin of the World Health Organization 62:271--91. Celia M. Bosqueiro. 1982. "Breastfeeding Promotion: Effect of an Educa- Garcia, Marito. 1988. "Food Subsidies in the Philippines: Preliminary Results." tional Program in Brazil." Studies in Family Planning 13(3):79-86. In Per Pinstrup-Andersen, ed., Food Subsidies in DevelopingCountries: Costs, Herrera, M. G. 1985. "Maternal Nutrition and Child Survival." Paper pre- Benefits, and Policy Options. Baltimore: Johns Hopkins University Press. sented to UNICEF Conference on Child Health and Survival, Harvard Garcia, Marito, and Per Pinstrup-Andersen. 1987. The Pilot Food Price Subsidy School of Public Health. May 30-31, Boston. Scheme in the Philippines. Its Impact 07o Income, Food Consumption, and Herrera, M. G., J. 0. Mora, B. de Paredes, and M. Wagner. 1980. "The Effects NutritionalStatus. Research Report 61. International Food Policy Research of Nutritional Supplementation and Early Education on Physical and Institute, Washington, D.C. Cognitive Development." In Ralph R. Turner and Hayne. W. Reese, eds., Ghal, 0. P., and V. N. Jaiswal. 1970. "Relationship of Undemutrition to Life-Span Developmnental PsNchology: Intervention. New York: Academic Diarrhea in Infants and Children." IndianJournal ofMedical Research 58:789. Press. Gibbonis, Gayle, and Marcia Griffiths. 1984. "Program Activities for Improv- Ho, Teresa J. 1985. Economic Issues in Assessing Nutrinon Projects: Costs, ing Weaning Practices." Information for Actioni Issue Paper Prepared for AffordahilitvandCost Effectiveness. Population, Health, and Nutrition Tech- UNICEF World Federation of Public Health Associations, Geneva. nical Note 85-14. World Bank, Washington, D.C. Gilmore, J. W., C. C. Adelman, A. J. Meyer, and M. C. Thome. 1980. Hornik, R. C. 1985. Numtion Education--A State of the Art Review. State of "Morocco: Food Aid and Nutrition Education." Project Impact Evaluation the Art Series. Nutrition Policy Discussion Paper 1. Administrative Com- 8. United States Agency forInternational Development, Washington, D.C. mittee on Coordination, Subcommittee on Nutrition, United Nations, Gomez, F., R. Galvan, S. Frenk,J. C. Munfoz, R. Chavez, andJ. Vazquez. 1956. New York. "Mortality in Second and Third Degree Malnutrition." Journal of Tropical Huffinan, Sandra, and Adwoa Steel. In press. "The Impact of Child Survival Pediatncs 2:77. Interventions on Cisild Growth." In Per Pinstrup-Andersen, David Pelle- Gonzalez-Cossio, T., Jean-Pierre Habicht, H. Delgado, and K. M. Rastiussen.. tier, and Harold Alderman, eds., Beyond Child Survival: Enhancing Child 1991. "Food Supplementation during Lactation Increases Infant Milk In- Growth and Numnoi in Devetoping Countries. take and the Proportion of Exclusive Breast Feeding." Federation of Anier- Johnson, David W., and Roger T. Johnson. 1985. "Nutrition Education: A ican Societies for Experimental Biology. FASEF lournal. Niodel for Effectiveness, A Synthesis of Research." Journal of Numnton Gopalan, C., and Meera Chatterjee. 1984. Use of Grou th Charts for Promoting Educatioii 1 7(supplement 2)1-44. ChildNurrition:A Re'iewofGlobalExperience. Special Publication 2. Nutri- Jumaan, A. O., M. K. Serdula, D. F. Williansson, M. J. Dibley, N. J. Binkin, tioii Foundation of 1India, New Delhi. and J.J. Boring. 1989. "Feeding Practicesand Growth inYemeniiChildren." Gopalan, C., M. C. Swaminatlhan, V. K. Krishna Ktiniari, D. Hanumrantha Jounsol of Tnpical PediatTics anid Environmental Child Health 35:82-86. Rao, and K. Vilayaraghavan. 1973. "Effect ofCalorie Supplementation on Kasongo Project Team. 1983. "Anthropometric Assessment of Young Growth oBf Undemokirished Children." American jvournal of Cllilncal Numntion Children's Nutritional Status as an Indicator of Subsequent Risk of Dying." 26:56 3-66. Journal of Tropical Pediatrics and Environmental Child Health 29:69-75. Grantham-McGregor, Sallv, William Schofield, and Linda Harris. 1983. Kate., R. W., Robert S. Chen, T. E. Downinig, Jeanne X. Kasperson, "Effect of Psychosocial Stimulation on Menital Development ot Severely Ellen Messer, and Sarah R. Niillnsais. 1988. The Hunger Report: 1988. Malnourished Children: An Interim Report." Pediatrics 7(22):239-43 Brown University, Alan Shawn Feinstein World Hunger Program, Provl- Grantham-McGregor, Sally, Marie E. Stewart, Christine Powell, and W. N. dence, R.I. Schofield. 1979. "Effect of Stimulation on Mienital Development of Mal- Keninedy, Eileeni T. 1989. The Effects of Sugarcasse Production o7 Food Secunt's, noLurished Child." Lancet 2:200-1. Health, and Nutrition in Kenya: A Longitudinal Analysis. Research Report 78. Grantham-McGregor, Sally, Marie E. Stewart, and W. N. Schofield 1980. Intemational Food Policy Research Institute, Washington, D.C. "Effect of Long-Term Psychosocial Stimulation on Mental Development of Kennedy, Eileen T., and Flarnld Alderman. 1987. Comparative Analysis of Severely Malnourished Children." ILancet 2:785-89. Nutritional Effectiveness of Foood Subsidies and Other Food-Related Interven- Graves, P. L. 1976. "Nutrition, Infant Behavior, and Maternal Characteristics: tions International Food Policy Research Institute, Washington. D. C. A Pilot Study in West Bengal, India." AmericanJournalofClinicl Nuntionl Kennedv, Eileen T., and Bruce Cogill. 1987. Incomne and Nutrition Effects of 29:305-19. Suga,caIe Production in Southwestern Keriya. Research Report 63. Intema- Griffiths, Marcia. 1985. "Gr,wth Monitoring ofPreschool Children: Practical tional Food Poilicy Research Institute, Washington, D.C. Consideratiotns for Prinsary Health Care Projects." Information for Action Kennedy, Eileen T.. and Milton Kotelchuck. 1984. "The Effect of wlc Sup- Issue Paper prepared for t.NICEF World Federatioln of Public Health Ass."- plemental Feeding on Birth Weight: A Case-Control Analysis." American ciations, Geneva. Journal of Clin1ical Nutrition 40(3):579-85. Owatkin, Davidson, Janet R. Wilcox, and Joe D. WraTy. 1980. Can Health and Kennedy, Eileen T., C. Overholt, and C. Haddral. 1984. Effect ofa Milk Subsidy Nutrition Intiervention Make a [i .. . Ov ersea; Development Council, on the Dismbution of Benefirs within the Household. International Food Policy Washington, D.C. Research Institure, Washington. D.C. Protein-Energy Malnutrition 417 Keusch, G. T.. and N. S. Scrimshaw. 1986. "Control of Infection to Reduce mentation: Effects on Child Stunting Because of Diarrhea." Amencan Malnutrition." In J. A. Walsh and K. S. Warren, eds., Strategies for Prrnary Joumal of Ctinical Nutntion 50:1-8. Health Care. Chicago: University of Chicago Press. MacCormack, C. P. 1988. "Health and the Social Power of Women." Bnnsh Kielmann, A. A., C. E. Taylor, R. L. Parker, W. A. Reinke, D. N. Kakar, and Journal of Science and Medicine 26:677-83. R. S. S. Sarma. 1983. "Child and Matemal Health Services in Rural India: IMcDonald, E. C., E. Pollitt, W. Mueller, A. M. Hsueh, and R. Sherwin. 1981. The Narangwal Experiment." In Integrated Nutrition and Health Care. 1. "The Bacon Chow Study: Effect ofNutritional Supplementation on Mater- Baltimore: Johns Hopkins tJniversity Press. nal Weight and Skinfold Thicknesses During Pregnancy and Lactation." Kielmann, A. A., and C. McCord. 1978. "Weight-For-Age as an Index of Risk British Journal of Numtion 51:357469. Of Death in Children." Lancet2:1247-50. McGuire, Judith S. 1988. Malnuttion: Opportunities and Challenges for Klein, R. E., H. E. Freeman, J. Kagan, Charles Yarbrough, and Jean-Pierre A t D. United States Agency for International Development, Washing- Habichr. 1972. "is Big Smart? The Relation of Growth to Cognition." ton, D.C. Journal of Healthand Social Behavior 13:219. MCGuire, Judith S., and J. E. Austin. 1987. "Beyond Survival: Children's Koster, F. T., C. C. Curlin, K. M. A. A7zt, and H. Azizul. 1981. "Svnergistic Growth for National Development." In AssignmentChildren. 2:1-51. Impact ofMeaslesandDiarrheaonNutritionandMortalityinBangladesh." McGuire, Judith S., and Barry M. Popkin. 1990. Helping Women Improve Bulletin /fthe World Health Orgaiiizanon 59:901. Numtion in the Developing World: Beating the Zero Sum Game. Technical Kramer, 5M. S. 1987. "Determinants of Low Birth Weight: Methodological Paper 114. World Bank, Washmgton, D.C. Assessment and Meta-Analysis." Bulletin of the World Health OrganiZation McKay, H., and L. Sinisterra. 1974. "Intellectual Development of Malnour- 65:663-737. ished Preschool Children in Programs of Stimulation and Nutritional Kumar, S. K. 1979. Impact ofSubsidized Rice on Food Consuniption and Nutrition Supplementation." In Joaquin Cravioto, Leif Hambraeus, and Bo Bahlquist, in Kerala. Research Report 5. International Food Policy Research Institute, eds., EarlY Malnutritnon and Mental Development: Symposia of the Swedish Washington, D.C. Nuttinon Foundation 12. Swedish Nutrition Foundation Stockholm: Alm- Kurz, K. MN., K. M. Rasmussen, and Jean-Pierre Habtcht. 1990. "Matemal quist and Wiksell. Nutritional Status Has a Small Influence on Length of Postpartum Amen- McKay, Harrison, Leonardo Sinisterra, Arlene McKay, Hemrando Gomez, and orrhea among Guatemalan Women." Federation of American Societies for Pascuala Lloreda. 1978. "Improving Cognitive Ability in Chronically De- Experimental Biology. FASES Journal 4(4):A 1160. prived Children." Science 200:270-78. Lambert, J. 1988. "Pakistan: Update on Breastfeeding." Mothers and Children McLaren, Donald Stewart, Emmanuel Shirajian, Hermine Loshkajin, and 7(2):5-6. Sossy Shadarevian. 1969. "Short-Term Prognosis in Protein-Calorie Mal- Lamptey, Peter, and Fred Sai. 1985. "Integrated Health/Nutrition/Population nutrition." Amencan Jounmal of Clinical Numtion 22:863-70. Programmes." In Margaret Biswas and Per Pinstrup-Andersen, eds., Num- Marks, G. C. 1989. "The Relationships of Body Weight and Composition in non and Development. London: Oxford Universitv Press. Preschool Children." Ph.D. diss., Cornell University. Latham, M . 1985. "The Relationship ofNutrition to Productivity and Wellbe- Martines, J. C., Jean-Pierre Habicht, A. Ashworth, and B. R. Kirkwood. 1988. mg of Workers." Paper presented to the Workshop on Political Economy of "Exclusive Breast-Feeding and Growth: When Should Supplements be Nutrition Improvement, June 10- 13, 1985, Coolfront, W.V. Introduced." Typescript. Lawrence, M. and R. G. Whirehead. 1988. "Physical Activity and Total Martorell, Reynaldo. In press. "Promoting Healthy Growth: Rationale and Energy Expenditure ofChild-BearingGambian Village Women." European Benefits." In Per Pinstrup-Andersen, David Pelletier, and Harold Alder- JourLal ofClinical Numtnon 42:145-60. man, eds., Beyond Child Survival: Enhancing Child Growth and Nutrinon in Lechtig, Aaron, Jean-Pierre Hahicht, Herman Delgado, Robert E. Klein, DevelopingCountries. Charles Yarbrough, and Reynaldo Martorell. 1975. "Effect of Food Martorell, Reynaldo, and G. Arroyave. 1984. "Malnutrition, Work Output, Supplementation During Pregnancy on Birth Weight." Pediarrics 56: and Energy Needs." Paper presented at the International Union of Biolog- 508-20. ical Sciences Symposium, Ciba Foundation, London. Lechtig, A., and R. E. Klein. 1979. "Maternal Food Supplementation and Martorell, Reynaldo, Jean-Pierre Habicht, and R. E. Klein. 1982. "Anthropo- lnfant Health: Results of a StuLdy in Rural Areas of Guatemala." In Hugo metric Indicators of Change in Nutritional Status in Malnourished Aebi and Roger G. Whitehead, eds., Maternal Nunition During Pregnancy Populations." In Barbara Underwood, ed., Proceedings Methodologies for and Lactation. Bern: Hans Huber. Human Population Studies in Nutrntion Relaoed to Health. National Institutes Leslie,Joanne. 1982. "Child Malnuitrition and Diarrhea: A Longitudinal Study of Health Publication 82-2462. Washington, D.C.: U.S. Govemment from Northeast Brazil." Ph.D. diss , Johns Hopkins University School isf Printing Office. Hygiene and Public Health. Martorell, Reynaldo, Jean-PierTe Habicht, Charles Yarbrough, A. Lechtig, Leslie, Joanne. In press. "Improving the NuItrition of Wonmen in the Third R. E. Klein, and K. A. Western. 1975. "Acute Morbidity and Physical World." In Per Ptnstrup-Anderseni, David Pelletier, and Harold Alderman, Growth in Rural Guatemalan Children." Amencan Journal of Diseases of eds., Beyvonid Child Survival: Enhancitig Child Growrh anid Numrioil mt Decel- Children 1291(1):12 96-1301. oping Countries. Martorell, Reynaldo, R. E. Klein, and H. Delgado. 1980. "Improved Nutrition Lindskog. U., P. Ltndskog, j. Carstensen, Y. LarSson. and M. Gebre-Medhin. and Its Effects on Anthropometric Indicators of Nutritional Status." Num- 1988. "Childhood Mortality in Relation to Nutritional Status and Water tion Reports International 21(2):219-30. Supply: A Prospective Study from Rural Malawi." Acta Pediatrica Niartorell, Reynaldo, J. Rivera, and C. K. Lutter. 1988. "Interaction of Diet Scandinavica 77:26L-68. and Disease in Child Growth." Paper presented at the 4th Intemational Lurter, C. K. 1987. "Nutritional Supplementation: When is It Most Effective Human Lactation Meeting, Nov. 16-19, 1988, San Jose, Costa Rica. in Promoting Growth Among Malnourished Infants and Children." Ph.D. Martorell, Reynaldo, and Charles Yarbrough. 1983. "The Energy Cost of diss., Cornell Universitv. Diarrheal Diseases and Other Common Illnesses in Children." In Lincoln Lutter, C. K.. J.O. Mora,Jean-Pierre Habicht, K. NI. Rasmussen, D. S. Robson. C. Chen and Nevin S. Scrimshaw, eds., Diarrhea and Malnumntion: Interac- and M. G. Herrera. 1990. "Age-Specific Responsiveness of Weight and tions, Mechanisms, and Interventions. New York: Plenum Press. Length to Nutritional Supplenmenitation." AmencarJournal of Cfinical Nu- Mason, John B., Jean-Pierre Habicht, H. Tabatabai, and V. Valverde. 1984. trition 51:359464. NumnonalSurveillance. Geneva: World Health Organization. Lutter, C. K., J 0. .Mora, Jean-Pierre Habicht, K. M. Rasmussen, D. S. Robson, Mata, L. J. 1978. The Children of Santa Maria Cauque: A Prospective Field Studoy S. G. Sellars, C. NI. Super, and M. G. Herrera. 1989. "Nutitional Supple- of Health and Growth. Cambridge, Mass.: Ml t Press. 418 Per Pinstrup-Andersen, Susan Burger, Jean-Pierre Habicht, and Karen Peterson - . 1983. "Promotion of Breastfeeding, Health, and Growth Among Payne, Philip R. 1985. "The Nature of Malnutntion." In Margaret Biswas and Hospital-BornrNeonates, and Among Infants of a Rural Area ofCosta Rica." Per Pinstrup-Andersen, eds., Nutrition and Development. Oxford: Oxford In Lincoln C. Chen and Nevin S. Scrimshaw, eds., Diarrhea and Malnutrn- University Press. ion: IlnteTrCtions,Mechanisms, and InterTenaons New York: Plenum Press. Pelletier, David. 1991. The Relationship Betweeni Child Anthropornetri and Mata, L. J., R. A. Kromall,J. J. Urrutia. and B. Garcia. 1977. "Effect otlntection Mortality in DeelopingCountries: Implications for Polic, .Programs, and Future on Food Intake and the Nutritional State: Perspectives as Viewed from the Research Cornell Food and Nutrition Policy Program Monograph Series 12. Village." American Journal of Clinical Nutnnon 30:1215-271. Ithaca, N.Y. Mata, LeonaTdo J., Juan Jos6 Urrutia, Constantino Albertazzi, Olegario . In press. "The Role of Information in Community-Based Nutrition Pellecer, and Eduardfo Arellano. 1972. "Influence of Recuirrent Infections Programs." In Per Pinstmup-Andersen, David Pelletier, and Harold Alder- on Nutrition and Growth of Children in Gulatemala." Amencain Journal of man, eds., Beynond Child Survival: Enhancing Child Growth and Nutrition in Clinical Nutrition 25:1267-75. Developing Countries. Mebrahtu, Saha, David Pelletier, and1 Per Pinstrnp-Andersen. In press. "Agri- Peterson, K. E., and D. A. Frank. 1987. "Feeding and Growth of Premature culture and Nutrition." In Per Pinstrup-Anderseni, David Pelletier, and and Small-ForGestational-Age Infants." In William Taeusch and Michael Harold Alderman, eds., BeNond Child Survital Enhancing Child Growth anid Yogman, eds., Folflow-Up Management of the High-Risk Infant. Boston: Little, Nutntioni in Developing Cutiines. Brown. Merchant, Kathleeni, Reynaldo Mariorell, and Jere Haas. 1990a. "Conse- Peterson, K. E., M. G. Herrera, and C. M. Super. 1988. "Sequelae of Growth quences for Maternal Nutrition of Reproductive Stress Across Consecutive Delay in Colombian Children Aged 0-36 Months." Paper presented to the Pregnancies." American journal (f Clinical Numrion 52:616-20. Annuial Meeting o f the American Public Health Associationt, Boston. . 1990b. "Maternal and Fetal Responses to the Stresses of Lactation1 Peterson K. E., C. M. Super, and M. G. Herrera. 1988. "Biological and Social Conrcurrent with Pregniancv and of Short Rectiperat ive Intervals." American Predictors of Nutritional Status at 6 Months of Age in Colombian Children Journal ofClinical Nutntion 52:280-88 at High Risk Of Malnutrition." Proceedings of the Fifth International Merchant, Kathleen, and others. 1988. "Frequent Reproductive Cycling." Congress of Auxology, Exeter, England. Progress in Foiod and Nutrition 12(4). Pinstrup-Anidersen, Per. 1982. Agricultural Research and Technology in Economic Mora, Jose. 1983. "Supplementary Feeding During Pregnaiicy: Impact on Development. New York: Longman. Motlierand Childin Bogota, Colombi,." In B. A. Underwood, ed., NuiTnon ---. 1985. "Food Prices and the Poor in Deselisping Countries." European Intervention Strategies in National Devel)pment New York: Nutrition Fotin- Review of Agricultural Economics 12(78):69-81. dation, Academic Press. - -, ed. 1988. Food Subsidies in Developing Countries: Costs, Benefits, and - . In press. "Integrated Progranss foT Primary Health Care and lnsproved Policy Optionis. Baltimore: Iohns Hopkins Universiry Ptess. Nutrition." In Per Pinstrup-Andersen, David Pelletier, and Harold Alder- Pinstrup-Andersen, Per, and Marito Garcia. 1990. "Data on Food Consump- man, eds., BeNond Child Survival: Enhancing Child Grow th and Numtion in tion bv High-Risk Family Members: Its Utility foTr Identifying Target House- DevelopingCountries. holds foir Food and Nutrition Programmes." In Beatrice Lorge Rogers and Mora, J. O. .1. Clement, N. Christiansen, J. Suescun, M. Wagner, and M. G. Nina P. Schlossman, eds., Intra-Household Resource Allocatoii. Issues anid Herrera. 1978. "Nutritioilal Supplementation and the OuItcome of Preg- Methods for Development Policy and Planning Tokvo: United Nations Uni- nancy. 3. Perinaral and Neonatal MNortality." Nunitisn Reports lntemational versity Press. 18(2):167-75. Pinstrup-Andersen, Per, and Peter B. R. Hazell. 1985. "The Impact of the Mora, J. O., L. de Navarro, J. Clement, M. Wagner, B). de Paredes, and M. G. Green Revolutiisn and Prospects for the Future." Food Reviews International Herrera. 1978. "The Effect ofNtirritional Supplemenitatiinon Calorie and (l):1-25. Protein Intake of Pregnant Women." Nutrition ReporTt Ititernational Pollitt, E. 1987. "A Critical View ofThree Decades of Research on the Effects 17(2):217-28. of Chronic Energy Malnutrition on Behavioral Development." In Blat Mora, J. O., B. de Paredes, M. Wagner, L. de Navarro, J. Suescun, N. Schurch and Nevini S. Scrimshaw, eds., Chronic Energy Deficiency Comse- Christiansen, and M. G. Herrera. 1979. "Nutrition Supplementation and quences anid Related Issues. Lausanne, Switzerland: International Dietary the Outcome of Pregnancy. I. Birth Weight." Anierican lournal f Clinical Energy Consultancv Group (IDEC). Nutrition 32:455-62' Prentice, A. M., T. J. Cole, F. A. Foord, W. H. Lamb, and R. G. Whitehead. Mora,J. O., M. G. Herrera,J. Suescun, L. de Navarro, and M. Wagner. 1981. 1987. "Increased Birth Weight After Prenatal Dietary Supplementation of "The Effects of Nutritional SupplementatioIn on Physical Growth ofChil- Rural African Womeni." AmericanrJournal of Clnical Nutrition 46:912-25. dren at Risk of Malnutrition." Amencan Journal of Clinical Nutnntion Prentice, A. M., S. B. Roberts, and A. Prentice. 1983. "Dietarv Supplementa- 34:1885-92. tion of Lacrating Gambian Women. 1. Effect on Breast-Milk Volume and National Academy of Sciences. 1988. Amno Acid Fortification of Cereals: Qulalty." Human Numtion: Clinical Numntion 37:53-64. Results and Interpretation of Trials in Three Countries: A Report of the Task Prentice, A. M., Susan B. Roberts, M. Watkinson, R. G. Whitehead, Alison Force onAminoAcidFornfication o(f Cereals. Cornell Ilnternatinal Nutrition A. Paul. Ann Prentice. and Anne A. Watkinison. 1980. "Dietary Supple- Monograph Series 20. Committee on International Nutrition Programs, inentation of Gambian Nursing Mothers and Lactational Performance." Food anid Ntitrition Board, Commission on Life Sciences, National Re- Lancet 2(October-December 1980):886-88. search Council. Ithaca, N.Y. Prentice, A. NI., R. G. Whitehead, M. Watkinson, W. H. Lanmb, and T. I. Cole. Naylor, A. J., and R. A. Wester. 1985. "Lactation Specialist Training for 1983. "Prenatal Dietary Supplementation of African Women and Birth- Health Professionals From Developing Natisns." StImmnary Report of Ses- Weight." Lancet 1:489-92. sions 14, August 1983-January 1985. San Diego Lactation Program. P'uffer, R. R., and C. V. Serrano. 1987. Patterns of Birth Weight. Scientific Nutrition Economics Group. 1982. Intra-Farmniy Foiid Distribution: Revlew of PtLblication 504. Pan American Health Organization, Washington, D).C. the Literature and Policy Implicanons United States Departnient of Agricul- Rasmussen, K. M., N. B. Mock, and Jean-Pierre Habichi. 1985. "The Biolog- ture, Office of Internatioinal Cooperation and Developpment, Technical ical MeaningofLow Birth Weight and the UseoifData on Low Birth Weight Assistance Division, Washington, D.C. for Nutritional Surveillance." Working Paper Series 27. Ithaca, N.Y.: Cor- Overholt, Catherine, Stephen G. Sellers, Jiise 0. Mora, Belen de Paredes, and nell Nutritional Surveillance Program. M. Guillermo Herrera. 1982. "The Effects of Nutritional Supplementation Rathhun, J M., and K. E. Peterson. 1987. "Nutrition in Failure-to-Thrive." In on the Diets of Low-Income Families at Risk of Malnutrition." Anerican R. F. Grand,J. L. Sutphen, and W. H. Dietz, eds., PediamcNumtion. Boston: Journal of Clinical Numntio 36:1153-61. Butterwiirths. Protein-Energy Malnutrition 419 Richardson, S. A. 1980. "The Long Ranige Consequences of Malnutrition in Stephenson, L. S. 1987. The Impact of Helminth lnfectons on Human Nunition: Infancv: A Study of Children in Jamaica, West Indies." In B. Wharton, ed., Schistosomes and Soil-Transmitted Helminths. London: Taylor and Francis. Topics In Pediamcs 2. Nutrition in Childhood. Tunrlridge Wells, Eng.: Strauss,John.1985."TheImpactotfImprovedNutrition onLaborProductivity Pitmani Medical. .nd Humani Resource Development: An Economic Perspective." Paper Rivera, Juan, Jean-Pierre H,ibichr, and Douglas Robson. 1991. "Effect of presented to the Workshopon the Political EconomyofNutrition Improve- Supplementary Feeding Upon Recovery from Mild-to-Moderate Wasting slicnt, ]Jue 10-13, 1985, Coolfront, W.V. in Preschool Children." AmericanJournal of Clinical Nutrition 54(l):62-68. . 1986. "Does Better NLutrition Raise Farm Productivity"' Journal of Rivera,J., Reynaldo Martorell,and C. K. Lutter. 1989. Interaccidnde la Ingesta Political Ecootnoy 94:297-320. Dietetcay la EfemiedadDiarreicaenelC( recimtenti)de 'os Nai's Institute oif Super, C. M., S. G. Sellers, ! . NMora. B. de PaTedes, and M. G. Herrera. 1989. Nutrition nof Central America and Panama, Guatemala Citv, Guatemala. "Weaning, Diarrhea, and the Transirion to a Family Diet Among Colonm- Roberts, S. B. 1982. "Seasonal Changes in Activity, Birth Weight. and bian Infants at Risk of Malnutrition." Typescript. Lacrational Performance in Rulral Ganibian Women." Transactions rcf the Svedberg, Peter. 1989. "Unidemtitrition in Sub-Saharan Africa: Is There a Sex Rwyal Society of Tropical Medicine and Hygiene 76:668-78. Bias?" Helsiniki: Wider. Rothe, G. E., K. MN. RasmTIssen, and Jean-Pierre Habicht. 1988. "The Predic- Tinner,]. Ni., and A. M. Thompson. 1970. "Standards for Birth Weight at tors of Growth Respone to Supplementation in Malnourished Childreni." Gestation Periods From 32 to 42 Weeks, Allowing for Matemal Height and Types,cript. Weight-" Archites of Disease in Childhood 45:566. Rowland, M. G Ni., S. G. 1. (G. Rowland, and T. I Cole. 1988. "Impact of Taylor, C. E., and E. M. Taylor. 1976. "Multifactorial Causation of Malnutri- Infection on the Growth of Childreni From 0 to 2 Years in an Urban West tion." In D. S. McLaren, ed., Numtioii in the Community. London: John African Community." AmencanJlumnal of Clinical Nutrition 47:134.-38. Wilev and Sons. Rowland, M. G. M., T. J. Cole, and R. G. Whirehead. 1977. "A Quantitative Thompson, A. MN., W. Z. Billewic:, and F. E. Hytten. 1968. "The Assessment Studv into the Role of Intection in Determining Nutritional Status in of Fetal Growth." BntishJountal of Obstetrics and Gynecology 75:903-16. Gambian Village Children." BritishJournal o] Numtion 37:441-50 T omkins.Andrew. 1981. "NutritionalStatusandSeveritvyofDiarrhoeaamong Ruel, Marie. In press. "Grlsh Morniroring as an Educational Tool. an Inte- Preschiool Children in Rural Nigeria." Lancet 1 860-62. grating Strategy, and a Source of Information: A Review of Experience." lII Torche, A. 1981. Evalulcitin Econ6mica del Programa Nacioiial de Almentiacto Per Pinstrup-Aindersen, David Pelleticr, and Harold Alderman,eds., Beyond Compimantana (rr'Ar. Institute of Economics, Ctholic University of Child Survival. Enhancing Child Grouth aiid Nutrirtion iti DevelopingCounries. Chile,Santiag. Rush,D.,Z.Stein,and. NI.Susser I980a.'DietinePregznancy: A Randomized Tucker, Katherineg David Pelletier, Kathleen Rasmussen, Jean-Pierre Controlled Trial of Nwtritional Supplenents." (Original Article Series) Habicht, Per Pinstrup-Anderseni, and Frederick Roche. 1989. Advances in Birth Defects 16:13). New York: Alan Liss. Inc. Nutritional Surveillance The Cornell Nutritional Surveillance Program 1981- 1980h. "A Randumized Coritrolled Trial of Prenatal Nutritional 1987. Monograph 2. Ithaca, N.Y.: The Cornell Food and Nutrition Policy Supplementation in New York City." Pediatrics 65:683--97. Program. Sahn, David E., ed. 1989. Seasonal aranabilitc its Third World Agriculture. LNK:EF(UnitedNationsChildren'sFund). 1988. 1983-1988 EvaluanonReport Baltirnore: Johns Hopkmits Universitv Press. of the Joint tV-u/tj,\'fo.EF Nutnition Support Programme in lnnga, Tanzania. -- . 1990. Mlalnutrition im C5thd'lvoire. Preailenceand Determimiin.nts. Work- Government of the United Republic of Tanzania, World Health Organiza- ing paper on Social Dimensions of Adlustrnent in Suh.Saharan Africa 4. tion, and United Nations Children's Fund, Dar es Salaam. World Bank, Washingtoni, D.C. - 1989. The State of the World's Children. New York: Oxford University Sahn, David E., and Harold Alderman. 1988. "The Effects of Human Capital Press. onWagesand theDeterininantsoiLaborSupply ina DevelopingCountry." United Nations. 1987. "First Report on the World Nutrition Situation." journal of Development Ecottmiics 29(12): 157-84. Administrative Committee on Coordination, Subcommittee on Nutritioni, Scrimshaw,N.S.,C.E.Taylor,anidj E.GGordon. 1968.IiiteracmionsoiNurntion New York. and Infectiots. MonIograph 57. World Health Organization, Geneva. . 1990. "Summarv Reporr of the Sixteenth Session of the A cc./Subconm- Selowsky, Marceloi. 1981. "Nutrition, Health, and Education: The Economic mittee on NutritiJmn, FebRiary 19-23, 1990, Paris." UNE,CO. Administrative Significance of Complementarities at an Early Age." Journul of Dev.elopment Comiimtitee on Coordination, Subcommittee on Nutrition, Uniited Nations, Econoimics 9:331-46. New York. Selowsky, Marcelo, and Lance Taylor. 1973. "The Economics of Malnourished . 1992. "Seconid Report on the World Nutrition Situation." AUCC/SCN, Children: An Example of Disinvestment In Humanl Capital." Economic New York. Development and Cultural Chatige 22(2): 17-30. *SAID (United States Agency for Intemattional Development). 1988. Growth Sepilveda, Jainie, Walter Willett, and Alvaro Mufioz. 1988. "Malnuitrition Nionitonng and Nutrition Ediucanon: Impact Evaluation of an Effective Applied and Diarrhea." Anericati Journal of Epidemiology 127:365-76. Nutrition Program in the Dominican Republic. Bureau for Science and Tech- Shekar, Nicera. 1991. The Tamril Nadu Integrated Nutrition Project: A Review of nology, Office of Nutrition, Washington, D.C. the Project with Special Emnphasiss on the Morittotimig and litformation System. Van Esterik, P. 1984. Intra-Family Food Distribution: Its Relevtance for Maternal Working Paper 14. Cortiell Food and Nutrition Policy Program, Ithaca, N.Y. and Child Nutrition. Working Paper 18. The Comell Nutritional Surseil- Sommer, Alfred, and Nlatthew S. Loewenstein. 1975. "Nutritional Status and lance Program, Ithaca, N.Y Mortality: A Prospective Validation of the QU.Ai: Stick." Arnencan Jounial Victora, C. G(., J. Patrick Vaughan, C. Lombardi, S. M. C. Fuchs, L. P. Gigante, of Clinical Nutrition 28:287-92. P. G. Smith, L. C. Nobre, A. M. B. Teixeria, L. M. Moreira, and F. C. Banos. Spun, G. B. 1983. "Nutritional Status and Physical Work Capacity." Yearbook 1987. "Evidence for Protection by Breastfeeding Against Infant Deaths irom of Physical Anthropol)gy 26:5-35. Infectious Diseases in Brazil." Lancet2(August):319--21. . 1984. "Physical Activ'ity, Nutritional Status, and Physical Work Villar, J., L. .Altobelli, E. Kestler, and J. Belizan. 1986. "A Health Priority for Capacitv in Relation to Agricultural Productivity." In Ernestr Pollitt and Developing Countries: The Prevention of Chronic Fetal Malnutrition." Peggy Amanitr, ed,., EieTg-y Intake and Activity. N ew York: Alan R. Liss. Bulletin of the World Health OTganiezatiuTi 64:847-51. Spurr, G. B., NI. Barac-Nieto, and M. G. Maksud. 1977. "Productivitv anld Villar,J., andJ. Belizan. 1982. 'The RelativeContribution ofPrematurity and Maximal Oxygen Consumption in Sugar Cane Cutters.." American Journal Fetal Growth Retardationi to Low Birth Weight in Developing and Devel- of Clnical Nuxtrition 30:316-21. oped Societies." Anericanjournal of Obstetrics aid Gynecology 37:142 44. 4210 Per Pinstrnp-Andersen, Susan Burger, Jean-Pierre Habicht, and Karen Peterson V\ilIar, J., V. Smerighlo, Reynaldo Martorell, C. H. Brown, and R. E. Klein. sion, Technical Department, Latin America and the Caribbean Region, 1984. "Heterogeneous Growth and Mental Development of Intrauter- Washington, D.C. ine Growth-Retarded Infants During the First 3 Years of Life." Pediarncs . 1990. "INDIA-Tamil Nadu Integrated Nutrition Project Comple- 74783. rtion Report." World Bank, Washington, D.C. V\on Braun, Joachim, Hartwig de Haen, and Jurgen Blanken. 1991. Commer- WHO (World Health Organization). 1979. Measurement of Nutn-itonal Impact. cialization of Agriculture under Conditions of Population Pressures. Production, Report 79. 1. Geneva. Consumption, and Nutritional Effects in Rwanda. Research Report 85. Inter- . 1981. "The Treatment and Management of Severe Protein-Energy national Food Policy Research Institute, Washington, D.C. Malnutrition." Geneva. Von Braun, Joachim, David Hotchkiss, aid Maarten Immink. 1989. Nontra- - 1986. "Use and Interpretation of Anthropometric Indicators of ditional Export Crops in Guatemala: Effects on Production, Income, and Nutri- Nutritional Status." Bullentin of the World Health Organization 64:929-41. tion Research Report 73. International Food Policy Research Institute, Washington, D.C. ~~~~~~~~~~~Wyon, J. B., and J. F. Gordon. 1971. The Khanna Study. Cambridge, Mass.: Washingtoin. D.C.Harvard University Press. Von Braun, Joachini. Detlev Puetr, and Patrick Webb. 1989. Irriganon Tech- HarvadOUiversityaPress. nologs and Commercialization of Rice in The Gambia: Effects oni Income and Yambi Olivia. 1988. "NutritionalStatus andthe Risk ofDeath: AProspective Numtion. Research Report 75. International Food Policy Research Insti- Study of Children Six to Thirty Months Old in Iringa Region, Tanzania." rote, Washington, D.C. ~~~~~~~~~Ph.D. diss., Cornell University. Lute, Washingtoii, D.C. Warerlow, J. C., M. H. N. Golden, and J. Patrick. 1978. "Protein-Energy Yambi, Olivia, Urban Jonsson, and Bjorn Ljungqvist. 1989. The Role of Malnutririon: Treatment." In John W. T. Dickerson and H. A. Lee, eds., Government in Promoting Community-Based Nutrition Programs: Experience The Clinical Management of Disease. London: Edward Amold. From Tanzania and Lessons for Afnca. Pew/Comell Lecture Series on Food Whitehead, R. G. 1973. "Nutrition Education Research." World ReviewJ of and Nutrition Policy. Ithaca, N.Y.: Cornell Food and Nutrition Policy Nutn-tion and Dietetics 17:91-149. Program. Zeitlin, Marian, and Candelaria S. Formaci6n. 1981. "Nutrition Education." . 1977. "Infection and Developnent of Kwashiorkor and Marasnius in In Nutrition Interventions in Developing Countries. Cambridge, Mass.: Africa." Ainencanjournal of Clinical Nutrinon 30:1281-85. Oelgeschlager, Gunn, and Ham. Williams, Cicely D. 1962. "Malnutrition." Lancet 2:342-44. Zeitlin, Marian, M. Griffiths, R. K. Manoff, and T. M. Cooke. 1984. "House- Winick, Myron, Knarig K. Meyer. and Ruth C. Harris. 1975. "Malnutrition hold Evaluation, Nutrition Communication, and Behavior Change Com- and Environmental Enrichment by Early Adoption." Science 190:1173-75. psnent." Indonesian Nutrition Development Program. New York: Manoff World Bank. 1989. Feeding Larn America's Children. Hunsan Resources Divi- Intemational. 19 Micronutrient Deficiency Disorders Henry M. Levin, Ernesto Pollitt, Rae Galloway, and Judith McGuire Throughout the developing world high incidence of disease is Iron Deficiency associated with inadequate intake or absorption of micronutri- ents. Relatively minute quantities of each of the micronutri- Iron is a mineral present in the body as a constituent of ents are required for normal health status and well-being. If hemoglobin and in some enzyme and electron carriers. Because these nutrients are deficient, there are serious consequences it cannot be made in the body, iron, like all essential nutrients, for health, mental and physical. Because of known prevalence must be obtained from food. rates of deficiencies and ways of addressing them, we will focus in this chapter on three micronutrients: iron, iodine, and MAGNITUDE AND DISTRIBUTION. It is generally thought that vitamin A. Deficiencies in other micronutrients are biologi- iron deficiency anemia is the most common nutritional defi- cally significant-such as the effect of zinc on growth in certain ciency in many developing countries, second only to protein- disease states (Nishi and others 1980; Daeschner and others energy malnutrition (PEM) (Florentino and Guirriec 1984). 1981), and the effect of vitamin B-6 on cellular immunity Age and physiological status determine the degree of vulnera- (Talbot, Miller and Kerkulirt 1987; Tomkins and Watson bility of the individual: rapidly growing infants, children, and 1989)-but we know little about the extent of these deficien- pregnant and lactating women are at high risk for deficiency. cies or how to attack them. Our purpose in this chapter, The Subcommittee on Nutrition of the Administrative Com- therefore, is to focus on iron, iodine, and vitamin A deficien- mittee on Coordination (ACC/SCN) of the United Nations cies and to assess global significance, public health importance, estimates that 1.3 billion people suffer from iron deficiency methods of prevention, and priorities within an economic anemia (United Nations 1990). In table 19-1 this figure is framework in which costs, cost-effectiveness, and cost-benefit disaggregated by region. Table 19A- 1 contains available infor- analyses are used. Much of what we learn about iron, iodine, mnation on the prevalence of iron deficiency anemia by coun- and vitamin A will be useful in addressing other essential try, and table 19A-2 includes definitions for iron deficiency by deficiencies in the future. age and sex. We will begin by discussing the global and public health significance of deficiencies in these micronutrients, including CAUSES. Severe iron deficiency results in anemia (low he- the magnitude and distribution of the deficiencies, their moglobin level), which impairs the transport of oxygen and causes, and the implications for human health and develop- basic cell functions. People with mild iron deficiency may not ment. We will then describe what is currently known about have low hemoglobin (Hb) levels but yet have reduced body short- and long-term strategies and the proportion of potential iron stores (ferritin) (Shils and Young 1988). "Iron deficiency" beneficiaries actually covered by these programs. In the final refers to any depletion of ferritin. An individual may be iron section we will discuss issues for governments and other agen- deficient without manifesting iron deficiency anemia, but all cies in setting priorities for operations, institutions, allocation those with iron deficiency anemia are iron deficient. Anemia of resources, and research. can also result from other nutrient deficiencies (for example, folate or vitamin B- 12) or genetic abnormalities (for example, Public Health Significance thalassemia). Folate deficiency is very commonly associated with iron deficiency anemia, so therapy usually includes a Micronutrient deficiencies are manifested by an array of disor- combined iron-folate tablet. ders that have serious social, private, and economic costs to In populations in which the prevalence of iron deficiency society. anemia is high, the deficiency usually is the result of the 421 422 Henry M. Levin, Ernesto Pollitt, Rae Galloway, and Judith McGwure Table 19-1. Estimated Prevalence of Anemria by Geographic Region, Age, and Sex, 1980 Children Children Men Pregnant women All women 0-4 years ____ 5-12 years 15-59 years 15-49 years 15-49 years Number Number Number Number Number Country Percent (millions) Percent (millions) Percent (millions) Percent (millions) Percent (millions) Africa 56 48.0 49 47.3 20 23.4 63 11.3 44 46.8 Latin America 26 13.7 26 18.1 13 12.8 30 3.0 17 14.7 East Asian 20 3.2 22 5.6 11 6.1 20 0.5 18 8.4 SouthAsia 56 118.7 50 139.2 32 123.6 65 27.1 58 191.0 Developingregionsa 51 183.2 46 208.3 26 162.2 59 41.9 47 255.7 Note: Anemia is defined as a hemoglobin concentration below WHO reference values for age, sex, and pregnancy statuLs. a. Excluding China. Source: DeMaeyer and Adiels-Tegman 1985. interaction between dietary factors, chronic iron loss due to maternal and infant iron deficiency. Increased prenatal and parasitic infections (forexample, hookworm orschistosomiasis perinatal risk (low birth weight, prematurity, and mortality) [Stephenson 1987]), or elevated needs (for example, during has been associated with low levels of hemoglobin and hema- pregnancy and periods of rapid growth). tocrit in the mother (Murphy and others 1986; Lieberman and It is currently believed that diet is the most important factor others 1987, 1988; Brabin 1988). determining iron status. Dietary factors include insufficient In infants (6-24 months) and preadolescent children (9-11 iron in the diet and poor bioavailability of dietary iron. There years), iron deficiency anemia is associated with mild growth are two types of iron in foods: heme iron (present in animal retardation (Lozoff 1982; Aukett and others 1986; Chwang, flesh), of which about 20 to 30 percent is absorbed, and Soemantri and Pollitt 1988). Treatment of iron-deficient ane- nonheme iron (present in plant sources), of which less than 5 mic infants (age 17-19 months [Aukett and others 19861) and percent is absorbed. Iron in breast milk is highly bioavailable, preadolescent children (age 8.2-13.5 years [Chwang, and up to 50 percent is absorbed (INACG 1979). People in Soemantri, and Pollitt 1988]) has resulted in increased growth developing countries derive most of their iron from nonheme during the period of intervention. The causes of this growth sources, whereas those in industrial countries consume greater retardation may be related to the general role of iron as an amounts of heme iron. The key dietary difference in iron status essential metabolic cofactor, its relation to immunocompe- is its bioavailability rather than the absolute amount of iron in tence (Higashi and others 1967; Klebanoff 1970; Chandra foods (INACG 1989). Absorption of nonheme iron can be 1973; Mata 1977; Chandra and Puri, 1985; DalIman 1987), or improved up to 18 percent (DeMaeyer 1989) by the addition its role in appetite, which decreases during iron deficiency to the diet of ascorbic acid or foods containing ascorbic acid (Basta and others 1979). (Hunt and others 1990) or other acids, the addition to the diet Iron deficiency anemia poses a developmental risk for cog- of foods containing heme iron , or the removal from the diet nitive dysfunction (for example, attention and concentration) of substances that inhibit iron absorption (DeMaeyer 1989). in preschool and school-age children, and that risk factor is Some types of food processing, such as the fermentation of soy sufficiently severe to jeopardize educational attainment. The products, also seem to improve iron bioavailability (Macfar- strongest and most consistent evidence of the effects of iron lane and others 1990). The recommended intake of iron for deficiency on cognition is found in clinical trials of preschool people by age and sex is given in table 19A-3. The upper value and school children who have been assessed for specific mental is needed by people who consume mainly nonheme iron. processes (for example, attention and concept formation) and Although parasitic infections also contribute to iron defi- school achievement. One example is a study by Soewondo, ciency, the treatment of such infectious parasites alone is not Husaini, and Pollitt (1989), which showed that anemic chil- the most cost-effective means of addressing iron deficiency dren three to six years old learned faster and formed appropri- anemia, because unless the parasites are removed, reinfection ate concepts more efficiently after they were supplemented will take place. Hygiene education, footwear, and improved with iron than their placebo-fed anemic controls. In another water supply and sanitation are needed (DeMaeyer 1989). set of studies, researchers in India (Seshardri and Gopaldas 1989) showed that when anemic children age five to fifteen IMPLICATIONS FOR HUMAN HEALTH AND DEVELOPMENT. It had were supplemented with iron, they performed better on tests been assumed that iron deficiency in pregnant women did not for intelligence quotient (JQ), memory, visual perceptual orga- put the fetus at risk because the fetus would have priority access nization, and clerical tasks improved than did anemic children to matemal iron stores (Bothwell and Charlton 1981). In a who received a placebo. study in Benin, however, Hercberg and others (1987) found Similar effects have been observed in the mental develop- that when multiple indicators of iron status were used to assess ment scale scores of iron-deficient anemic infants (Lozoff and maternal anemia there was a positive correlation between Brittenham 1985; Pollitt 1987). Iron-deficient anemic infants Micronument Deficiency Disorders 423 perform more poorly than iron-replete infants on the Bayley ity in picking tea leaves. In Kenya, Stephenson and others Scale of Mental and Motor Development (Lozoff 1989; Walter (1985) found an association between the intensity of the 1989), but the behavioral response to iron therapy is not infection with Schistosoia haematobium, a direct cause of iron consistent across different studies (Lozoff and Brittenham deficiency anemia, and physical fitness among school-age 1985; Lozoff and others 1982; Oski and others 1983; Aukett children. and others 1986; Lozoff 1989). No changes in mental test performance have been found among infants and children Iodine Deficiency whose iron stores and circulating iron are depleted but whose Hb levels have remained constant. Conceivably, these Iodine, a mineral, is a component of two thyroid gland hor- changes are too subtle to be detected in small samples. mones which are necessary for normal metabolism. Higher morbidity has been noted in anemic pregnant women (Fleming 1989). Reasons for this may be that iron MAGNITUDE AND DISTRIBUTION. The number of people esti- deficiency influences the risk of infection in distinct ways. It mated to be at risk of disorders from severe iodine deficiency is associated with abnormalities in cell-mediated and nonspe- is 680 million in Asia, 227 million in Africa, and 60 million cific immunity (Higashi and others 1967; Klebanoff 1970; in Latin America (see table 19-2). Prevalence is high in Chandra, 1973; Prasad 1979; Chandra and Puri 1985; Dallman mountainous and flood-prone areas where iodine-deficient 1987). The product ion of T cells is specifically compromised soils prevail. Although no age group or sex is immune to iodine (Srikantia and others 1976; Bagchi, Mohanram, and Reddy deficiency, the fetus, women, and children seem to be most 1980), and the capacity of neutrophils to kill bacteria is signif- vulnerable to serious and irreversible consequences of defi- icantly diminished during iron deficiency (Walter and others ciency (Herzel 1988). A detailed disaggregation of the preva- 1986). Excess free iron in the serum has been associated with lence of iodine deficiency by country is given in table 19A-4. predisposition to infection, but this state results largely from injected iron rather than ingested iron (DeMaeyer 1989). CAUSES. The term "iodine deficiency disorders" (IDDs) covers Because iron deficiency anemia compromises im- the breadth ofsequelae and is not limited to severe deficiency munocompetence, it is likely to increase mortality among (Het:el 1983). Iodine deficiency disorders result from inade- high-risk groups. This increase may not be attributable to quate intake of iodine either because the soils and water are immunodeficiency alone but also to circulatory failure (INACG iodine deficient or because certain naturally existing 1989). The exact role that iron deficiency plays in mortality "goitrogens" in foods interfere with the individual's use of needs further definition. iodine. Iodine is essential for the formation of thyroid hor- There is a good deal of evidence that relates maternal mones [thyroxine (T4) and 3,5,3'-triiodothyronine (TDJ, mortality to severe anemia. In Maharashtra, India, 90 percent which are necessary for normal growth and development and of all maternal deaths occurred in women with Hb levels of less for proper metabolic function. Recommended daily intake of than 7 grams per hundred milliliters of blood (Masani 1969, iodine is about 150-300 micrograms (see table 19A-5). When cited in Fleming 1989). In Nigeria, Harrison (1975) found that the thyroid gland does not obtain enough iodine to make these 4 percent of mothers with severe anemia (Hb levels of less than hormones, it increases in size to compensate for the deficiency. 5 grams per hundred milliliters of blood) died in childbirth. The enlarged thyroid gland is called goiter. Some evidence suggests that 20 percent of all maternal deaths in West Africa and India (when blood transfusion was not IMPLICATIONS FOR HUMAN HEALTH AND DEVELOPMENT. Iodine available) were directly attributable to anemia and that addi- deficiency affects reproduction. Some types ofanovulation can tional mortality resulting from hemorrhage was indirectly be reversed by desiccated thyroid, which confirms the relation- caused by maternal anemia (Fleming 1989). Iron deficiency impairs work performance through its effects Tae 19-2. Estimated Prevalence of Iodine on Hb and, possibly, myoglobin, which is involved in the Deficiency Disorders and Population at Risk, transport of oxygen in muscle. There is a high negative corre- L Re lation between Hb levels (grams of Hb per 100 milliliters of (millegion blood) and the percentage of increase in heart rate of a person on a treadmill (Gardner and others 1977). A strong positive Overt relationship exists between Hb levels and potential maximum Regio __ At risk Goiter cretinism workload as measured in the Harvard Step Test. Performance Southeast Asia 280 100 4.0 improves after iron supplementation increases Hb to the ex- Rest of Asia 400 30 0.9 pected normal level (Scrimshaw 1984). Africa 227 39 0.5 Iron deficiency anemia adversely affected the work produc- Latin America 60 30 0.3 tivity of Indonesian rubber tappers until they received iron Eastem Mediterranean 33 12 supplementation (Basta and others 1979). In a recent study in Total 1,000 211 5.7 Indonesia, Suhardjo (1986) found that, following iron treat- -Negigible. ment, iron-deficient anemic w omen increased theirproductiv- Snurce: WHO 1990. 424 Henrn M. Lein, Errnesto Pollitt, Rae Galloway, and Judith McGuire ship, known even in ancient times, between the thyroid gland phylaxis found significant positive effects of the intervention. and fertility (McMichael, Potter, and Hetzel 1980). In animals The difference in these outcomes is probably related to the there is evidence of a significant increase in spontaneous nature of the particular function of the central nervous system abortions and stillbirths and a marked reduction in brain that was assessed and to the mechanism and timing of the growth in the fetus when the mother is iodine deficient (Hetzel deficiency. Some hearing loss associated with acquired hypo- and Potter 1983). Conclusive evidence of these effects on thyroidism is correctable by increased iodine intake. Still, mild human reproduction is weak; however, reduction in stillbirths mental retardation resulting from impaired structural develop- and perinatal mortality and increased birth weight in Zaire ment of the brain during fetal life is probably irreversible. Thus, (McMichael, Potter, and Hetzel 1980; Thilly 1981, as cited in addressing iodine deficiency in reproductive-age women is of Hetzel 1987), Papua New Guinea, Ecuador, and Peru highest priority. (Clugston and others 1987) have been observed after im- Because of the relation between IDD and intellectual capac- plementation of iodine deficiency control programs. In Zaire ity, productivity is adversely affected by iodine deficiency. the infant mortality rate for mothers given iodine supplemen- After a salt iodization program in one Chinese village, the tation during pregnancy was significantly less than for those average income increased from 43 yuan per person in 1981 to who were not given iodine (Thilly and others 1980). 223 in 1982 to 414 in 1984, which was higher than the average Severe iodine deficiency in utero can result in postnatal for the district (Levin 1987). In addition, because mental and dwarfism, and people with goiter can be retarded in their physical fitness improved, cereals were exported for the first physical development (Hetzel 1988). In addition to hindering time and men were fit enough to join the army after salt growth, iodine deficiency also increases morbidity rates in iodization. In Ecuador, Greene (1977) found that people with children, especially from respiratory infection (Tomkins and moderate iodine deficiency were consistently paid less for Watson 1989). Phagocyte dysfunction (Chandra and Au agricultural work than normal individuals. In summary, iodine 1981) and delayed immune response (Marani,Venturi, and deficiency has been associated with impaired reproduction, Nasala 1985) have been reported. severe and mild mental retardation, growth inhibition, and It is known that the thyroid of the fetus does not become reduced productivity. activated until the tenth week (Delong 1987); before that, development depends on the thyroid hormone of the mother Vitamin A Deficiency (Hetzel and Potter 1983). Deleterious effects of matemal io- dine deficiency on the fetus during this time include reductions Like all essential vitamins, vitamin A is an organic substance in brain DNA and RNA (Hetzel and Potter 1983). Extreme which the body cannot produce. Vitamin A is essential for deficiency results in severe mental retardation, known as cre- normal vision, growth, and immune function and to maintain tinism, which can take several forms. The seriousness of cre- epithelial cells. tinism is obvious, but from an economic development perspective the greatest concern in endemic areas is the possi- MAGNITUDE AND DISTRIBUTION. An estimated 42 million bility that even noncretinous children may be mentally and children under the age of six have mild or moderate xeroph- neurologically handicapped. This milder impairment might rhalmia (West and Sommer 1987). About 250,000 to 500,000 remain unnoticed within the community, but it can limit children go blind annually and approximately 50 to 80 percent the social and economic growth of these communities ofthosethatgoblinddiewithinoneyear(Sommer1982;IvAcG (Stanbury 1987). 1989). In tables 19-3 and 19A-6 the prevalence of vitamin A In endemic areas the performance of noncretinous children deficiency is given by region and country. The age groups at in cognitive and motor tests correlates positively with the highest risk for vitamin A deficiency are young children be- mother's thyroid levels during the pregnancy of the respective yond weaning age (six months to six years), although older offspring. For example, in the Westem highlands of Papua children and pregnant and lactating women are also affected. New Guinea, serum thyroxine (T4) during pregnancy was Prevalence peaks among two- to-four-year-old children (East- related to the offspring's performance at twelve years of age on man 1987). In some parts of the world it appears that boys are tests of visual perceptual organization and visual motor coor- at higher risk than girls (Sommer 1982; Tielsch and Sommer dination (Pharoah and others 1984). 1984; DeMaeyer 1986), which may be a reflection of different Given the serious effects of iodine deficiency on the proper cultural practices in rearing children or physiological differ- functioning of the brain, it is not surprising to learn that even ences (Sommer 1982). Prevalence is also greatest in low- mild deficiency may have irreversible effects. A study in the income groups and during those seasons when food sources of highlands of Bolivia of goitrous children age five and a half to the vitamin are scarce (Mamdani and Ross 1988). In table twelve years failed to show a clear effect of iodized oil supple- 19A-2, definitions for vitamin A deficiency are listed. Note mentation given twenty-two months earlier on IQ, visual motor that the population percentages in this table refer to those with coordination, and school performance (Bautista and others severe vitamin A deficiency. There are currently no values for 1982). In contrast, the authors of a study in the Guizhou mild vitamin A deficiency. province of China (Yan-You and Shu-Hua 1985) of the effects of iodized salt on the hearing of otherwise normal seven- to- CAUSES. Xerophthalmia and its cure through the diet were eleven-year-old children one, two, and three years after pro- recognized in ancient times. It is only in the early part of this Mic-ronutrient Defrczencs Disorders 425 Table 19-3. Estimated Prevalence of Vitamin A cited in Mamdamni and Ross 1988); partly by an inadequate Deficiency, 1984-85 source of fat, which facilitates the absorption of the carot- Children with enoids (United Nations 1985); and partly by protein inade- Countries Children aged mild to moderate quacy, which hinders vitamin A absorption, release, and with 1-4 years deficiencyb transport (Mamdami and Ross 1988). Vitamin A transport is Region ___ deiciency' (millions) (millions) affected only when protein deficiency is severe (Somnmer Africa 16 53 7.9 1982). Intestinal abnormalities caused by bacterial infection Americas 5 38 5.6 or parasites can also interfere with the absorption of vitamin Southeast Asiac 5 51 7.7 A. Respiratory infection and other diseases can increase the India I 111 16.7 requirement for the vitamin and interfere with its intake Mediterranean 3 7 1.0 through decreased appetite (United Nations 1985). In Africa, Western Pacific 4 20 3.0 imeasles is often a precipitating factor in blindness due to Total 34 280 41.9 vitamin A deficiency (Eastman 1987). ------ - - - -- ----- In addition, there are social and economic factors related to a. Countries in WHO category A (significant proiblem wilth control pro- grams in place 1984-85): Bangladesh, El Salvador. Haiti, India, Indonesia, the deficiency (Mamdami and Ross 1988). For example, in Nepal, Philippines, Sri Lanka. Category B (significant problem but no con- urban Bangladesh a clear negative association has been noted trol program 1984-85): Benin, Brazil (northeast states), Burkina Faso, Ethi- between per capita income and vitamin A deficiency (Stanton opia, Malawi, Mali, Mauritania, Mexico, Oman, Sudan. Tanzania, Vier Nam, Zambia. Category C (probable problem but no assessment or program and others 1986). Within low-income groups, however, such 1985-85): Afglhanistan, Angola, Botivia, Burma, Chad (north), Ghana a correlation is often less obvious. Intrafamily food distribution (north), Kanipuchea. Ken"a, Lao P.DR., Mozambique, Niger, Nigeria patterns and, as noted, the low social value attributed to foods (north), Uganda. h. Assuimes 15 percent prevalence. rich in vitamin A rnay determine risk in vulnerable groups. c. Excluding India. Seasonal availability of vegetables and fruits often acts Source: West and Sominer 1987. synergistically with other factors to precipitate deficiency. In countries such as the Philippines and Indonesia, dark green, leafy vegetables are generally unavailable when infec- century, when vitamin A was discovered, that the deficiency tions peak. was first described in connection with physical growth and, later, with vision. In the more recent past greater attention has IMPLICATIONS FOR HUMAN HEALTH AND DEVELOPMENT. When been placed on the association between the deficiency and pregnant women are deficient in vitamin A, severe xeroph- infant and child morbidity and mortality. Traditionally, vita- thalmia may develop in utero (Sommer 1982), which increases min A deficiency has been defined as a severe reduction in vulnerabilitytoinfectionanddeath.VitaminAdeficiencyalso vitamin A reserves along with clinical signs of the deficiency. decreases fertility (Eastman 1987). Congenital malformations Milder depletion may also be defined as deficiency, even have also been related to experimentally induced vitamin A though it does not result in changes in the eyes, because it may deficiency in animals. Less is known about the effects of still have an important relationship to morbidity and mortality vitamin A deficiency on congenital malformations in human (West and Sommer 1987). Biochemically, even mild signs of beings, although some suggestive clinical observations exist deficiency in children are detected by a decrease in vitamin A (Wallingford and Underwood 1986), such as congenital xe- reserves, with liver and serum levels less than 20 micrograms rophthalmia, anophthalmia, microphthalmia, and other ocu- per deciliter (see table 19A-2). lar defects (IVACG 1986). Vitamin A overdose during early Dietary sources of vitamin A include preformed vitamin A pregnancy may cause fetal absorption (Eastman 1987). For this (retinol) from animal sources and beta-carotene and other reason it is generally recommended to give pregnant women carotenoids found in plant sources, which can be converted to no more than 1,000 micrograms of vitamin A at a time or to vitamin A in the body. Retinol is the most active form of treat existing deficiencies with dietary sources of vitamin A vitamin A, followed by beta-carotene and then the other (IVACG 1986). carotenoids (see table 19A-5 on how these three compare). The ocular signs of vitamin A deficiency fall under two Vitamin A deficiency is caused by dietary inadequacy (see categories: (a) night blindness due to the interruption of the table 19A-5 for recommended intakes), by increased physio- dark adaptation process in the visual cycle; and (b) structural logical requirements, and by cultural factors which determine changes in the surface of the eye (Bitot's spots, drying, kerato- individual availability and consumption. In countries in which malacia, ulceration, and so on) due to loss of secretory function the staple food is rice, a cereal without vitamin A, low-income in the mucosal epithelium in the conjunctiva of the eye and groups are at high risk of vitamin A deficiency. Even in changes in the differentiation or maturation of specific epithe- countries in which sources of the carotenoids exist, there may lial cell types (Mamdami and Ross 1988). Night blindness, be high incidence of deficiency. In parts of Indonesia, for conjunctival dryness, and Bitot's spots are generally reversible, example, dark green, leafy vegetables (a rich source of beta- but more advanced stages are not and can result in lesions carotene) are commonly available, yet vitamin A deficiency is which cover the entire cornea, causing partial or total blind- highly prevalent. This coincidence is partly explained by the ness. Once the disease has reached this extreme level of low social value attributed to green vegetables (WHO 1982, as severity, the life of the child is endangered. 426 Henry n.M. Letm, Ernesto Poll,tt Rae GallowaiY, andJudith McGuire Both mild and severe forms of vitamin A deficiency are vitamin A deficiency (Pollitt 1990). This question is particu- associated with increased morbidity, especially from respira- larly troublesome in Africa, where comeal scarring associated tory and diarrhealdisease (Sommer, Katz, and Tarwotjo 1984). with measles is frequent (WHO/EPI 1988b). In Malawi and The effects of vitamin A deficiency on cell-mediated immu- Tanzania, for example, half of the children attending schools nity, antibodies, and secretory antibodies have been docu- for the blind reported a history of measles preceding total mented using both animal models and clinical data (Nauss blindness (WHO/EPI 1988b). 1986; Olson 1986; Chandra 1988). Conclusive information on To summarize, vitamin A deficiency has been linked with its adverse effects on phagocytosis is not available. interference with ocular function, impaired growth and repro- It is known that measles and vitamin A deficiency have a duction, and increased morbidity and mortality. Deficiencies complex reciprocal interaction. Vitamin A status is a determi- in the three micronutrients are manifested by overlapping nant of the outcome of measles, especially in Africa, and conditions. In table 19-4 we review the deficiency conditions measles, in turn, is a forceful precipitating factor in blindness for all three and show the similarities in the conditions for the (Eastman 1987). For example, measles was related to bilateral micronutrients. comeal ulceration in 78.9 percent of these cases (Ksanga, Pepping, and Kavishe 1985, as reported by Eastman 1987). In Micronutrient Interactions fact, the peak for vitamin A deficiency coincides with the peak for measles in children (Eastman 1987). An association be- The interaction ofmicronutrients can be viewed in two differ- tween vitamin A deficiency and measles has also been found ent ways. Deficiencies of the three micronutrients under dis- in Asia. In studies in Bangladesh (Cohen and others 1985) and cussion interact according to their geographic setting and Indonesia (Sommer 1982), 10 percent and 37 percent, respec- according to how the micronutrients are metabolized. tively, of children with keratomalacia had had measles within Iron, iodine, and vitamin A deficiencies often occur in the previous four weeks. The intensity of the infection, and countries in which poverty limits dietary sources and in which not a specific characteristic of measles, seems to be the deter- geography limits the composition of food that normally would mining factor in the high mortality observed among vitamin contain these micronutrients. Generally, these three micronu- A deficient children (WHO/UNICEF 1987; WHO/EPI 1988b). The trient deficiencies occur simultaneously in certain areas of release of vitamin A from storage in the liver is apparently Africa, the Andes of South America, and in many parts of Asia hindered by infection (DeMaeyer 1986). The effect on the (see tables 19A-1, 19A-4, and 19A-6). immune system explains, in part, a well-documented associa- Metabolism of the micronutrients may also be affected by tion between vitamin A deficiency and growth retardation dietary components. Protein-energy malnutrition interferes in animals (McLaren 1966, cited in West and Sommer with iodine metabolism (Ingenbleek and De Visscher 1979; 1987; Eastman 1987). In experimental animals vitamin A Gaitan, Mayoral, and Gaitan 1983). Other nutrients may also causes a cessation of bone growth along with loss of appetite increase or inhibit the absorption or use of these micronutri- (Eastman 1987). ents. Vitamin A and zinc deficiencies might interact synergis- Although conclusive information on mortality risks among tically (Baly and others 1984). Vitamin A deficiency also children with all levels of vitamin A deficiency is not yet affects anemia (Bloem and others 1990). For example, in available (Wittpenn and Sommer 1986), it has been estimated Guatemala, fortification of sugar with vitamin A resulted in that of the children with keratomalacia who remain untreated, improvement of hemoglobin levels (Mejia and Arroyave 60 percent will die (West and Sommer 1987). Some studies 1982). Because vitamin A seems to affect hemoglobin levels show that even children with mild vitamin A deficiency have but not body stores, it might be involved in the synthesis of higher mortality rates than matched controls (Sommer and hemoglobin and red blood cells (Mejia andChew 1988). Other others 1983; Sommer and others 1986; Rahamathullah and components in foods may inhibit the use of dietary micronu- others 1990). The cause of this is probably related to vitamin trients, such as phytates in some plants, which inhibit iron A's role in maintaining healthy mucosal tissue throughout the absorption. body and in the immune function. To confirm these findings similar studies are under way in other countries, and results Prevention should provide conclusive evidence on whether vitamin A deficiency increases the mortality risk in children (National Adequate consumption of the micronutrients through food is Academy of Sciences 1987). the best way of preventing micronutrient deficiencies. Recom- There are no known effects of vitamin A deficiency on the mended daily intake by age and sex is presented in table 19A-3 growth and development of the brain and intelligence. Blind- (iron) and table 19A-5 (iodine and vitamin A). ness or partial blindness would obviously affect learning, espe- When the intake of these foods is limited, specific interven- cially in the classroom setting. To the extent that it increases tions are needed to prevent and address micronutrient defi- morbidity, even mild vitamin A deficiency may affect school ciencies. Most micronutrient interventions represent both performance and productivity. At issue here is how many preventive and curative therapies. High-dose supplements, in preschool- or school-age children with a history of xerophthal- particular, can be used to treat severe deficiency and to prevent mia are left out of formal schools because of blindness due to deficiency in vulnerable age groups. Micronutnient Deficiency Disorders 427 Table 19-4. Functional Effects of Essential Micronutrient Deficiencies Deficiency Effect Iron Iodine Vitamin A Morbidity Immune function Yes Unknown Yes Prevalence Yes Yes No Incidence No Unknown Yes Duration No Unknown Yes Severity No No Yes Mortality Infant Yes Yes Yes Child Unknown Yes Yes Maternal Yes Unknown No Other (fetal, early adult) No Yes No Mental development and learning disorders Brain development Unknown Yes No Aptitude Yes Yes No Intelligence quotient Yes No No Exploratory behavior Yes No No Attention span Yes Yes No Memory Unknown Yes No School achievement Yes Yes Unknown Learning disability (blindness, deafness) No Yes Yes Sensory impairment Yes Yes No Productivity Spontaneous activity Yes Yes No Endurance Yes No No Max imum aerobic capacity Yes No No Occupational productivity Yes Yes No Disability3 No Yes Yes Growth Yes Yes Yes Reproduction Fertility No Yes Yes Miscarriage and stillbirth Yes Yes Yes Intra-uterine growth retardation Yes Yes No Prematuritv Yes No No Congenital deformities (birth defects) No Yes Yes a. Such as blindness, mental retardation, or lack of motor coordination. Source: Iron-DeMaever 1989; lodine-Hetzel, Dunn, and Stanburg 1987; Vitamin A-West and Sommer 1987, Eastman 1987. There are two main types of interventions to reduce micro- three to five years, as is the case for iodized oil. The frequency nutrient deficits: supplementation (the administration of pills, depends mainly on the ability of the body to store the micro- capsules or injections containing one or more of the micronu- nutrient (substantial in the case of iodine and vitamin A). trients) and fortification (the addition of micronutrients to Although iron deficiency is usually treated with supple- foods in processing). Other interventions, such as nutrition ments from one of several iron compounds, ascorbic acid education and agricultural programs, can be used over the long increases the absorption of nonheme iron from existing sources term to promote the intake of these micronutrients by vulner- like maize, rice, wheat, or sorghum, in which the iron content able groups. is adequate but is in an unabsorbable form. Vitamin C supple- mentation and fortification can therefore be considered to be Supplementation an iron intervention. When taken with such foods, ascorbic acid can increase the absorption of available iron by about 30 Delivery of micronutrients through supplementation can be percent (Hallberg 1981; Berg and Brems 1986), or from about done in a variety of ways. The supplements can be taken orally 5 percent to about 6.5 percent. or by injection. Typical iron supplementation programs are A significant challenge for prophylaxis or treatment shown in table 19A-7. Some supplements must be taken daily through supplementation is compliance (taking the proper (such as oral iron), whereas others can be taken at intervals of dosage at appropriate intervals). This is a problem particularly 428 Henry M. Letin, Ernesto Pollitt, Rae Galloway, and Judith McGuire in iron supplementation, which requires the daily ingestion of Fortification iron tablets, sometimes with mild side effects (headache, nau- sea, and so on) in the initial weeks of supplementation. Side Where essential micronutrient deficiencies are prevalent, effects were thought to be an important detractor to compli- dietary fortification is generally considered preferable to sup- ance, but in recent iron studies in Thailand and Burma, only plementation as a long-term strategy in controlling micronu- 10 to 15 percent of women taking iron complained of side trient deficiencies. The advantages of dietary fortification are effects (Charoenlarp and others 1988), and only a small pro- that compliance is ensured if the appropriate carrier is selected. portion of those women failed to take supplements because of The cost of delivery of the micronutrients through food staples side effects. In another study, Griffiths (1980) found that when is far less than through the health system and can be partially women were warned of possible side effects they were more or fully borne by the consumer. A crucial step in fortification likely to continue taking their iron supplements when side is choosing the right food to fortify. Several criteria are used in effects occurred. Ensuring compliance with iron therapy has selecting a particular food vehicle for fortification (Beaton and been most successful in situations in which supplements are Bengoa 1976; Baker and DeMaever 1979): provided and ingestion is supervised, such as in the workplace It must be a food that is consumed by the vast majority and schools. But for persons who do not participate regularly * t tabe ation at in adequate a s. in such institutions or who live in outlying areas, compliance t tare pulato an in a amouns. and tablet availability are serious obstacles to success. A recent * It must be able to be fortified on a large scale and at study of iron ingested in slow-release capsules (gastric delivery relatively few centers, so the fortification can be adequately system [GDSI) showed that side effects did not differ between superv'sed. placebo, ferrous sulfate, and GDS. In addition, compliance did * It must be stable under the extreme conditions likely to not seem to differ between ferrous sulfate or GDS even though be encountered in storage and distribution. the dosage for GDS was one pill per day and that for ferrous * It must not interfere with the use of the nutrient, and the sulfate was two (Simmons 1990). The CDS iron was better nutrient must not interfere with the food (that is, it rnust absorbed, however. not be detrimental to flavor, shelf life, color, texture, or When available, long-lasting, megadose supplements are cooking properties). superior to low-dose supplementation with respect to both cost and compliance. This is particularly true for injections of Table 19-5 shows typical fortification programs for the three iodized oil, which provide protection for as long as four to five micronutrients under discussion. The usual food vehicles are years, and oral doses of iodized oil, which are good for two years salt or sugar because they tend to meet the four criteria. The (Underwood 1983; Berg and Brems 1986). Large doses of food chosen for fortification is site specific; thus successful iodine can cause adverse reactions (thyrotoxicosis, with such fortification programs have included wheat flour, skimmed symptoms as increased heart rate, trembling, sweating, and milk, monosodium glutamate (MsG), infant foods, beverages, weight loss [Hetzel 19881), although this is not generally a salt, and condiments (Clydesdale and Wiemer 1985). Foods significant problem because in most cases spontaneous remis- commonly eaten only by specific subpopulations may prove to sion will occur (Medeiros-Neto and others 1987). The highest be satisfactory vehicles for targeting high-risk groups, such as risk of thyrotoxicosis is in people over forty, so giving supple- small children. For example, in many industrial countries ments only to younger adults avoids most of this toxicity (Berg weaning foods are fortified with iron. and Brems 1986). High doses of vitamin A are both toxic and The fortification of salt with iodine has been practiced teratogenic. Great care must be taken not tt) give vitamin A extensively in industrial countries. Many Latin American capsules at higher dosages than recommended or to women countries have passed legislation requiring iodization of all who might be pregnant (generally 1,000 micrograms or less). salt. Unfortunately, maintaining supplies of iodized salt has Vitamin A can be toxic even in children, so an alternative remained a problem in these countries. Political commitment delivery scheme based on low, frequent doses is being investi- at all levels of government and the community is needed for gated (Underwood 1989). As with any essential drug, micro- such legislation to be effective. A successful national pro- nutrient supplements need to be handled properly to ensure gram in Bolivia organized small salt producers into coopera- their stability, potency, availability, and proper use. tives, which made them more able to compete with larger Table 19-5. Typical Fortification Program Nutrient Compound Vehicle Concentration Source Iron Ferric orthophosphate Salt 3.5 g/kg Working group on fortification of salt with iron 1982 NaFe EDTA Sugar 13 mg/100 g Viteri and others 1981 Iodine Potassium iodate Salt 15-40 ppm Mannar 1987 Vitamin A Retinol palmitate SLIgar 10 mg/g Arroyave and others 1979 Source See last column. Micronutrient Deficiency Disorders 429 producers. This action increased their productivity, which in Breastfeeding promotion is important for micronutrient turn increased the availability of iodized salt, improving con- nutriture (as well as other nutrition and health benefits) be- sumption of iodized salt and reducing the prevalence of goiter cause, if the mother is in good health, it is a good source of iron, (Pardo 1990). iodine, and vitamin A. In addition, a protein in breast milk, Fortification may not always be feasible, however, because lactoferrin, reduces free iron in the intestinal lumen and hence of the lack of an appropriate carrier food available to at-risk protects the infant against infection while simultaneously ren- groups, weak enforcement of fortification regulations, (Berg dering the iron more absorbable (Tomkins and Watson 1989). and Brems 1986), or excessive cost of the fortified food. The Breastfeeding promotion is also important because vitamin A most appropriate carrier may also be a food that has unrelated deficiency has been associated with the early cessation of health effects, such as hypertension and tooth decay, which breastfeeding (Stanton and others 1986). In Bangladesh the are linked with the overingestion of salt and the consumption risk of a child's developing one or more signs of vitamin A of sugar, respectively. In these cases other strategies may be deficiency is six times higherforachild younger than two years required. Fortification is the best method available for solving of age who is not breastfed (Mamdami and Ross 1988), and micronutrient deficiency problems in the long term because of length of lactation was found to influence the risk of vitamin low cost, good coverage, and technical feasibility. The success A deficiency in children in Ethiopia (de Sole, Belay, and of fortification in the long term hinges crucially on the regu- Zegeye 1987, as cited in Mamdami and Ross 1988) and Indo- larity and enforcement capacity in the governmental depart- nesia (Sommer 1982). ments responsible for food safety and quality. Without Promoting the increased production and use of nutrient-rich effective public and private oversight of the fortification sys- local foods through agricultural programs and policies (im- tem, the quality of a fortification program can deteriorate proved marketing, greater dietary diversity, increased rural rapidly. Sufficient incentives and penalties for private industry incomes) may prove invaluable to any program aimed at re- and public regulators need to be set up to ensure longevity. ducing the incidence of these deficiencies. Not surprisingly, in Double fortification of foods may be a way to address two urban Bangladesh, risk of vitamin A deficiency in children was micronutrient deficiencies in a cost-effective way. India has associated with poor intake of foods rich in vitamin A (Stanton experimented with double fortification, using iron and iodine. and others 1986). In Bangladesh, families without gardens Although the technology has not been completely worked out, are more likely to have children with xerophthalmia than are double fortification offers hope in dealing with multiple micro- those with gardens (Cohen and others 1985). nutrient deficiencies in areas where several coexist. Research and development of new technologies to increase the micronutrient content of raw and processed foods and new Other Interventions high-nutrient varieties are also needed. For example, in pro- cessed foods the absorption of iron could be greatly increased While supplementation and fortification are two of the main by the addition of ascorbic acid or by the decrease of sub- interventionsusedtoaddressmicronutrientdeficiencies,there stances that compete with iron for absorption. Public health are a number of other interventions-nutrition education, measures to alleviate environmental factors which exacerbate breastfeeding promotions, agriculture, food processing, public dietary deficiencies (Stanton and others 1986) may be recom- health measures-that should be used either alone or along mended also. with supplementation and fortification efforts. Program Coverage NUTRITION EDUCATION. Education of the consumer about nutrition is an important component of any micronutrient In tables 19A-1, 19A-4, and 19A-6 we give approximations intervention. It is needed along with every fortification, sup- for program coverage for iron, iodine, and vitamin A, respec- plementation, and food production program to guarantee that tively. As can be seen in table 19A-4, there is much activity the intended beneficiary actually consumes the nutrients. Fail- in planning and implementing national programs to combat ure to educate the public and politicians to support fortifica- iodine deficiency disorders. Control programs usually involve tion programs over the long term has been implicated as a either fortification or supplementation. Legislation to make significant reason for the failure of programs in Central and iodization of food mandatory is more infrequent except in some South America (Schaefer 1974, as cited in Thilly and Hetzel Latin American countries in which legislation was passed 1980). In both Guatemala and India, mass media and commu- several decades ago. It should be noted, however, that even nication campaigns were required to create demand for forti- with the passage of legislation, goiter has persisted in these fied salt (Thilly and Hetzel 1980; United Nations 1987). countries, which indicates the difficulty in controlling and More general nutrition education is also important to in- regulating industry to comply with the law. Knowledge of crease intake of the micronutrients from the existing food coverage for these iodization programs is scant, and even in a supplies. Lack of maternal knowledge of the need for children country such as Bhutan, where 100 percent of the salt is to consume leafy green and yellow vegetables is associated iodized, there are questions of whether or not this salt is with increased risk of nutritional blindness (Stanton and reaching remote endemic areas and of whether the stability of others 1986). iodine in the salt can be maintained at effective levels. 430 Henry M. Levin, Ernesto Pollitt, Rae Galloway, and Judith McGuire Vitamin A program activity has increased dramatically over graphic distribution, and causality need to be assessed in each the last several years, but gaps still exist in the knowledge of country. Priorities for resource allocation must be based on the actual coverage in countries. Only a few countries have under- nature of the problem (prevalence, severity, geographic distri- taken vitamin A fortification programs, making supplementa- bution, causality), the cost-effectiveness of altemative solu- tion the most frequent option at present. Pilot studies in the tions, the institutional capacity to carry out the interventions, Philippines for fortifying MSG with vitamin A have proved and the cultural acceptability of solutions. promising (Muhilal and others 1988; Muhilal, Muherdiyanti- ningsih, and Karyadi 1988). As for program coverage for io- Cost-Effectiveness Analysis dine, program coverage for vitamin A is still underreported. Until such information is available, it will be difficult to gauge Different delivery systems are associated with different costs progress in combating the deficiency. and effectiveness. In this context, the term "costs" refers to the Control programs for iron deficiency are not well docu- value of all resources required to deliver the micronutrients to mented, even though iron supplementation presumably is part the target population. The term "effectiveness" refers to pro- of standard practice in prenatal care. Considering the magni- gram and biological effectiveness. Program effectiveness is the tude of the deficiency problem and the effects of iron defi- efficacy of the delivery system in providing adequate dose and ciency, the lack of attention given to addressing the problem coverage to those with deficits. Biological effectiveness is the is surprising. Countries with active programs, such as India, efficacy of the dose to eliminate the deficiency. Cost-effective- have found difficulties in population compliance with a con- ness is the cost per unit of change in the outcome of interest sequence of low coverage. More than 80 percent of those (West and Sommer 1984). The choice of a delivery system dropping out of iron supplementation programs cited discon- should depend heavily upon its relative cost-effectiveness, the tinued supplies of tablets as the reason for noncompliance most desirable strategies being those with the highest effective- (United Nations 1990). Much more needs to be done to ness relative to cost. An additional economic criterion is that document exact prevalence and geographic location of iron of the cost-benefit relation for each intervention. Micronutri- deficiency so that control programs can be effectively targeted ent interventions have both costs-the value of resources to those most at risk. Better program design is also needed to required for delivery of micronutrients to the appropriate pop- meet present problems. ulations-and benefits-the improved functioning of those populations through the elimination of micronutrient defi- Case Management ciencies. For example, decreases in vitamin A deficiency will reduce blindness, allowing affected populations to care more Prevention is the best type of case management, but where fully for themselves, to reduce needless expenditures on health severe deficiencies of iron, iodine, and vitamin A exist, case care, to benefit from education, and to be more productive in management is best handled by trained health personnel and the workplace. Reduction of iron deficiencies improves may require hospitalization. Immediate attention should be educational outcomes and work output in both the household giventocorrectingdeficienciesupondiagnosis.Theexception and the workplace. Reduction of iodine deficiencies de- to this is cretinism, which is irreversible. Immediate attention creases the likelihood of cretinism and other disorders that should be given to the mothers of cretinous children to im- burden society. prove the outcomes of future pregnancies. With mild and One criterion for determining where to make investments moderate deficiency, case management can be supervised by in a resource-scarce situation is to allocate resources to those community health workers. In table 19A-2 we show how endeavors in which the ratio of benefits of the intervention to severe deficiencies in iron, iodine, and vitamin A can be costs exceeds altematives. Costs can provide a measure only of detected, and in table 19A-7 we give typical supplementation the cost of delivery and not of the program and biological programs for all these micronutrients. effectiveness. Any measure of cost-effectiveness must take into For anemic patients with circulatory failure or respiratory account both the cost and the effectiveness of the interven- distress, blood transfusion is required. Because blood loss of a tion. The methodologies involved in determining costs and severelydeficientpersoncanprecipitateshockorhearttfailure, benefits are included in appendix 19B. In appendix 19C we it is of vital importance to increase the hemoglobin level of describe the criteria of effectiveness. In appendix 19D general anemic pregnant women before and during labor. If a child methodologies for cost-benefit analysis are presented. experiences severe vomiting or diarrhea after taking oral vita- Any comparative analysis of cost-effectiveness should take min A capsules, water miscible retinol palmitate injections can account of costs properly accounted for to meet micronutrient be used. Oil-based injections should not be used because they needs among at-risk persons for a given period of time. In table are metabolized too slowly to be effective in acute, severe 19-6 the costs per person are given for different interventionis deficiency (West and Sommer 1987). in 1987 dollars (column 3). Column 4 presents the data from column 3 corrected for the duration of the dose. We stress in Assessment of the Effect of Interventions appendix B the need to use an ingredients or resource recovery method to estimate costs. Moreover, we indicate some of the In order to determine the best solution for addressing micro- reasons that the costs of a given intervention in a particular nutrient deficiencies, the prevalence, socioeconomic and geo- context cannot necessarily be generalized to other contexts. Micronutrient Deficiency Disorders 431 With respect to micronutrient requirements, there is no capsules must be taken per year, the cost per year of protection assurance that different interventions have the same success is twice that for the administration of a single capsule-the for the reasons stated in appendix B. The prevalence and cost basis in each of the studies. The difference in the costs severity of micronutrient deficiencies differfrom site tosite and between the two iron fortification programs using sugar as the affect the success (and costs) of an intervention. Different vehicle is due to the addition of ascorbic acid to the second interventions cover requirements for different lengths of time. program. Ascorbic acid costs fifteen to twenty times as much A year of fortification, for example, meets one year's require- as ferrous sulphate. The high estimated cost for delivery of ments; oral iodine covers needs for two years; injected iodine ferrous sulphate tablets is because of the relatively high per- covers needs for three to five years, depending upon the dosage; sonnel requirement for providing daily tablets and the active and vitamin A capsules are associated with four to six months supervision and motivation needed to obtain compliance. If of protection, although lower-dose vitamin A supplements are such tablets could simply be delivered every six months to available for monthly dosing. Clearly, the cost per year must households, the cost would fall much closer to the cost of be adjusted for the duration of protection provided by the distributing vitamin A capsules. intervention. Finally, the cost per year must be adjusted for the Over time the cost-effectiveness may change. Tilden and benefit leakage. In a large-scale fortification effort, not all of Grosse (1988) found that dietary modification programs were the recipients will be at risk, so the cost per person at risk will more cost-effective over twenty years than either supplemen- be higher than the cost per recipient. Targeting more finely tation or fortification programs and that supplementation was may not be cost-effective even if leakages are high. If only more cost-effective over time than fortification. In their anal- one-third of the population is at risk of micronutrient defi- ysis they showed that over a twenty-year time period dietary ciency, the cost per each at-risk person will be three times as modification is more effective than supplementation, which is high as the average cost per person. more effective than fortification in preventing blindness and One of these adjustments is made in the last numerical death from vitamin A deficiency. It should be kept in mind column in table 19-6, where the estimated cost per person in that fortified foods and water are consumed by large numbers the previous column is calculated on the basis of one year of of people who are not at risk. Thus, the cost per at-risk person protection. The enormous differences in estimated cost be- for supplementation and fortification is much closer than the tween fortification with iodine and iodized oil injections is differences shown in table 19-6, and the cost-effectiveness narrowed considerably when the five-year period of protection of the two strategies may not be very different once these for oil is taken into account. Even so, the differences in cost adjustments are taken into account. This is especially true for oil injections are substantial among studies, which may be for vitamin A deficiency; the numbers of those at risk are the result of real differences, of some of rhe site-specific differ- small (children under five) in relation to the entire popula- ences, or of poor data. With respect to vitamin A capsules, the tion, and thus targeted supplementation would be more cost- two studies show considerable agreement on costs. Because two effective. Table 19-6. Cost of Micronutrient Interventions Estirnated cost Estimated cost per person per person pet year Cost per person (1987 U.S. of protecton Nutrientform Country and wear (U.S. dollars) dollars) (U.S. dollars) Source Iodine Oil injecrion Peru 1978 1.30 2.30 0.46 Hetzel and others 1980 Oil injection Zaire 1977 0.35 0.67 0.14 Herzel and others 1980 Oil injection Indonesia 1986a 1.00 1.05 0.21 Irie and others 1986 Water fortification Italy 1986 0.04 0.04 0.04 Squarrito and others 1986 Salt India 1987 0.02-0.04 0.02-0.04 0.04 Mannar 1987 Vitamin A Sugar foTtification Guatemala 1976 0.07 0.14 0.14 Arroyave and others 1979 Capsule Haiti 1978 0.13-0.19 0.23-0.34 0.46-0.68 Austin and others 1981 Capsule Indonesta and Philippines 0.10 0.21 0.42 West and Sommer 1984 1975 Iron Salt fortification India 1980 0.07 0.10 0.10 Cook and Reusser 1983 Sugar forrification Guatemala 1980 0.07 0.10 0.10 Viteri and !9thers 1981 Sugar fortification Indonesia 1980 0.60 0.84 0.84 Levin 1985 Tablets Kenya and Mexico 1980 1.89-3.17 2.65-4.44 2.65-4.44 Levin 1985 a. Per injection. b. From data provided in Derman and others 1977. Source: See last column. 432 HenrY M. Levin, Ernesto Pollitt, Rae Gallowa.N and Judith McGuire An additional way of measuring cost-effectiveness across a The pecuniary value of additional work output was esti- variety of health and nutrition interventions is to use a com- mated by first ascertaining the probable rise in Hb associated mon outcome measure and estimate the costs per unit of that with particular interventions. This rise in Hb was converted outcome. Throughout this collection, the disability-adjusted into an increase in individual work output by applying the life-year gained is one measure of universal applicability. It is elasticities from the studies mentioned above. Because at least also possible to use cost per life saved or cost per unit of some of the rise in individual work output will simply replace economic productivity gained. the work output of others in a labor surplus economy, the social The cost-effectiveness of achieving these outcomes with benefits will be less than the sum of the individual increases in micronutrient interventions is shown in the tables in appendix productivity. That is, some workers will no longer be needed 19E and summarized in table 19-7. Using the current preva- and will be unemployed as the output of other workers in- lence of deficiencies commonly observed in developing coun- creases. Therefore, only half of the increase in work output was tries and certain assumptions about demographics, death and assumed to be a net increase in social productivity. disability, coverage and effectiveness (75 percent), the dis- The net increase in social productivity was valued according count rate (3 percent), and life expectancy (seventy years), we to the wages that would be required to produce that additional calculated the discounted cost per disability-adjusted life-year output. The total value of productivity was adjusted to a per gained based on available costs of micronutrient control pro- capita level by dividing by the entire population, including the grams.' For calculations of productivity, the annual wage rate portion of the population that was not economically active. was assumed to be $500, unemployment was assumed to be 25 Finally, the results were adjusted for the estimated effects of percent, and disabilities were assumed to be the same as the improved iron status on outputs other than work productivity. health disabilities in table 19E- 1. No adjustment was made for These include lower morbidity and mortality, greater physical increased cost of feeding a more productive worker or for the stature, higher productivity outside of the workplace, im- employment replacement effect of increased productivity. proved quality of leisure time, greater leaming and faster school advancement, and increased feelings of well-being. Cost-Benefit Analysis Especially important is the additional work output in the household and in peasant agriculture, which is not accounted Cost-benefit analysis for all micronutrients, although per- for by the value of additional output in labor markets. This formed on a limited number of countries, suggests that both adjustment raised the value of total benefits by 50 percent supplementation and fortification are good investments. above those of just the market-based work benefits. Costs were estimated for fortification strategies in which IRON. Levin (1985) estimated the benefits and costs of both both salt and sugar were used as examples of dietary vehicles. medicinal supplementation and dietary fortification with iron- Supplementation was based on the assumption that iron sup- deficient populations on the basis of data from Indonesia, plements were one of four different dietary or health interven- Kenya, and Mexico. Benefits accrued primarily from higher tions delivered by a health system. Both the average cost per work output associated with normalization of hemoglobin intervention was estimated as well as the marginal cost of the levels in anemic populations. A remarkable degree of consis- iron supplements alone. The cost of fortification was based on tency was found in eight studies of work output related to data from field trials. The cost of supplementation was based hemoglobin levels. The elasticity of work output with respect on the use of village health auxiliaries, various modes of to rises in Hb was between one and two: that is, an increase in transportation, facilities, and the cost of the supplements. An Hb of 10 percent was associated with a rise in work output of additional cost in both types of interventions resulted from the 10 to 20 percent. higher energy needs of workers who produce a larger work Table 19-7. Return on Nutrition Investments Discounted value Cost per disability- Cost per life saved of productivity gained adjusted life-year gained Intervention (U.S. dollars) per program dollar (U.S. dollars) Iron deficiency Supplementation of pregnant women only 800 24.70 12.80 Fortification 2,000 84.10 4.40 Iodine deficiency Supplementation of reproductive-age women only 1,250 13.80 18.90 Supplementation of all people under sixty years 4,650 6.00 37.00 Fortification 1,000 28.00 7.50 Vitamin A deficiency Supplementation of children under five years only 50 146.00 1.40 Fortification 154 47.50 4.20 Source: Based on table 1 9E- 1. Micronutrient Deficiency Disorders 433 output. This cost was estimated on the basis of the additional understanding of benefits and costs associated with IDD control calorie input required for the additional work output, calcu- in those areas of the world with substantial at-risk populations, lated from the cost of the additional rice o 50 Moderate Population 21-29 n.a. Individual n.a. 25-50 Severe Population > 29 n.a. Individual n.a. < 25 Vitamin A deficiency Population: children younger than six years (percent) Night blindness (xN)' 1.0 Bitot's spots (xlB)c 0.5 Corneal scars, ulceration (x3.A) 0.01 Xerophthalmia-related corneal scars 0.05 Serum retinol < 10 mcg/dL 5.0 Serum retinol Liver retinol Individual (mcg/dL) (mcg/g) Mild deficiency Children 20 < 20 Adults 30 < 20 Severe deficiency Children <10 < 10 Adults <10 < 10 n.a. Not Applicable. a. Hemoglobin values below wshich anemia is likely to be present in individuals living at sea level. b. Prevalence of endemic cretinism is 1-10 percent. c. Clinical abbreviation for stage of Xerothphalmia. Source: West and Sommer 1987; DeMaever and Adiels-Tegman 1985; Hetzel 1988; DeMaeyer 1989. Micronutrient Deficiency Disorders 437 Table 19A-3. Recommended Daily Intake of Iron, Based on Bioavailability in Diet Iron intake (mg/day) by quality of the dieta Age group Absorbed iron requiremenit (mg/day) Lou, bioavailabilit' Medium bioavailabilitY High bioavailabilitr 4-12 months 0.96 32 16 9 13-24 months 0.61 20 10 5 2-5 years 0.70 23 12 6 6-11 years 1.17 39 19 11 Girls 12-16 years 2.02 67 34 18 Boys 12-16 years 1.82 61 30 16 Adult Males 1.14 38 19 10 Adult Females Pregnant 3.6 120 60 33 Lactating 1.31 44 22 12 Menstruating 2.38 79 40 22 Postmenopausal 0.96 32 16 9 a. As defined by Monsen and others 1978, a diet with low bioavailability contains no meat, fish, or poultrv; none of the Iron Is heme Iron and 3 percent of total iron is absorbed. A diet with medium bioavailability contains I ounce of fish per day; four percent of the iron is heme iron, and 6 percent of total iron is absorbed. A diet with high bioavailability contains 3 ounces of beef per day; 21 percent of the iron is heme iron, and 11 percent of total iron is absorbed. Source: DeMaeyer 1989. Table 19A-4. Prevalence of Iodine Deficiency and Program Coverage, by Country Prevalence Fortification Supplementation Country Percent Areaa Program status Legislation (years) Coverage (percent) program status Af-ca Angola - Regional None None - None Benin - Regional None None - None Botswana 63 Regional None None - Planned Burkina Faso 7.7 National, None None - Planned Burundi 56 Regional Under way None - Planned Cameroon 59 Regional None None - Planned Central African Republic 25 Regional None None - None Chad 11 Regional None None - None Comoros 40 Regional None None - None Congo - Regional Under way 1988 - None C6te d'lvoire 18 Regional None None - Planned Ethiopia 34 Regional Under way None - Under way Gabon I National None None - None The Gambia Regional None None - None Ghana 13 Regional None None - Planned Guinea 15.4 National None None - None Guinea Bissau - Regional None None - None Kenya 15-72 National Under way 1970 50 Planned Lesotho 14.3 National Planned None - Planned Liberia - Regional None None - Planned Madagascar 18 Regional None None - Planned Malawi 30-70 Regional Planned None - Under way Mali 20 Regional None None - Planned Namibia - Regional None None - None Niger 13 National None None - Planned Nigeria 40 Regional None None - Planned Rwanda 19 Regional Under way None - Planned Senegal 33 Regional Under way None - None Sierra Leone - Regional None None - Planned Somalia - Regional None None - None Sudan 20 Regional None None - Planned Swaziland 26 National None None - None Tanzania 40 Regional Under way None - Under way Togo - Regional None None - Planned Uganda Regional None None - None Zaire - Regional None None - Under way Zambia 27-81 Regional Under way 1979 - Planned Zimbabwe 20 Regional None None - Planned (Table continues on the following page.) 438 Henrn M. Levin, Ernesto Pollitt, Rae Galloway, and Judith McGuire Table 19A-4 (continued) Prevalence Fortification Supplementation Count-v PercenT Areaa Program status Legislation (years) Coverage (percent) program status Asia Bangladesh 10.5 National Under way 1989 55 Under way Bhutan 64.5 National Under way None 100 Under way Burma 14.3 National None None - Under way Cambodia 30 National None None - Planned China - Regional Under way None 87 None India 7.3 National Underway 1962 12 Planned Indonesia 20 National Under way 1976 51 Under way Korea, Republic of - None None - Planned Lao PDR - None None - Planned Malaysia - Regional Under way None 0 None Nepal 46.1 National Under way None 72 Under way Papua New Guinea 40 Regional Under way 1972 - Planned Philippines 14.9 National Under way None - Planned Sri Lanka 19.3 National Planned None 0 Planned Thailand 14.7 National Under way None 2 None Viet Nam 34 National Under way None 5 None Middle East Afghanistan National None None - None Algeria - Regional Under way None Planned Egypt 70 Regional None None - None Iran 60 Regional None None - Planned Iraq 80 Regional None None - Planned Lebanon 50 Regional None None - Planned Libya 20 Regional None None - Planned Morocco - Regional None None - None Pakistan Regional Under way None 11-17 None Tunisia - Regional None None - None Latin Amenca Argentina 15.6 National Underway 1967 99 Underway Bolivia 61 National Under way 1967 20-80 Under way Brazil 14.7 Regional Under way 1977 - None Chile 18.8 Regional Under way 1968 85 Under way Colombia 1.8 Regional Under way 1947 - None Costa Rica 3.5 National Under way 1970 - None Cuba 30.3 Regional None None - Under way Dominican Republic 80 Regional None None - None Ecuador 36.5 Regional Under way 1968 75-80 Under way ElSalvador 48 National Underway 1961 17 Underway Guatemala 10.6 National Under way 1954 36 Under way Honduras 17 National Under way 1961 - Under way Mexico - Regional Under way 1962 Under way Nicaragua 20 National Under way 1969 - Under way Panama 6 National Under way 1966 47 Under way Paraguay 18.1 National Under way 1966 - Under way Peru 50-80 Regional Planned None 60 Under way Uruguay - Regional Under way 1961 - Under way Venezuela 21.3 National Under way 1968 - Under way - Data not available. Note The quality of data (sample size, ages, region) necessitates caution when comparing countries. a. Regional or sporadic goiter prevalence, or known regionally in some age groups; or national goiter prevalence or prevalent nationally in some age groups. b. Probable. Source. FAO no date; Hetzel 1987; United Nations 1987; PAHONIWHOIUNICEF 1988; Dunn (ed.) 1989a and 1989b; ICCIDD 1989; ICCIDDIWHO 1989; "Iodine De- ficiency" 1989; Pollitt 1989; "Status of IDD" 1989; WHO 1989. Micronutrient Deficiency Disorders 439 Table 19A-5. Recommended Daily Intake of Iodine and Vitamin A (mcg/day) Age group Intake Iodine All People 150-300 Vitamin A Infants younger than I year 300 1-3 years 250 4-6 years 300 7-9 years 400 10-12 years 575 13-15 years 725 16-19 years 750 Adult men and women 750 Lactating women (first six months) 1,200 Note: Micrograms or retinol equivalents (RE) are currently used to measure retinol and the carotenoids. Previously, intemational units were used. To con- vert: I RE = I mcg retinol = 6 mcg all-trans beta-carotene = 12 mcg other provitamin A carotenoids = 10 u provitamin A carotene = 3.33 iu of retinol. a. Including pregnant women and lactating women after the first six months. Source: FAO 1965; Hetzel 1988. Table 19A-6. Prevalence of Vitamin A Deficiency and Program Coverage by Country Prevalence Fortificaton Supplementation Countr-v Percent Areaa Program status Legislation program status Africa Angola - Probable None None None Benin 3.5 Regional None None None Botswana - Regional None None None Burkina Faso - National None None None Burundi - Probable None None Under way Chad - National None None None C6te d'lvoire - Regional None None Under way Ethiopia - Regional None None None Ghana - National None None Under way Kenya - Probable None None Under way Madagascar - Regional None None None Malawi 3.9 Regional None None Under way Mali - National None None Under way Mauritania 2.3 Regional None None Under way Mozambique - National None None Under way Niger - National None None Under way Nigeria - Regional None None Under way Rwanda - Probable None None None Senegal - Regional None None None Somalia 11.0 Regional None None Under way Sudan 1.6 Regional None None Under way Tanzania 1.6 Regional None None Under way Uganda - Regional None None Under way Zaire - Regional None None Under way Zambia - Regional None None Under way Zimbabwe - Regional None None None Asia Bangladesh 4.9 National None None Under way Burma - Regional None None Under way Cambodia Probable None None None China - National None None None (Table contnnues on the following page.) 440 Henm- M. Levin, Ernesto Pollitt, Rae Galloway, and Judith McGuire Table 19A-6 (continued) Prevalence Fortification Supplementation Country Percent Area' Program status Legislation program status Asia (continued) India 12-20 National None None None Indonesia 20.0 National Under way None Under way Lao PDR - Probable None None None Malaysia - Regional None None None Micronesia 10.0 n.a. None None None Nepal 1.0 National None None Under way Philippines - National Planned None Under way Sri Lanka - Regional None None Under way Thailand - Regional None None None Viet Nam - National None None Under way Middle East Afghanistan - Probable None None None Algeria - Regional None None None Egypt - Regional None None None Iran - Regional None None None Iraq - Regional None None None Jordan Regional Under way 1977 None Morocco - Regional None None None Pakistan - Probable None None Under way Syrian Arab Republic - Regional None None None Yemen, Republic of - Regional None None None Yemen, Arab 0.57 Regional None None None Latin America Bolivia - Regional None None None Brazil - Regional None None Under way Costa Rica - Regional Under way Under way None Ecuador - Regional None None None El Salvador - Regional None None Under way Guatemala - Regional Under way Under way Under way Haiti 0.81 Regional None None Under way Honduras - Probable Under way Under way None Jamaica - Regional None None None Mexico - Regional None None None Nicaragua - Regional Under way None None Panama - Regional Under way Under way None Peru - Regional None None None - Data not available. Note. The quality of data (sample size, ages, region) necessitates caution when comparing countries. a. Either regional or sporadic prevalence or known regionally, or prevalent nationallv, with clinical signs in children younger than six years. Source: Arroyave 1982; Cohen and others 1985; FAO no date; IVACG 1989; Mathur and Kushwaha 1987; Pollitt 1989; Solon and others 1983; Underwood 1983; UNICEF 1988; United Nations 1985; WHO 1988a and 1988b. Table 19A-7. Typical Supplementation Programs Deficiency Target group Compound Dose Frequency Iron Presumptive treatment Pregnant women Ferrous sulfate 200 mg Twice daily where prevalence of Pregnant women Folic acid 500 mcg Daily iron deficiency anemia Children six months to five years Ferrous sulfate 10 mg/kg Dailyb is moderate or high School-age children Ferrous sulfate 200 mg Daily Presumptive treatment Pregnant women Ferrous sulfate 200 mg Daily where prevalence of Pregnant women Folic acid 250 mcg Daily iron deficiency anemia Children six months to five years Ferrous sulfate 3 mg/kg Dailyb is mild School-age children Ferrous sulfate 100 mg Daily Treatment of severe iron Pregnant women Ferrous sulfate 200 mg Thrice daily for deficiency anemia four weeks (hemoglobin less than Pregnant women Folic acid 250 mcg Thrice daily for 7 g/dL) four weeks Micronutrient Deficiencv Disorders 441 Deficiency Target group Compound Dose Frequency Children six months to five years Ferrous sulfate 10 mg/kg Daily Adults Ferrous sulfate 200 mg Twice daily Treatment of moderate Pregnant women Ferrous sulfate 200 mg Twice daily iron deficiency anemia Pregnant women Folic acid 250 mcg Daily (hemoglobin between Children six months to five years Ferrous sulfate 3 mg/kg Daily 7 and 10 g/dL) Adolescents Ferrous sulfate 200 mg Twice daily Adults Ferrous sulfate 200 mg Twice daily Iodine All lodinated oil-oral 2 ml Two years All lodinated oil' 2 ml Three to five years All lodinated oil- I ml Three years intratranscular Vitamin A Children without xerophthalmia Oil solution 20,000 mcg Three to six months after age d ~~~~~~~~~~~~~~one year Children with xerophthalmia Oil solution 20,000 mcg Diagnosis 20,000 mcg Next day 20,000 mcg Two to four weeks later, at clinical deterioration, or at discharge Pregnant women without dietary Oil solution 1,000 mcg Daily sources Lactating women Oil solution 20,000 mcg At parturition 1,000 mcg Daily Note: All treatment is oral, unless othenvise specified. a. Children younger than four months should be given only breast milk, which provides adequate iron. After four months, iron-containing weaning foods should be given. Low-birthweight infants require iron from two months age. b. Short-course therapy. c. Injected. A caveat for use of any injections is increased risk of AIDS transmission where needles are commonly reused and probability of sterilization is low. d. Children younger than twelve months should receive half doses. Source: IVACO 1986; Dunn 1987; West and Sommer 1987; DeMaeyer 1989; United Nations 1990. Appendix 19B. Costs of Supplementation pays or should pay and how it should be financed is important, and Fortification it is the subject of a separate analysis. It is important to get an accurate determination of the cost before addressing its Before we review the costs of supplementation and fortifica- financing. tion, it is important to review briefly the appropriate method for measuring costs. Although the notion of costs is often used Cost Estimation quite casually in the health sector literature, it has a very specific meaning in the economics literature (Mishan 1976; Using the ingredients method of determining the cost of the Levin 1983; Mills 1985). It refers to the social value of all the delivery of medicinal supplements, such as iron, one must first resources, or ingredients, that are required to provide an inter- select a delivery model. The usual model is that of a commu- vention-even resources provided in kind. Theproper method nity- or village-based health care system in which there is a for ascertaining costs is first to specify the particular resources heavy use of local resources, such as health auxiliaries or that are needed for a nutritional intervention, such as the community health workers. There is a considerable literature facilities, personnel, materials, and micronutrients that are on village or community health workers (Djukanovic and required. Second, the value or cost of each of the ingredients Mach 1975; Hetzel 1978; WHO 1979; Bender and Yoder 1983). is derived from both market data, if these are available, and Such workers are people who have completed all or most of other determinants of economic value, such as shadow price primary school and are literate in basic reading, writing, and (Mills 1985). These costs for all the ingredients are summed to computational skills. They typically come from the local com- determine the total cost of an intervention. The total cost can munity, so they relate well to the populations they serve. They be divided by the overall population or another base to obtain can deliver nutritional supplements and provide information the cost per participant of the intervention. and advice on their use. They are able to offer inoculations and The determination of the cost of an intervention is inde- other specific health services for which they are qualified pendent of the issue of how it is financed. The cost is a measure through short training programs, and they periodically have of the value of resources that are used for the intervention, no contact with more highly trained staff. Other ingredients for matter who is paying for them. Although an analysis of who the supplementation model include facilities, equipment, 442 H1enri M. Levin, Ernesto Pollitt, Rae Galloway, and Judith McGuire transportation, and the micronutrient supplements them- intervention, dominance of particular commodity flows, selves (Hetzel and others 1980; West and Sommer 1984; Levin and other factors that do not reflect the true value of 1985, 1986). resources in that standard monetary unit. The result is that The costs of the fortification intervention are based on those some of the cost differences among interventions drawn of the micronutrient compounds as well as personnel, equip- from the experience of different nations may depend on ment, and special packaging required for the preparation and fluctuating exchange rates that are not in competitive mar- delivery of the fortified product (Arrojave and others 1979). ket equilibrium rather than on the "true" world value of The value of the food vehicle that is being fortified is not a cost those resources. of fortification but only the additional cost associated with the * Time. The usual cost comparisons among studies do not fortification process. To protect the micronutrient content take account of the fact that often the studies were carried associated with fortification, some products, such as fortified out at different times. Because prices change over time and salt, must be packaged in more expensive bags than their follow different patterns among countries and because there unfortified counterparts. are often no appropriate price deflators for standardizing them over time, the costs derived at different time periods Comparing Costs are not strictly comparable even after adjusting for price- level changes. The literature on micronutrient interventions contains cost estimates for both supplemetitatioti and fortification. Usually, o* Context. A cost comparison between two different coun- estiatesfor oth upplmenttio andfortficaion.Usualy, tries and situations will reflect the unique characteristics of these costs are expressed as the cost per person covered by the thos cntexts. rex le,t tsf le halthteams inteventon.Thus in theory , it wol apertan ol those contexts. For example, it costs far less for health teams intervention. Thnus, In theory, It would appear that one COUldI readily compare the costs of injected supplements for iodine or to go from village to village in a relatively flat region with readily compare the costs of injected supplements for iodine or good transportation than it does for them to travel in moun- vitamin A with the costs of orally administered supplements. . Or one could compare the costs of fortification of different food tainots or jungle regions. Differences in population density vehicles with iodine, such as salt or water, or with iron, such affect costs profoundly because of the difficulties of reaching as salt, sugar, wheat flour, or milk products. Then one could sparse and remote populations as compared with more com- select those interventions that had the lowest cost per person pact ones. Tropical climates may pose additional costs for for delivery. packaging and refrigeration of micronutrients than more For example, one of the most obvious features of table 19-6 temperate climates. The result is that some of the measured is the large differences in the cost per person, even when cost differences among interventions may derive from the isth larg difrne in th cos pe pesn eve when differences in circumstances rather than intrinsic differ- standardized to 1987 dollars. Even for a single intervention, such as injections of iodinated oil, the cost per person varied ences in costs. from $0.67 to $2.30. The costs per person among different * Local price differences. Some differences in costs among interventions for the same micronutrient differ even more. interventions in different countries and regions are due to differences in prices for the same commodities and labor. For Idiosyncratic Differences in Costs example, the price of iodinated oil was found to be about 50 percent higher in Bolivia in 1985 than in France in 1983, a difference that cannot be explained by inflation (Dunn In fact, comparisons of costs among studies cannot be used as 1987). Among three countries in 1979, the cost of persons a basis for determining the most efficient form of supplemen- trained as vaccinators varied from $2.24 a day in Indonesia tation or fortification. The reason is that the costs in any tor$5.90 a day in Tnc(Cese particula study wil-to sore degree-e idiosynratic be to $5.90 a day inThailand, a considerable difference (Creese partcula stuy wil-tosom degee-b idisyncatiche- and others 1982). This Is another reason that cost experi- cause of the methodology used and the time and setting in ences in one country for a particular intervention might not which it was carried out (Mills 1985). It is important to reflectthe costsofthat intervention inanother country. summarize the sources of these idiosyncrasies to show why cost results from one study cannot necessarily be compared with * Population base. Some of the interventions are targeted another. to only the populations at high risk, such as is the case of iodinated oil in villages with a high incidence of goiter and * Methodological differences. Different studies use different low iodine in the soil. Others, such as the iodination of salt, methodologies, from casual guesses at costs to rigorous cost are based on distributing a fortified product to the entire accounting methods. Even among the latter there are differ- population, those in need of the intervention and those not ences in assumptions and methods based on different judg- in need. If only one-third of the population is in need of a ments. Unless a uniform method is used among studies, particular intervention, the cost per person at risk is three comparison is inappropriate. times as high as the cost per capita in the population as a * Exchange rates. The usual practice is to convert costs in whole. Thus, the cost per person of interventions that reach local currencies to some standard monetary unit, such as only those who are at risk should not be compared with the U.S. dollars, that can be compared among studies. But much lower costs of those that reach the entire population, exchange rates are distorted by speculation, government including those who are not at risk. M cronument Deficiencv Disorders 443 Appendix 19C. Criteria of Effectiveness ments have benefits that exceed costs, it is also important to consider whether the relation of benefits to costs exceed Some interventions will have a high success rate in obtaining those-or are at least in the range of-alternative investments. repletion, suci as injected or oral iodinated oil or oral capsules Although it would be desirable to have a standard cost- of vitamin A. Once ingested or injected, these interventions benefitmethodologywithpreciserulesforcalculationforevery are almost invariably associated with iodine or vitamin A situation, this is not the present case. The cost methodology is repletion. In contrast, medicinal supplementation with iron or straightforward and is identical to the ingredients, or resource dietary fortification does not always ensure repletion. Because recovery, method that was outlined in appendix 19B for cost the capacity of the body to store iron is limited, iron supple- and cost-effectiveness studies. But although the conceptual mentation requires that the participant take iron daily. When methods for identifying and measuring benefits are well estab- administered in schools or workplaces, this compliance can be lished (Creese and Henderson 1980; Mills 1985), the applica- readily maintained. Wlhe it is necessary to depeiid oi house- tion of these methods depends crucially on a variety of holds continually to take iron supplements, it is not realistic judgimients on both the measurement of benefits and their to expect a high level of compliance. Thus, the cost of deliv- values. Some of the best work on cost-benefit analysis in the ering the iron to households is not equivalent to the cost of health sector is found in the area of immunization (Creese and obtaining iron repletion. Indeed, obtaining complianice may Henderson 1980; Creese 1983), and many of the methods used require continuing reinforcement through monitoring and there can be applied to micronutrients. perstiasion by village health teams and other educational The basic method of estimating benefits is to identify the efforts. positive effects of micronutrient interventions on such areas as The same is true with fortification. Not only is it necessary morbidity, work output, and educational benefits for children. for all persons at risk to consume adequate amounts of the The benefits of reduced morbidity are generally considered to fortified food, but the food must have sufficient amounts of the he the savings in health care and the value of lost productivity; micronutrient at the time of consumption. There may be a the benefits of work output can be measured with respect to compliance problem when unfortified, local products compete additional days of productive work (in the labor market or with the nationally or regionally distributed fortified ones. In household) and the additional productivity per day; and edu- Ecuadoritwasnecessarytomountasocialmarketingcampaign cational benefits include the value of additional student to increase use of a fortified product such as iodinated salt achievement and the reduction in the cost of special educa- because alternative salt sources were available at the local level tional services or grade repetition. Some of these benefits also (Manoff 1987). In tropical areas the hygroscopic nature of salt have implications for costs. For example, if iron-replete work- that is used for iodine fortification means that unless contained ers are able to put out more work effort to increase productivity, in watertight packaging until consumption, at least some of the they will also need additional food to compensate for the iodine will be lost. lodinated salt in jute bags showed a loss of higher expenditure of energy (Levin 1985, 1986). three-quarters of its iodine in nine months (Mannar 1987). As summarized in the earlier sections of this chapter, each The type of packaging, the time it takes to get to consumers, of the micronutrient interventions has an effect on health, and the use of open or closed containers by shops and consum- productivity, and other aspects of behavior. In theory, it is only ers will determine potency. In very humid climates with highly necessary to translate the effects into benefits and to place untdependable transportation and long periods before sale or monetary values on them to compare them with the costs of consumption in open containers, the salt may lose virtually all an intervention. Unfortunately, the lack of field trials that its iodine. incorporate data collection in the various benefit domains limits the application of cost-benefit analysis to this area. Nevertheless, there exist studies for each of the three micro- Appendix 19D. Cost-Benefit Analysis nutrients that are both informative and suggest high returns. These are discussed in the main text of the chapter. Cost-benefit analysis represents a technique for ascertaining whether micronutrient and other social interventions are worthwhile. Such interventions use scarce societal resources, Appendix 19E. Costs and Benefits which could be used to provide other types of social benefits. At a minimum, an intervention should not be undertaken The tables in this appendix show the costs and benefits of unless its benefits exceed its costs. But because many invest- various interventions. 444 Henr-y M. Levin, Ernesto Pollitt, Rae Galloway, and Judith McGuire Table 19E- 1. Assumptions in Calculating Costs per Disability-Adjusted Life-Year, Death Averted, and Income Enhancement Parameter Value Program effectiveness (percent) 75' Unemployment (percent) 25 Life expectancy (years) 70 Discount rate (percent) 3 Annual wage rate (U.S. dollars) 500 Population (number) 100,000 Age distribution (number) 0-1 year 3,900 1-2 years 3,250 2-3 years 2,340 3-4 years 1,950 4-5 years 1,560 5-9 vears 12,000 10- 14 years 9,000 15-59 years 57,000 60 vears and older 7,000 Malnutnrion rates (number and percent PEM Children vounger than five 3,900 (30) Adults stunted from childhood malnutrition 17,000 (30) l ron Anemic children under age 15 18,000 (50) Anemic adult men 7,250 (25) Anemic pregnant women 2,520 (63) Total population anemic 49,000 Iodine Population deficient 24,000 (24)d Cretinism 50 (0.4) Vitamin A Deficient children under six 1,950 (15) Severely deficient children under six 40 (.27) Severely deficient children under six dying 20 (.16) Partially blind children under six 81 (0.060) Totally blind children under six 41 (0.028) Annual deaths from malnutrition (number) PEM-related causes in children under five 160 Severe anemia in women at childbirth 10 Stillbirths related to iodine deficiency 10 Neonatal deaths related to iodine deficiency 10 Children under five with vitamin A deficiency 40 Degree of disability (percent) Undernutrition 10 Iron deficiency 20 Iodine deficiency 5 Cretinism 50 Partial blindness 25 Total blindness 50 a. Includes coverage as well as efficacy. b. Aduilts age 15-59. c. Includes 25,000 women of reproductive age, of whom 4,000 are pregnant. d. One child is hom with cretinism each year. e. Health and productivity disab[lity. Source: Based on authors' assumptions. Micronutrient Deficiency Disorders 445 Table 19E-2. Nutrition Program Costs for Population of 10,000 Annual per capita cost Annual program cost Intervention Target group (U.S. dollars) (U. S. dollars) Food supplements Pregnant women 46 620,540 Children 0-3 years Nutrition education Pregnant women 2 26,980 Food subsidy Bottom quintile 30 600,000 Integrated nutrition PHC Pregnant women 25 337,250 School feeding Children 5-9 years 12 144,000 Iron Supplement Pregnant women 2 8,000 Fortification Entire population 0.20 20,000 Iodine Supplement, selective Women 0.50 12,500 Supplement, total Entire 0.50 23,250 Fortification Entire population 0.10 10,000 Vitamin A Supplement Children 0-5 years 0.50 6,500 Fortification Entire population 0.20 20,000 Note: Based on assumptions in table 19A-8. a. Assumes six prenatal visits plus 200 iron tablets. Source: Ho 1985; Levin 1985; Kennedy and Alderman 1987. Table 19E-3. Costs and Effectiveness of Iron Intervention Parameter Iron supplementation of pregnant women Iron fortification Target group Pregnant women All people Number 4,000 100,000 Average rate (percent) b 63 50 Per capita cost (U.S. dollars) 2 0.20 Program effectiveness (percent) 75 75 Deaths averted 10 10 Immediate productivity gains (percent) 20 20 Program duration (days) 200 Year round Program costs (U.S. dollars) 8,000 20 000d Discounted wage gains (U.S. dollars) 221,280 1,682,720 DALY gained 624 4,520 Wage gains divided by program cost 27.7 84.1 Cost per DALY (U.S. dollars) 12.80 4.40 Cost per death averted (U.S. dollars) 800 2,000 Note: Based on assumptions in table 19E-1. a. Rate of anemia for iron supplementation of pregnant women; rate of iron deficiency for iron fortification. b. Per pregnancy for iron supplementation; per participant for iron fortification. c. Calculated as the product of the number of anemic participants times disability times wage times effectiveness times employment, plus the product of number ofdeaths times wage times employment tinies productive life expectancy; ([0.63 x 39901 x 0.2 x 500 x 0.75 x 0.75) + (IQ x 500 x 0.75 x 21.3) = 141,400 - 79,880 = 221,280. d. Calculated as the product of the number of adult participants times the rate of anemia times disability times effectiveness times employment times wage, plus the product ofthe numberofdeaths times wage times employment times productive life expectancy; (56,990 x 0.5 x 0.2 x 0.75 x 500) + (10 x 500 x 0.75 x 21.3) = 1,602,840 + 79,880 = 1,682,720. e. Calculated as the product of the number of deaths times life expectancy, plus the product of disability times number of malnourished participants times ef- fectiveness; (10 x 24.7) + (0.2 x 0.63 x 3990 x 0.75) = 247 - 377 = 624. f. Calculated as the product of number of adult participants times the rate of anemia times disability times effectiveness, plus the product of the number of deaths times life expectancy; (56,990 x 0.5 x 0.2 x 0.75) + (10 x 24.7) = 4270 + 250 = 4520. Source: Based on authors' assumptions. 446 Henrn M. Levin, Ernesto Pollitt, Rae Galloway, and Judith McGuire Table 1 9E-4. Costs and Effectiveness of Iodine Intervention Iodine supplement: Iodine supplement: lodization of salt Parameter targeted coverage mass coverage or water Target group Reproductive-age women Everyone under age sixty Everyone Number 25,000 93,000 100,000 Average rate of iodine deficiency (percent) 24 24 24 Per capita cost (U.S. dollars) a 050b 0.50 0.10 Program effectiveness (percent) 75 75 75 Deaths averted 10 10 10 Productivity loss (percent) Normal population 5 5 5 Cretins 50 50 50 Program duration Year round Year round Year round Program costs (U.S. dollars) 12,500d 46,500 100,000 Discounted wage gains (U.S. dollars) 172,000 280,000 280,000: DALY gained 660 1,2709 1,335 Wage gains divided by program cost (U.S. dollars) 13.8 6.0 28 Cost per DALY (U.S. dollars) 18.90 37 7.50 Cost per death averted (U.S. dollars) 1,250 4,650 1,000 Note: Based on assumptions in table 19E-1. a. Per participant per year. b. Prevents both neonatal death and cretinism. c. Neonatal. d. Calculated as the product of the number of participants times the rate of deficiency times disability times wage times effectiveness times employment rate, plus number who died times productive life expectancy times employment times wage for ten cretins, plus the product of frequency times productive life expectancy times employment times wage for ten deaths; (25,000 x 0.24 x 0.05 x SOO x 0.75 x 0.75) + (IQ x 0.5 x 15.5765 x 0.75 x 500) + (lQ x 15.5765 x 0.75 x 500) = 84,380 + 29,210 + 58,410 = 172,000. e.Calculatedas innoted; (57,000 x 0.24x 0.05 x 0.75 x 0.75 x 500) + (IQ x Q.5 x 15.5765 x0.75 x 500) + (10 x 15.5765 x 0.75 x 500) = 192,380 + 29,210 + 58,410 = 280,000. f. Calculated as the product of the number of participants times the rate of deficiency times disability times effectiveness, plus the product of disability times life expectancy for ten cretins, plus the life expectancy for ten deaths; (25,000 x 0.24 x 0.05 x 0.75) + (10 x 0.5 x 29) + 10 x 29 = 225 + 145 + 290 = 660. g. Calculated as in note f; (93,000 x 0.24 x 0.05 x 0.75) + (10 x 0.5 x 29) + lQ x 29 = 837 + 145 + 290 = 1270. h. Calculalated as in note f; (99,980 x 0.24 x 0.05 x 0.75) + (10 x 0.5 x 29) + 10 x 29 = 900 + 145 + 290 = 1335. Source: Table 19E- I using methodology described in d (above). Table 19E-5. Cost and Effectiveness of Vitamin A Intervention Parameter Vitamin A supplementation' Vitamin A fortification Target group Children under five Entire population Number 13,000 100,000 Average rate of vitamin A deficiency (percent) 15 15 Per capita cost (U.S. dollars): 0.50 0.20 Program effectiveness (percent) 75 75 Deaths averted (number) 20 20 Blindness averted (number) Total 4 4 Partial 8 8 Productivity loss (percent) Totally blind 50 50 Partially blind 25 25 Program duration Year round Year round Program costs (U.S. dollars) 6,500d 20 000d Discounted wage gains (U.S. dollars) 140,188 140,188 DALY gained 6961 696 Wage gain divided by program cost 21.6 7.0 Cost per DALY (U.S. dollars) 9.3 29 Cost per death averted (U.S. dollars) 325 1,000 Note: Based on assumptions in table 19E-I. a. Semiannual mass dose. b. In children under five. c. Per participant. d. Does not include losses due to excess child morbidity. Calculated as the product of the number of deaths averted times the productive life expectancy times employment times wage, plus the product of the number of total blindness averted times productive life expectancy times disability times employment times wage, plus the product of the number of partial blindness averted times productive life expectancy rimes disability times employment times wage; (20 x 15.5765 x 0.75 x 500) + (4 x 15.5765 x 0.5 x 0.75 x 500) + (8 x 15.5765 x 0.25 x 0.75 x 500) = 116,824 + 11,682 + 11,682 = 140,188. e. Calculated as deaths averted times discounted remaining life expectancy plus total blindness times disability times discounted remaining life expectancy plus partial blindness times disability times discounted remaining life expectancy; (20 x 29) + (4 x 0.5 x 29) + (8 x 0.25 x 29) = 696. Source: Based on authors' assumptions. Micronutrient Deficiency Disorders 447 Notes Beaton, G. H., and J. M. Bengoa. 1976. Nutrinon in Preventative Medicine. Geneva: WHO Nutrition Unit. 1. In developing countries, 50 percent of children from birth to fourteen Bender, D., and D. Yoder. 1983. The Village Health Worker in Review: An years, 25 percent of adult men and women, and 63 percent of pregnant women Annotated Bibliography. Monticello, 111.: Vance Bibliographies. are iron deficient; 24 percent of the total population are iodine deficient, and Berg, Alan, and Susan Brems. 1986. "Micronutrient Deficiencies: Present 0.4 percent is cretinous because of iodine deficiency; 15 percent of children Knowledge on Effects and Control." Technical Notes 32. Population, under five are deficient, 2.5 percent are severely deficient in vitamin A, 1.4 Health and Nutrition Department, World Bank, Washington, D.C. percent die of severe deficiency, 0.28 percent become totally blind, and 0.66 Bloem, M. W., M. Wedel, E. J. van Agtmaal, A. J. Speek, S. Saowakortha, and percent become partially blind. W. H. P. Schreurs. 1990. "Vitamin A Interventions: Short-Term Effects of Thirteen percent of the population of developing countries are under five a Single, Oral, Massive Dose on Iron Metabolism." American Journal of years of age; 21 percent are five through fourteen; 57 percent are economically Clinical Nutmrinon 51 :76-79. active adults, 44 percent of whomare women of reproductive age of whom 16 Bothwell, T. H.. and R. W. Charlton. 1991. Iron Deficiency in Women. percent are pregnant; 7 percent are sixty years of age or over. Adults were International Nutritional Anemia Consultative Group, Washington, D.C. assumed to be economically active between fifteen and fifty-nine years of age. Brabin B. . 1988. "Consequences of Materal Anemia on Fetal and Early Excess mortality due to severe vitamin A deficiency was assumed to be I Brant Dequen a Matemic Area onatal andiEarly percent of the target age group, or about 41 percent of the severely deficient Infant Development in a Malaria Endemic Area: Perinatal Morbidity and children; excess mortality due to iron deficiencv was assumed to be I out of Mortalit."GPaperpresenteoatAtheinternatinalaNutrtionalbAemia Con 400 pregnant women; excess mortality due to iodine deficiency was assumed sut8v Gr t nemr to be 1 neonatal death out of 96 pregnancies of iodine-deficient women. 1988. Geneva. Partial blindness: 25 percent disability; total blindness: 50 percent disability; Chafkin, S. 1984. "A Note on Iron Deficiency and World Bank Consideration iron deficiency: 20 percent disability; iodine deficiency: 5 percent disability; Thereof." Unpublished report. Population Health and Nutrition Depart- cretinism: 50 percent disability. ment, World Bank, Washington, D.C. Iron supplementation: $2.00 per pregnancy; iron fortification: $0.20 per Chandra, R. K. 1973. "Reduced Bactericidal Capacity of Polymorphs in Iron capita per year; iodine supplementation: $0.50 per recipient per year; iodine Deficiency." Archives of Disease in Childhood 48:864466. fortification: $0.10percapita peryear; vitamin A supplementation: $0.50 per , ed. 1988. NumritionandImmunology. New York: Alan B. Liss. recipient per year; vitamin A fortification: $0.20 per capita per year. Chandra, R. K., and B. Au. 1981. "Single Nutrient Deficiency and Cell-Me- diated Immune Responses. 3. Vitamin A." Nutrition Research 1:181-85. Chandra, R. K., and S. Puri. 1985. "Trace Element Modulation of Immune References Responses and Susceptibility to Infection." In R. J. Chandra, ed., Trace Elements in Nutrition and Children. Nestle Nutrition Workshop Series, no. ArToyave, G. 1982. "The Program of Fortification of Sugar with Vitamin A in 8. New York: Raven Press. Guatemala." In N. S. Scrimshaw and M. Wallerstein, eds., Nutrition Policy Charoenlarp, P., S. Dhanamitta, R. Kacwvichit, A. Silprasert, C. Suwanaradd, and Implementation. New York: Plenum Press. S. Na-Nakom, P. Prawatmuang, S. Vatanavicharn, U. Nutcharas, P. Arroyave, G., J. R. Aguilar, M. Flores, and M. A. Guzman. 1979. "Evaluation Pootrakul, V. Tanphaichitr, U. Thanangkul, T. Vaniyapong, Thane Toe, of Sugar Fortification with Vitamin A at the National Level." Scientific A. Valyasevi, S. Baker,J. Cook, E. M. DeMaeyer, L. Garby, and L. Hallberg. Publication 384. Pan-American Health Organization, Institure of Nutrition 1988. "A WHO Collaborative Study on Iron Supplementation in Burma and of Central America and Panama, Washington, D.C. Thailand." Americanjournal of Clinical Numtion 47:280-97. Assami, M., S. Hercberg, S. Assami, P. Galan, A. Assami, and G. Potier de Chwang, L., A. G. Soemantri, and E. Pollitt. 1988. "Iron Supplementation Courcy. 1988. "Iron and Folate Status in Algerian Pregnant Women." and Physical Growth of Rural Indonesian Children." American Joural of Ecology of Food and Nutrition 21:181-87. Clinical Numtion 47:496-501. Aukett, Ni. A., Y. A. Parks, P. H. Scott, and B. A. Wharton. 1986. "Treatment Clugston, G. A., E. M. Dulberg, C. S. Pandav, and R. L. Tilden. 1987. "lodine with Iron Increases Weight Gain and Psychomotor Development." Archives Deficiency Disorders in South East Asia." In B. S. Hetzel, ]. T. Dunn, and of Disease in Childhood61:849-57. J. B. Stanbury, eds., The Prevention and Control of Iodine Deficiency Disorders. Austin, J. E., T. K. Balding, D. Pyle, F. S. Solar, T. L. Femandez, M. D. Latham, New York: Elsevier. and B. M. Popkin. 1981. Nutrition Intervention in Developing Countries: Clydesdale, F. M., and K. L. Wiemer, eds. 1985. Iron Fortification of Foods. Fortification. Study 3. Cambridge, Mass.: Gunn and Hain. Orlando, Fla.: Academic Press. Bagchi, K., M. Mohanram, and V. Reddy. 1980. "Humoral Immune Response Cohen, N., H. Rahman, J. Sprague, M. A. Jahil, E. Leemhuis de Regt, and M. in Children with Iron-Deficiency Anemia." Bnrtish Medical Journal Mitra. 1985. "Prevalence and Determinants of Nutritional Blindness in 280:1249-51. Bangladeshi Children." World Health Statistics Quarterly 38:317-30. Baker,S.J.,andE. M.DeMaeyer. 1979."NutritionalAnemia:ltsUnderstand- Cook, J. D., and M. E. Reusser. 1983. "Iron Fortification: An Update." ing and Control with Special Reference to the Work of the World Health American Joumal of Clinical Nutrition 38:648-59. Organization." AmericanJournal of Clinical Nutrition 32:368-417. Cortea, H. 1980. "A Cost-Benefit Study of Iodine Supplementation Pro- Baly, D. L., M. S. Golub, M. E. Gershwin, and L. S. Hurley. 1984. "Studies of grams for the Prevention of Endemic Goiter and Cretinism." In J. B. Marginal Zinc Deprivation in Rhesus Monkeys. 3. Effects of Vitamin A Stanbury and B. S. Hetzel, eds., Endemic Goiter and Endemic Cretinism. New Metabolism." AmericanJournal of Clinical Numtion 40:199-207. York: Wiley. Barden, H. S., and R. Kessel. 1985. "The Costs and Benefits of Screening for Creese, A. L. 1983. "The Economic Evaluationof Immunization Programmes." Congenital Hypothyroidism in Wisconsin." Social Biology 31:185-200. In K. Lee and A. M ills, eds., The Economics of Health in DevelopingCountries. Basta, S., D. Sockirman, D. Karyadi, and N. S. Scrimshaw. 1979. "Iron New York: Oxford University Press. Deficiency Anemia and the Productivity of Adult Males in Indonesia." Creese, A. L., and R. H. Henderson. 1980. "Cost-Benefit Analysis and AmericanJournal of Clinical Nutrition 32:916-25. Immunization Programmes in Developing Countries." Bulletin of the World Bautista, A., P. A. Barker, J. T. Dunn, M. Sanchez. and D. L. Kaiser. 1982. Health Organization 58:491-97. "The Effects of Oral Iodized Oil on Intelligence, Thyroid Status, and Creese, A. L., N. Sriyabbaya, G. Casabal, and G. Wiseso. 1982. "Cost-Effec- SomaticGrowth inSchool-AgeChildrenfrom anAreaofEndemicGoiter." tiveness Appraisal of Immunization Programmes." Bulletin of the World American Journal of Chnical Nutrition 35:127-34. Health Organization 60:621-32. 448 Henrs- M. Lein, Ernesto Pollitt, Rae Galloway , and Judith McGuire Dallman, P. R. 1987. "Iron Deficiency and the Immune Response." Arerican Griffiths, M. 1980. Concept Testing Nutrition Communication and Behavior Journal of Clinical Numnton 46:329-34. Change Components. Vol. 1. Unpublished report on the Indonesia Nutri- Daeschner, C. W., 111, M. C. Matustik, U. Carpentieri, and lvi. E. Hagggard. tion Development Program. Manoff International. 1981. "Zinc and Growth in Patients with Sickle Cell Anemia Disease." Hallberg, L. 1981. 'Effect of Vitamin C on the Bioavailability of Iron from Journal of Pedianics 98:778-80. Food." In J. N. Counsell and D. H. Hornig, eds., Vitamin C. Englewood, Delong, R. 1987. "Neurological Involvement in Iodine Deficiency Disorders." N.J.: Applied Science Publishers. In B. S. Hetzel, J. T. Dunn, and J. B. Stanbury. eds., The Pretention arid Harrison, K. A. 1975. "Maternal Mortality in Anemia in Pregnancy." West Control of lodine Deficiency Disorders. New York: Elsevier. Africani Medicai Journal: 27-31. DeMaeyer, E. 1986. "Xerophthalmia and Blindness of Nutritional Origin in Hercberg, S.. P. Galan, M. Chauliac, A. M. Masse-Raimbault, M. Devanlay, the Third WVorld." Children in the Tropics 165. Intemational National S. Bileoma, E. Alihonou, 1. Zohoun, J. P. Christides, and G. Potier de Children's Centre, Paris. Courcy. 1987. "Nutritional Anaemia in Pregnant Beninese Women: Con- .1989. "Preventing and Controlling Iron Deficiency Anaem ia through sequences on the Haematological Profile of the Newbom." Bntish]ournal of Pnmary Health Care." WHO Nutrition Unit, Geneva. Nutrition 57:185-94. DeMaeyer, E., and M. Adiels-Tegman. 1985. "The Prevalence of Anemia in Het-el, B. S., ed. 1978. Basic Health Care in Developing Countries: An Epidemi- the World. ' W'orld Health Stansucs QuorterN 38:302-16. ological Perspective. Oxford: Oxford University Press. the WOTid." W"orld Health Swtis___ 183."Ioine efiieny Dsordrsend hei Eraicaion" Lnce Derman, D. P., M. H. Sayers, S. R. Lynch, R. W. Charlton, T. H. Bothwell. 193. "iodine Deficiency Disorders and Their Eradicatio9." Lancet and F. G. H. Mayet. 1977. "Ir(on Absorption from Cereal-Based Meai 2(8359):1126-29. Containing Cane Sugar Fortified with Ascorbic Acid." 3rinsh Journul of . 1987. "An Overview of the Prevention and Control of Iodine Defi- Nutrition (2)261-69. ciency."In B. S. Hetzel,J. T. Dunn, and J. B. Stanbury, eds., The Prevetion Djukanovic, V., and E. P. Mach, eds. 1975. Alternative Approaches to Meetng and Control of Iodine Deficiency Disorders. New York: Elsevier. Basic Health Needs in DevelopingCountries. Joint United Nations Children's .1988. "The Prevention and Control of IDD." State-of-the-Art Series, Fund/World Health Organization Study. Geneva. Nutrition Policy Discussion Paper 3. Administrative Committee on Coor- Dunn, J. T. 1987. "Iodized Oil in the Treatment and Prophylaxis of IDD." In dination, Subcommittee on Nutrition, United Nations, New York. B. S. Hetzel, J. T. Dunn, and J. B. Stanbury, eds., The Prevention andControl lFletzel, B. S., and G. F. Maberley. 1986. "Iodine." Trace Elernents in Human of Iodine Deficiency DLSorders. New York: Elsevier. and Animal Nutrition 2:139-208. Dunn J., ed. I 989a. "Status of inD." ID) (International Council for Control of Hetzel, B. S., and B.J. Potter. 1983. "lodine Deficiency and the Role ofThyroid Icodine Deficiency Disorders) Newsletter 5:1. Hormones in Brain Development." In 1. E. Dreosti and R. M. Smith, eds., Neurobiology of the Trace Elerments. Vol. 1, Trace Element Neurobiology and .1989b. "IDD in Southeast Asia." lIDD (International Council for Deiinis Clfo.N HuaaPes Contrcol of Iodine Deficiency Disorders) Newsletter 5(2):1-24. Deficiencies. Clifton, N.J.: Humana Press. Hetzel, B. S., C. H. Thilly. R. Fierro-Benitez, E. A. Pretell, l. H. Buttheld, and 1989c. "Staus of IDD Control in Ten East, Central and South African J. B. Stanhury. 1980. "Iodized Oil in the Prevention of Endemic Goiter and Countries." IDI (International Council for Control of Iodine Deficiency Cretinisnm." In J. B. Stanbury and B. S. Hetzel, eds., Endemic Goiter and Disorders) New'sletter 5(1): 1-8. Endemic Cretinism. New York: John Wiley and Sons. Eastman, S. J. 1987. "Vitamin A: Deficiency and Xerophthalmia: Recent Higashi O., Y. Sato, M. Takamura, and M. Oyama. 1967. "Mean Cellular Findings and Some Programme Implications." Assignrnent Children 1987-3 Peroxidase (McP) of Leukocytes in Iron Deficiency Anemia." TohokuJournal UNICEF, New York. of Expenmental Medicine 93:105-9. FAO (Food and Agriculture Organization). 1965. "Report of a Joint FAO/WHO Ho, T. J. 1985. "Economic Issues in Assessing Nutrition Projects: Costs, ExpertGroup on Vitamin A." FAO Food Policy and Nutrition Division/wl-o Affordability, and Cost Effectiveness." Technical Note 85-14. Population, Nutrition Unit. Health, and Nutrition Department, World Bank, Washington, D.C. - 1970. "Report of a Joint FAo/wHO Expert Group on Iron." FAO Food Hunt, J. R., L. M. Mullen, G. 1. Lykken, S. K. Gallagher, and F. H. Nielsen. Policy and Nutrition Division/WHco Nutrition Unit. 1990. "Ascorbic Acid: Effect on Ongoing Iron Absorption and Status in 1989. "NutritIonCountry Profiles, 1988-89." Food Policy andNutri- Iron-Depleted Young Women." American Jounal of Clinical Nurntion tion Division. 51:649-55. Fleming, A. F. 1989. "Consequences ofAnaemia in Pregnancy on the Moth- IcCIDDWrHv (World Health Organization). 1989. "Report on the Develop- ers." Liverpool School of Tropical Medicine, Departmenit of Tropical Med- ments in XDD Control in Africa Region." Paper presented at the IccIDo 4th icine and Infectious Diseases, Liverpool, Eng. Annual Meeting, March 11-12, 1989, New Dehli. Florentino, R. F. 1988. "Nutritional Anemia Control Program in the Philip- INACG (Intemational Nutritional Anemia Consultative Group). 1977. Guide- pines." Case Study presented at the Intemational Nutritional Anemia lines for the Eradication of Iron Deficiency Anemia. Washington, D.C. Consultative Group (INAC), November 14. - 1979. Iron Deficiency in Infancy and Childhood. Washington, D.C. Florentino, R. F., and R. M. Guirriec. 1984. "Prevalence of Nutritional . 1989. GuidelinesfurtheControlofMaternalNumnonalAnemia. Wash- Anemia in Infancy and Childhood with Emphasis on Developing Coun- ington, D.C. tries." In Abraham Stekel, ed.. Iron Numtion in Infancy andChildhood. Nestle Ingenbleek, Y., and M. De Visscher. 1979. "Hormonal and Nutritional Status: Nutrition Workshop Series 4. New York: Raven Press. Critical Conditions for Endemic Goiter Epidemiology?" Metabolism Gaitan,J. E., L. G. Mayoral, and E. Gaitan. 1983. "Defective Thyroidal Iodine 28(1):9-19. Concentration in Protein Calorie Malnutrition." Journal of Clinical Endo- Irie, M., Kuroda, K. Nakamura, K. Tsuboi, S. Takeda, K. Inonie, K. Enomoto, crnology and Metabolism 57:327-33. i. Arifs, P. Adji, and R. Djokomoeljanto. 1986. "Study of Endemic Goiter Gardner, G. W., V. R. Edgerton, B. Sersewiratne, R. J. Barnard, and Y. Ohira. In Indonesia: Thyroid Function, GoLiter Prevalence, andControl Programs." 1977. "Physical Work Capacity and Metabolic Stress in Subjects with Iron In G. Neto-Medeiros, R. M. B. Maciel, and A. Halpem, eds., Iodine Deficiency Anemia." Amencan Journal of Clinical NumHnon 30:910-1 7. Deficiency Disorders and Congenital Hypothyroidism. Ache, Indonesia. Greene, L. S. 1977. "H yperendemic Goiter, Cretinism, and Social Organiza- IVACci (International Vitamin A Consultative Group). 1989. "Report on the tion in Highland Ecuador." In L. S. Greene, ed., Malnumntion, Behavior, and National Symposium and 13th IVACG Meeting," Katmandu, Nepal, Social Organization. New York: Academic Press. November 5-IQ. Micronutrient Deficiency Disorders 449 Kennedy, E. T., and H. H. Alderman. 1987. "Comparative Analyses of Jhaveri,eds Boimbay:FederationofObstetricandGynaecologicalSocieties Nutritional Effectiveness of Food Subsidies and Other Food-Related Inter- ot India. ventions." Joint WH0,MyNICEF Nutrition Support Program. International Mata, L. 1977. TheChildren on SantaMariaCauque. Cambridge, Mass.: MiT Press. Food Policy Reseatch Institute, Washington, I).C. Mathur, G. P., and K. P. Kushwaha. 1987. "Vitamin A Deficiency: Review Klebanoff, S. ]. 1970. "Myeloperoxidase: Contribibtion to the Microbicidal Arricle." Indian Pediatrics 24:573-81. Activity nf Intact Leukocytes." Science 169:1095-97. Medeiros-Neto, G., R. Djokomoeljanto. M. Benmiloud, M. C. DeBlanco, M. Ksanga, Pauline, Free Pepping, and Festo Kavishe, coinps. 1985. "Proceedings Irie, T. Z. Lui, C. R. Ackardt. 1987. "TThe Monitoring and Evaluation of of a Workshop in the Control of Vitamin A Deficiency and Xerophthalmia Iodine Deficiency Control Programs: Report of an ICCIDD Committee." In in Tanzania." Tanzanian FOod and Nutrition Center Report 980. Dar es B. S. Hetzel, J. T. Dunn, and J. B. Stanbury, eds., Prevention and Control of Salaam, Tanzania. Iodine Deficiency Disorders, New York: Elsevier. Levin, H. M. 1983. Cost-Effectiveness: A Pnmer. Beverly Hills, Calif.: Sage Mejia, L. A., and G. Arroyave. 1982. "The Effect of Vitamin A FortificatLon Publications. of Sugar on Iron Metabolism in Preschool Children in Guatemala." Amer- . 1985. A Benefit-Cost Analysis of Nutninonal Intervenntios for Anemia ican journLal of Clinical Numtion 36:87-93. Reduction. Population, Health, and Nutrition Technical Note 12. Popula- Mepia, L. A., and F. Chew. 1988. "Hematological Effect of Supplementing tion, Health, and Nutrition Department, World Bank, Washington, D.C. Anemic Children with Vitamin A Alone and in Combination with Iron." . 1986. "A Benefit-Cost Analysis of Ntiritional Programs for Anemia AmericaniJournal ofClinical Nutntion 48:5954600. Reduction." W'orld Bank Research Observer 1 (2):21946. Mills, A. 1985. "Economic Evaluation of Health Programmes: Application of . 1987. "Economic Dimensions of lodine Deficiency Disorders." In B. the Principles in Developing Countries." W'orld Health Statistcs Quarterly S. Hetzel, J. T. Dunn, and J. B. Stanbury, eds., The Prevention and Control 38:368-82. of Iodine Deficiency Disorders. New York: Elsevier. Mishan, E. 1976. Cost-BenefitAnalysis. New York: Praeger Publishers. Lieberman, A., K. J. Ryan, R. R. Monson, and S. C. Schoenbaum. 1987. "Risk Monsen, E. R., L. Hallberg, M. Layrisse. D. M. Hegsted, ]. D. Cook, W. Metz, Factors tor Racial Differences in the Rate of Premature Births." Neu, England and C. A. Finch. 1978. "Estimation of Available Dietary Iron." Amencan Journal of Medicine 317:743-48. Jo urnal of Cluttical Nutninon 31:134- 41. 1988. "Association of Maternal Hemarocrit with Premature Labor." Muhilal, A. Mudeana, E. Azis, S. Saidin, A. B. Jahari, and D. Karyadi. 1988. American Journal of Obstetrics and Gynecolog 159:107-114. "Vitamin A-Fortified MSG and Vitamin A Status: A Controlled Field Trial." Lozoff, B. 1989. "Methodological Issues in Studying Behavioral Effects on Arnerican]ournalofClinicalNutrinon48:1265-70. Infant Iron-Deficiency Anemia." American Journal of Clinical Nurition Muhilal, D. Permeisih, Y. R. Idjradinata, Muherdiyantiningsih, and D. 50(supplement):641-54. Karyadi. 1988. "Vitamin A-Fortified MSG and Health, Growth, and Survival Lozoff, B., and G. M. Brittenham. 1985. "Behavioral Aspects of Iron Defi- ofChildren:AControlledFieldTrial."Amencan]ournalofClincalNutrtion ciency." Progress in Hematology 14:23-53. 48:1271-76. Lozoff, B., G. M. Brittenham, F. E. Viteri, and 1. J. Urrutia. 1982. "Behavioral Murphy,]. F.,. O'Riordan, R. G. Newcombe, and E. C. Coles. 1986. "Relation Abnormalities in Infants with Iron Deficiency Anemia." In E. Pollitt and of lHaemoglobin Levels in First and Second Trimesters to Outcome of R. Leibel, eds., Iron Deficiency: Brain Biochemistry and Behavior. New York: Pregnancy." Lancet 1:992-94. Raven Press, 183-95. National Academy of Sciences. 1987. "Prevention and Control." Food and Lozoff, B., G. M. Brittenham, F. E. Viteri, A. W. Wolff. and J. J. Urrutia. 1982. Nutrition Board, Committee on International Nutrition Programs, Sub- "The Effects of Short-Term ODral Iron Therapy on Developmental Deficits committee on 'v'itamin A Deficiency, Washington, D.C. in Iron-Deficient Anemic Infants."Journal of Pediamcs 100:351-57. Nauss, K. M. 1986. "Influence of Vitamin A Status nn the Immune System." In J. C. Bauemfeind. ed., Vitamin A Deficiency and ILs Control. London: Macfarlane, B. J., W. B. Vander Riet, T. H. Bothwell, R. D. Baynes, D. Academic Press. Sieganberg, U. Schmidt,A.Tal,J. R. N.Tayler, and F. Mayet. 1990. "Effects ofiraditional Oriental Soy Pr.ducts o Iron Absorptio." A-].,d Nishi, Y., F. Lifshitz, M. A. Bayne, F,. Daum, M. Silverberg, and H . Aiges. 1980. of Climncal Nution 51 :873-80.p mcicanourna "Zinc Status and Its Relation to Growth Retardation in Children with Chronic Inflammatory Bowel Disease."AmencanjowrualofClinicalNumnon 33:2613-21. McMichael, A. J., J. D. Potter, and B. S. Hetzel. 1980. "Iodine [)eficiency, O Thyroid Function, and Reproductive Failure." In J. B. Stanbury and B. S. Olson A "P inlogic. a nd ed., Vitasin a Siens of lletzel, eds., EndernicGoiterandfEndermcCretinism. New York: Wiley. 445460. Vitamin A DeficLency." In], C. Bauernfeind, ed., ViaPnAsDeficiencs and Ito ContrDol. Loncdon: Academic Press. MLaren,D. s.1966." PresentKnowledgeofthe Roleof VitaminAl MiHeailth Oski, F. A., A. S. Honig, B. Helu, and P. Howanitz. 1983. "Effect of Iron and Disease." Transacntios of the Royal Society of Tropical Medicine and Thrp on BeairPromnei'oaei,Io eiin nat. H-ygiene. 60:4 36-62. Therapy o n Behavior Performance in Nonanenmic, fron Deficient Infants." Hygiene. 60:436-62. ~~~~~~~~~~Pediatncs 71:87 7-80. Mamdansi, NI., and D. A. Ross. 1988. "Vitamin A Supplementation and Child PAHO (Pan-American Health Organization)/WHto (World Health Organiza- Survival: Magic Bullet or False Hope ?" Evaluation and Planning Centre for tion)/UNICEF (United Nations Children's Fund). 1988. "Expanded Program Health Care, 19. London School of Hygiene and Tropical Medicine, for the Control of Iodine Deficiency Disorders in Latin America." Wash- Mannar, M. G. Venkatesh. 1987. "Control of Iodine Deficiency Disorders by ington, D.C. lodination of Salt Strategy for Developing Countries." In B. S. Hetzel, J. Pardo. A. S. 1990. "The Pronalcobo: The National Program to Fight against T. Dunn, and ]. B. Stanbtiry, eds., The Prevention and Control if Iodine Goiter in Bolivia." A Case Study prepared for the Population, Health, and Deficiency Disorders. New York: Elsevier. Nutrition Division, World Bank, Washington, D.C. Manoff, R. K. 1987 "Social Marketing: New Tool to Combat Iodine Defi- Pharoah, P. 0. D.. K. J. Connolly, R. P. Ekins, and A. B. Harding. 1984. ciency Disorders." In B. S. Hetzel, J. T. Dunn, and J. B. Stanbhiry, eds., The "Matemal Thyroid Hlormone Levels in Pregnancy and the Subsequent Prevention and Control of Iodine Deficiency Disorders. New York: Elsevier. Cognitive and Motor Performance of the Children." Clinical EndoLrinotogy Marani, L., S. Venturi, and R. Nasala. 1985. "Role of Iodine in Delayed 21:265-70. Immtine Respionse." Israeljournal of Medical Science. 21:864. Pollitt, Emesto. 1987. "Effects of Iron Deficiency on Mental Development: Masani, K. M. 1969. Proceedings of the International Seminar on Maternal Methodological Considerations and Substantive Findings." In F.Johnston, Morbidity, Familny Planning, Biology of Reproduction. Purandare and C. L. ed., Nutritional Anthropology. Neu York: Alan R. Liss. 450 Henrn Mi. Levin, Ernesto Pollitt, Rae Galloway, and Judith McGuire .1989. 'Report of a Mission on a Nutrition Component for the World Srikantia, S. G., C. Bhaskaram, J. S. Prasad, and K. A. V. R. Krishnamachari. Bank'c Basic EdUcation Project in the Dominican Repiblic." Population 1976. "Anaemia and Immune Response." Lancet 2:307-9. and HUman Resources Department. World Bank, Washington, D.C. Stanhury, J. B. 1987. "The Iodine Deficiency Disorder: Introduction and 1990. Mainutirinon and Infection in the Classroom. Paris: UNESec'. General Aspects." In B. S. Hetzel, J. T. Dunn, and J. B. Stanbury, eds., The Pollitt, Emesto, and R. Leibel, eds. 1982. Iron Deficiency- Brami Biochemisrrn- Prevention and Control of Iodine Deficiency Disorders. New York: Elsevier. and Behavior. New York: Raveni Press. Stanton, B. F.,]. D. Clemens, B. Wojtyniak, and T. Khair. 1986. "Risk Factors Popkin, B. MN., F. S. Solon T. Fernandez, and Mi. C. Latham. 1980. "Benefir- for Developing M ild Nutritional Blindness in Urban Bangladesh." Anerican Cost Anialvsis in the Nutrition Area: A Project in the Philippines." Social Jounal of Diseases of Chidshood 140:584-88. Science andMedicine 14:207-16. Stekel, A. 1987. Iron Nutrition in Infancy and Childhood. Nestle Nutrition Prasad, A. S. 1979. "Leucocyte Function in Iron-Deficiency Anemia." Amer- Workshop Series. New York: Raven Press. icanJounial of Clinical Nutrition 32:550-52. Stephenson, L. S. 1987. The Impact of Helmninth Infectons on Hutman Numntion. Prual, A., P. Galan, L. De Berris. and S. Hercherg. 1988. "Evaluiation of Iron London: Tavlor and Francis. Status in Chadian Pregnant Women: Consequences of Maternal Iron Stephenson, L. S., M. C. Latham, K. M. Kurz, D. Miller, S. N. Kinoti, and M. Deficiency on the Haemotipolieric StatusofNewbooms."TropicalGeograph- L. Oduori. 1985. "Urinary Iron Loss and Physical Fitness of Kenyan Chil- ical Medicme 40:1-6. dren with Urinary Schistosomiasis." Amnerican Jounsal of Tropical Medicine Rahmathullah, Laxmi, Barbara Underwood, Ravilla Thulasiral, Roy Milton, anid Hygiene 34:322-30. Kala Radinaswamy, Raheeiss Rahismathulleh, and Ganieesh Babu. 1990. "Re- Suliardjo. 1986. The Effect of Iron Intervention on Work Producdtith of Tea Pickers. duced Mortality among Children in Southern India Receiving a Small Bogor Agricultural University, Faculty of Agriculture, Bogor, Indonesia. Weekly Dose of Vitamin A." NeuwEnglaridjoumnlofMedicine 32( 134):929-35. Talbot. M. C., L. T. Miller, and N. I. Kerkvlirt. 1987. "Pyridoxine Supplemen- Schaefer. A. E. 1974. "Status of Salt lodination in PAHO Member Countries." tanton: Effect on Lymphocyte Responses in Elderly Persons." Amencan In !. T. Dunn and G. A. Medeiros-Neto, eds., Enderric Goiter and Cretinism: Journal of Clinical Nutrntion 46:65944. Continuing Threants to World Health. Report on the 4th Meeting of the The Task Force for Child Survival. 1989. "Iodine Deficiency." Wlorld Im- Pan-American Health Organization Technical Group in Endemic Goiter. munization Neus 5(4):24-25. Pan-American Health Organization, Pan-Americani Sanitary Bureau, Re- Thilly, C. H. 1981. "Goitre et crktinisme end6m1ques: role etiologique de la gional Office of the World Health Organization, Washington, D.C. consomimation de manioc et strategie d'6radtcatmon." Bulletzn de Acackmie Scrimshaw, N. S. 1984. "Functional Consequences of Iron Deficiency in Roale deMedecine Belgique 136:389-412. Human Populations.'JournalofNumw.onalSaencesand Vitanrunology 30:47463. Thillv, C. H., and B. S. Hetzel. 1980. "An Assessment of Prophylactic Seshadri, S., and T. Gopaldas. N.d. "Magnitude anid Implication of the Programs: Social, Political, Cultural, and Economic Issues." In J. B. Problem of Nutritional Anemia." University of Baroda, India, Department Stanbury and B. S. Hetzel, eds., Endemic Goiter and Endemic Cretinism. New of Food and Nutrition. York: John Wiley and Sons. _ 1989. "Impact of Iron Supplementation on Cognitive Functions in Thilly, C. H., R. Lagasse, G. Roger, P. Bourdoux, and A. M. Ermans. 1980. Preschool and School-Aged Children: The Indian Experience." American "ImpairedFetalandPostnatalDevelopnientandHighPerinatalDeath-Rate Joumal of Clinical Nutrition 50(supplement):675-86. in a Severe Iodine Deficient Area." In J. R. Stockigt and S. Nagataki, eds., Shils, N-M. E., and V. R. Youing. 1988. Moderi Nutrition in Health and Disease. Thyroid Research 8. Proceedings of the Eighth International Thyroid Congress. Philadelphia: Lea and Fehiger. Canberra: Australian Academy of Science. Simnmons, W. K. 1990. "Evaluation of a Novel Delayed-Release Formulation Tielsch, J. M., and A. Sommer. 1984. "The Epidemiology of Vitamin A for Iron Supplementation in Pregnancy." Report 5. Intemational Center for Deficiency and Xerophthalmia." Annual Review of Numnon 4:183-285. Research on Women. Washington, D.C. Tilden, R. L., and R. N. Grosse. 1988. "Vitamin A Cost-Effectiveness Model." Soewondo,N M., Ni. HUsaimi, and E. Pollitt. 1989. "Effects of Iron Deficiency School of Public Health, University of Michigan, Ann Arbor. Typescript. On Attention and Learning Processes in Preschool Children: Bandung, Tomkins, A., and F. Watson. 1989. Malnutrinon and Infection. State-of-the- Indonesia." AmencanJournalofClinicalNumtion 50(supplenient):667-74. Art ser., Nutrition Policy Discussion Paper 5. Administrative Commit- de Sole, B., Y. Belay, anid B. Zegeye 1987. "Vitamin A Deficiencv in Southem tee on Coordination, Subcommittee on Nutrition, United Nations, Ethiopia." Americatn Journal of Clinical Nutrition, 45:780-84. New York. Solon, F., R. Florentino, M. Latham, T. Fernandez, [. Panopio, and R. Guiirriec. Underwood, B. A. 1983. Nutrition Intervention Strategies in National Dev'elop- 1983. "Pilot MSG Fortification for the Control of Vitamin A Deficielicy in ment. New York: Academic Press. the Philippines." In V. Tanphaichitr, W. Dahlan, V. Supharkarn, and A. - 1986. "The Safe Use of Vitamin A bv Women during Reproductive Valyasevi, eds., Human Nutrition: Better Nutritiot, Better Life Proceedings of Years." Pamphlet from IVACG Secretariat, Washington, D.C. the 4thAsian Congress of Nutrition. Bangkok. Bangkok: Aksornsmai Press. . 1989. Paper presented at Intemational Vitamin A Consultative Sommer, A. 1982. Nutritional Blindness: Xerophthalmia and Keratomalacia. New Group (IVACtG) Annual Meeting, Kathmandu, Nepal. York: Oxford University Press. UNICEF (United Nations Children's Fund). 1988. "Support to Vitamin A Sommer, A., E. Djunaedi, A. A. Loeden, 1. Tarwotjo, K. P. West, R. Tilden, Deficiency Control Programs." New York. and L. Mele. 1986. "Inspact of Vitamin A Supplementation on Childhood United Nations. 1985. "Prevention and Control of Vitamin A Deficiency, Mortality." Lan1cet 1: 1169-73. Xerophthalmia, and Nutritional Blindness." Proposal for a ten-year program Sommer, A., J. Katz, and 1. Tarwotjo. 1984. "Increased Risk of Respiratory of support to countries. Administrative Committee on Coordination, Sub- Disease and Diarrhea in Children with Preexisting Mild Vitamin A Defi- committee on Nutrition, New York. c iency. " Amencan]Journal of Clinical Nutrition 40:1090-95. .___- 1987. "A Global Strategy for the Prevention and Control of lodine- Soimmer Alfred, Ignatius Tarwotjo, Gusti Hussatni, and Djoko Susanto. 1983. Deficiency Disorders." Proposal for a ten-year program to support countries. "Increased Mortality in Children with Mild Vitatiin A Deficiency." Lancet Administrative Committee on Coordination, Subcommittee on Nutrition, (September 10):585-88. New York. Squatrito, S., R. Vigneri. F. Runello, A. M. Ermans, R. D. Polley, and S. H. . 1990. "Controlling Iron Deficiency-Report of Joint ACC/SCN Ingbar. 1986. "Prevention and Treatment of Endemic lodine-Deficiency Workshop on Iron Deficiency Control, June 6-8, 1990. Dublin, Ireland." Goiter by lodination of a Municipal Water Supply." Journal of Clinical Administrative Committee on Coordination, Subcommittee on Nutrition, EndocrinoIogy and Metabolism 63:368-75. New York. Micronutrient Deficiency Disorders 451 Valyasevi, Arec. 1988. "Delivery System for Iron Supplementationi in Preg- . 1982. "Control of Vitamin A Deficiency and Xerophthialmia." Tech- nant Women-Thailand Experience." Paper presented at the INACG (In- nical Report 672. WHO Nutrition Unit, Geneva. ternational Nutritional Anemia Consultative Group) Workshop - 1988. "Global Status for Vitamin A Deficiency." Geneva. November 14-16. Geneva. - . 1989. "Global Status of Iodine Deficiency Disorders." 1989. Report VLteri, F. E., E. Alvarez, J, Bulux, H. Gonzalez, 0. Pineda, L. M1ejia, R. Batres, for World Health Assemblxy. WHO Nuttition Unit, Geneva. Typescript. and B. Torun. 1981. "Iron Fortification in Developing Countries." Progress . 1990. "Global Status of IDD." Report for World Health Assembly. WHO in Clinical and Biological Research 77:345-54. Nutrition Unit, Geneva. Typescript. Wallingford, 1. C., and B. A. Underwood. 1986. "Vitamin A Deficiency in WHO (World Health Organization)/EP (Expanded Programme on Immuniza- Pregnancy, Lactation, and the Nursing Child." In J. C. Bauemfeind, ed., tion). 1987. Potential Conmbuton of the Expanded Prograni7m on Immuniza- Vitamin A Deficiency and Its Control New York: Academic Press. non to the Control of Vitamin iA Deficiency and lodine Deficiency Disorders. Walter, T. 1989. "Infancy: Mental and Motor Development."AinencaniJournal Paper presented at the EPI Global Advisory Group Meeting, November of Clinical Numtion 50:65 5-61. 9-13, 1987, Washington, D.C Walter, T., S. Arredondo, M. Arevalo, and A. Stekel. 1986. "Effect of Iron . 1988a. Global Situation of Vitamin A Deficiency. Geneva: WHO Nutri- Therapy on Phagocytosis and Bactericidal Activity in Neutrophils of Iron- tion Unit. Deficient Infants." AmericanJournal of Clinical Numnion 44:877-82. _ . 1988b. Programnmes for the Control ofVitamin A Deficiency The Role of West, K. P., Jr., and A. Sommer. 1984. Periodic, Large, Oral Doses of Vitarmn the EPI in New Initianves for the 1 990s. Geneva: WHO Nutrition Unit. A for the Prevention of Vitamin A Deficiency and Xerophthailmia A Summary WHO (World Health Organizat ion)/UNICEF (United Nations Chi dren 's Fund). of Expenences. Report of the International Vitamin A Consultative Group. 1987. "Joint Statement on Vitamin A for Measles." International Nursing The Nutrition Foundation, Washington, D.C. Review 35(1):21. 1987. Delivery of Oral Doses of Vitamin A to Prevent Vitamin A Working Group on Fortification of Salt with Iron. 1982. "Use Of Common Deficiency and Nuritinonal Blindness. State-of-the-Art Series, Nutrition Pol- Salt Fortified with Iron in the Control and Prevention of Aneniia-A icy Discussion Paper 2. Administrative Committee on Coordination, Sub- Collaborative Study." Anerican Journal of CLnical Numtion 35:1442-51. committee on Nutrition, United Nations, New York. Yan-You, W., and Y. Shu-Hua. 1985. "Improvement in Hearing among Wlttpenn, J., and A. Somnser. 1986. "Clinical Aspects of Vitamin A Defi- Otherwise Nornial Schoolchildren in Iodine-Deficient Areas of Guizhou, c iency." In J. C. Bauemfeind, ed ., Vitamin A Deficiency and Its ControL New China, Following Use of lodised Salt." Lancet 2:518-19. York: Academic Press. Yepez, R_. A. Calle, P. Galan, E. Estevez, M. Davila, R. Estrella, A. Masse- WHO (World Health Organization). 1979. Training Unlization of Auxiliary Raimbault, and S. Hercberg. 1987. "Iron Status in Ecuadorian Pregnant Personnel for Rural Health Teams in Developzing Countries. Technical Report Women Living at 2800 m. Altitude: Relationship with Infant Iron Starus." 633. Geneva. InternationalJounial of Vitamin and Numrion Research 57:327-32. PART FOUR Emerging Problems Sexwlly Transrrmted Disases Camcers Diabetes Cardivar Disease Injury Cataract Oral Health Schizophrerna and Manic-Depressive Illness 20 HIV Infection and Sexually Transmitted Diseases Mead Over and Peter Piot The health programs of developing countries have not tradi- were identified only during the last decades, partly as a result tionally accorded a high priority to the prevention and control of improved laboratory techniques. of diseases which are predominantly transmitted by sexual intercourse. With the realization that sex is the primary mode Distinctive Features of STD Epidemiotogy of transmission for the human immunodeficiency virus (HIV), however, intemational donors are helping national health A list of common sexually transmitted agents and the diseases ministries of developing countries allocate large human and they cause are presented in table 20- 1. In this chapter, we focus financial resources to the fight against at least one sexually on HIV infection and on selected CSTDs, including gonorrhea, transmitted disease (STD). In many cases these programs for the genital chlamydial infections, syphilis, and chancroid. The prevention of acquired immunodeficiency syndrome (AIDs) are contribution of the sexually transmitted human papilloma large enough to rival preexisting programs to prevent other viruses to the causation ofcervical cancer will not be discussed, diseases, like malaria and measles, which currently kill more although both in Africa and in Latin America the incidence people in most of these countries. By supporting this proposed of cervical cancer is among the highest in the world, and sexual expenditure pattem, the international donors and national activity is the main risk factor for this common neoplasm health ministries have implicitly raised the priority attached (Reeves, Brinton, and Brenes 1985; Rosenberg, Schultz, and to the prevention of STDs far above the position formerly Burton 1986; and Reeves, Rawls, and Brinton 1989). occupied by this class of diseases. In addition, because of a An in-depth discussion of the biology and epidemiology of growing awareness that at least one important cancer (cervical even the main STDs is beyond the scope of this chapter but can cancer) and a significant proportion of maternal morbidity and be found in several textbooks and monographs (Osoba 1987; mortality result from STDs, there is a renewed interest in STD Arya, Osoba, and Bennett 1988; Holmes and others 1989). control in the public health community. Appendix 20A includes a summary of medical information on Our objective in this chapter is to examine the case for the individual diseases considered here. assigning a high priority to the prevention (primary and sec- The epidemiology of STDs is distinctive because of common ondary) of the spread of STDs, including AIDS and its causative behavioral and biological features. First, STDs typically have agent, HIV. Although it would be possible and in some ways long latent or incubation periods before symptoms become more convenient to separate the discussion of AIDS from that apparent, during which transmission can occur. Second, the of other STDS, a central theme of this chapter is the examina- genetic structure of most STD agents varies so much that re- tion of the epidemiological, medical, and economic arguments searchers have been unable to design a vaccine against them. for integrating AIDS prevention eftorts with efforts to combat Third, STDs are primarily spread by a class of behavior which is other STDs. Furthermore, the transmission dynamics of all STDs inherently resistant to change, because it is highly motivated, have strong similarities, which benefit from a common analyt- often clandestine, and varies so much both within and between ical examination. These considerations lead us to address AIDS social and ethnic groups. and "classic STDs," or CSTDs, in the same chapter but often to Acommonbiologicalfeatureofmanyofthemicroorganisms separate the discussions into different sections. causing STDs is their unique and often exclusive adaptation to humans, the main mode of transmission being genital mucosal The Epidemiology of STDs contact-for example, sex in most instances.Whether a micro- bial agent is mainly sexually transmitted in a given population, Both CSTDS and Hiv are mainly transmitted through sexual however, depends not only on its biology but also on behav- intercourse, although in many cases they may also be transmit- ioral and environmental conditions. Thus, in many developing ted vertically from mother to child or by blood contact. More societies some infections are mainly acquired in childhood than fifty CSTDs have now been recognized, many of which because of low hygienic standards or poor living conditions, 455 456 Mead Over and Peter Piot Table 20-1. Important Sexually Transnitted Agents and Diseases Agents _ ___ Disease or syvndrome Bacteria NeisseriagonorThoeae Urethritis, epididymitis, proctitis, bartholinitis, cervicitis, endometritis, salpingitis and related sequelae (infertility, ectopic pregnancy), perihepatitis; complications of pregnancy (e.g., chorioamnionitis, premature rupture of membranes, premature delivery, postpartum endometritis); conjunctivitis; disseminated gonococcal infection (txJI) Chlamydia trachomnats Same as N. gonorrhoeae, except for DGI; also trachoma, lymphogranuloma venereum, Reiter's syndrome, infant pneumonia Treponema pallidium Syphilis Haemophilus ducreyi Chancroid Mycoplasna hominis Postpartum fever, salpingitis Ureaplasma urealy ticum Urethritis; low birth weight,", chorioamnionitis' Gardnerella vaginalis and others Bacterial vaginosis Calymmatobacteriuum granulomatis Donovanosis Group B 0-hemolytic streptococcusa Neonatal sepsis, neonatal meningitis Viruses Herpes simplex virus Primary and recurrent genital herpes; aseptic meningitis; neonatal herpes and associated mortality or neurological sequelae; spontaneous abortion, premature delivery Hepatitis B virus Acute, chronic, and fulminant hepatitis B, with associated immune complex phenomena and sequelae, including cirrhosis and hepatocellular carcinoma Cytomegalovirus Congenital infection; gross birth defects and infant mortality, cognitive impairment (e.g., mental retardation, sensorineural deafness); heterophile-negative infectious mononucleosis; protean manifestations in the immunosuppressed host Human papilloma virus Condyloma acuminata, laryngeal papilloma in infants; squamous epithelial neoplasias of the cervix, anus, vagina, vulva, and penis Molluscum contagiosum virus Genital molluscum contagiosum Human immunodeficiency virus A[mS and related conditions HTLV- I (Human T-lymphotropic virus) T-cell leukemia, lymphoma; tropical spastic paraparesis Protozoon: Trichomonas vaginalis Vaginitis; urethritis,a balanitisa Fungus: Candida albicans Vulvovaginitis, balanitis, balanoposthitis Ectoparasites Phthirius pubis Pubic lice infestation Sarcoptes scabiei Scabies a. Causative relationship uncertain. Source: Based on compilation of the literature. whereas in industrial countries the same infections are mainly disruption of traditional social structures, increased mobility sexually transmitted among adults (hepatitis B, cytomegalovi- for economic or political reasons, poor medical facilities, a rus infection). In general, infections become more often sexu- large proportion of the population composed of teenagers and ally transmitted with an increasing standard of living, because young adults (who have the highest incidence of STD), and high opportunities for person-to-person transmission are decreasing unemployment rates are all contributing to the high incidence during childhood. of STDs and their complications and sequelae (Piot and Holmes Risk factors for STDs are directly related to patterns of sexual 1989). behavior. They include a large number of sex partners, a history of STDS, urban residence, being single, and being young (Piot STDs Are Communicable and Meheus 1983). Prostitutes are named by up to 80 percent of male patients as the source of infection in some, but not all, Preventing or curing one case of an STD often prevents many developing countries as compared with less than 20 percent in other cases. This obvious consequence of the fact that STDs are Europe and North America (Rajan 1978; D'Costa and others communicable introduces a complication into the analysis of 1985) and are probably an important reservoir of STDs in many the priority to assign to their prevention. It is not sufficient to parts of the world. Still, significant differences in sexual behav- weigh against the cost of preventing a case only the benefits of ior patterns exist within continents and even within countries. preventing that single case; the so-called "dynamic benefits" The highest rates of STDs are found in urban men and women that accrue to others than the immediately affected individual in their sexually most active years, that is, between the ages of must also be included. I fifteen and thirty-five. On the average, women become in- A key epidemiological concept in this connection is that of fected at a lower age than men. Increasing urbanization with the "reproduction rate."2 Defined as the number of new (or HIV Infection and Sexuallyv Trarnmitted Diseases 457 secondary) cases infected by an average case, the reproduction of any communicable disease were described earlier. The rela- rate can he used to Multiply the number of prevented primary tionship R = QaD can be applied directly to characterize the cases in order to obtain a crude measure of the total beneficial course of an STD epidemic, where a is interprered as the rate of effect of the prevention effort.3 Clearly the inclusion of these acquisition of new sexual partners and that rate is the same for extra cases among the benefits of an STD prevention program all members of the population. If each individual in the popu- will increase the measured cost-effectiveness of preventive lation, however, has a different rate of sexual partner change, efforts. It can be argued that the failure of decisionmakers to a, then it is not sufficient to use the average of these rates in consider sufficiently the benefits of the prevention of second- order to estimate R. Instead, Anderson and May (1988) have ary and subsequetnt cases has contributed to the undervalua- shown that the heterogeneity in the sexual behavior, as mea- tion of the priority to assign to CSTL) control in developing sured by the variance of the ai, adds substantially to the countries. In contrast, it is clear that the present attention reproductive rate, and thus to the likely future rate of growth allocated to AILDS prevention is almost entirely due to fear of a of the epidemic.4 In addition, it is increasingly clear that there high reproduction rare. is considerable heterogeneity over time in the infectivity of In addition to its importance in estimating the benefits of individuals with HiV infection, adding to the complexity of the preventing a communicable disease, the reproduction rate dynamics of viral spread in populations. plays a key role in the analysis of the future course of an epidemic. To understand this, consider that a communicable Current Levels and Trends in the Developing World disease characterized by a reproductiotn rate less than unity is headed for extinction as each individual case fails to replace Reflecting the low level of priority assigned to CSTDs in most itself entirely in the population. Contrarily, a disease whose countries, data on the levels of CSTD infection in many popu- reproduction rate is greater than unity can be predicted to lations in the developing world are poor and largely confined explode geometrically. For any communicable disease in the to selected groups (and samples of convenience). Therefore, early stages of an epidemic, the value of the reproductive rate the figures presented here should be considered as approximate (R) can be simply calculated as the product of three parame- and not necessarily representative for the general population. ters: the probability of infection on each contact (Q), the From a public health perspective, however, the overall preva- number ofcontacts per time period between an infected person lence or incidence rates in the general population may not be and a susceptible one (a), and the duration of infectivity of the as critical as the size of the segment of the population that is infected person (D). Note that the first of these three values is at risk and the rate of infection in each risk group. For example, primarily determined by characteristics of the disease, whereas even if the nationwide prevalence ofitiiv infection in a country the second is primarily behavioral. The third parameter, the is low, there may still be a significant problem in the cities.5 duration of infectivity, is typically affected by both the biology of the particular disease and the effectiveness of public health CLASSIC SEXUALLY TRANSMITTED DISEASES. Two parameters strategies for either curing or isolating the infective individual. important in estimating the burden ofcSTDs are the prevalence Later in this chapter we use estimated reproduction rates or of infection and the rate of complications and sequelae. their analogues both to characterize epidemic patterns and to The degree of health-seeking behavior and the quality of estimate benefits of case prevention. health services and STD control programs directly control the latter and, by reducing transmission, indirectly control the The Dynamics of Sexual Transmission former. In table 20-2, we present selected data on the prevalence of A turning point in the public health perspective toward STDs gonococcal and genital chlamydial infection and serologic occurred with the realization in the late 1970s that a key evidence of present or past syphilis among samples of adult distinction between STDs and other epidemnics is the import- women in different parts of the world in the 1970s and 1980s. ance of the heterogeneity of sexual behavior in understanding With the exception of the population-based surveys of women the disease process. The simplest useful characterization of in Senegal and Uganda, the samples are drawn from the heterogeneous sexual behavior, introduced to the analysis of population of pregnant women. Although pregnant women gonorrhea epidemics in 1978 by Yorke, Hethcote, and Nold imperfectly represent all adults, these data are as representative (1978), is to posit two separate groups, a "core group" ofhighly of the general population as one can get in the literature. sexually active individuals and a "noncore group," which is Because infertile women are excluded from these series, there much less so. The characterization becomes more realistic as may be a bias to lower prevalence rates. Still, because the the number of groups is increased or as the behavior of indi- clinical manifestations of such CSTDs as gonococcal and viduals is allowed to vary within a group. However character- chlamydial infection are less specific in women than in men, ized, the heterogeneity of sexual behavior plays an extremely and their diagnoses therefore technically more complex, the influential role in determining both the course of an STD prevalence rates of infection, but not necessarily the incidence epidemic and the choice of control strategy. rates, are usually higher in women than in men. Partly for this The definition of the reproduction rate and its importance reason, morbidity and mortality rates from all CSTDs except as a simple indicator of the likely future course of an epidemic syphilis are also much higher in women. 458 Mead Over and Peter Piot Table 20-2. Prevalence of Gonococcal and Genital Chlamydial Infection and Positive Serologic Test for Syphilis among Urban Pregnant Wornen Neisseria Chiamydia vDRL/RPR Country gonorThoeae trachomatis TPH.A/FTA-Abs ' Source Africa Cameroon 15 na.. i0 Nasah and others 1980; Kaptue and others 1990 Ethiopia 9 n.a. n.a. Penne and others 1980 Gabon 5.5 8.3 n.a. Yvert and others 1984; Leclerc and others 1988 The Gambia 6.7 6.9 17.5, 7.2 Mabey and others 1984 Ghana 3.4 7.7 n.a. Bentsi and others 1985 Kenya 7 8.9 3.8 Laga and others 1986b; Temmerman and others 1990 Mozambique n.a. n.a. 6.3 Liljestrand and others 1985 Nigeria 5.2 6.5 2.1 Okpere, Obaseiki-Ebor, and Oyaide 1987; Aladesanmi, Mumtaz and Mabey 1989; Fakeya, Onile, and Odugbemi 1986 Rwanda 5 16 4 Senvonyi 1987; Dabis and others 1989 Senegal 1.5 7 7.5 de Schampheleire and others 1990; Ndoye 1991b Somalia n.a. n.a. 3 Jama and others 1987 South Africa 11.7 13 n.a. Welgemoed and others 1986; Ballard, Fehler, and Piot 1986 Swaziland 3 n.a. 13.1 Meheus and others 1980; Guiness and others 1988 Tanzania 6 n.a. 16.4 COoper-Poole 1986 Uganda 18/2 n.a. n.a. Arya, Taber, and Nsanze 1980 Zaire 1.8 6.3 0.9 Luyeve and others 1990 Zambia 11.2 n.a. 14.3, 12.5 Ratnam and orhers 1982 Americas Chile 2 n.a. n.a. Donos) and others 1984 Jamaica 11 n.a. n.a. George 1974 Unired States I/9d n.a. n.a. Mtimavalye 1987 Asia India 10 n.a. n.a. Jha and others 1978 Malaysia 0.5 n.a. n.a. Mtimavalye 1987 Thailand 12 n.a. n.a. Mtimavalye 1987 n.a. Not applicable. a. Test for Treponema palidum, the etiologic agent for syphilis, was used for both diseases. Acronynms-VDRL: Venereal Disease Research Laboratoary/RPR: Rapid Plasma Reagin/TPHA: Teponemiia Pallidtim Hemagglutinarion Assav/FtA: Fluorescrent Teponema Antibody. b. Women in general population. c. Low fertility district/high fertility districr. d. White/black. Sources: See last colunmn. The morbidity of CSTDs occurs mostly between the ages of limited in severity and duration (one to five weeks), delay fifteen and forty-five years-not only the sexually most active before seeking treatment may be as long as two years (Popula- period in life but also the most economically and demograph- tion Information Program 1983). Mild to severe urethral stric- ically productive age. Geographic differences in prevalence are ture may occur after urethritis in up to 3 percent of men, with obvious from table 20-2, but extrapolations on the scale of a the time between onset of urethritis and acute urinary reten- continent cannot be made. Prevalence data on genital ulcer- tion rangingbetweenafewdays and several years(Bewes 1973; ations in the general population are not available from the Osegbe and Amaku 1981). Treatment is difficult and time developing world, but rates of 4 to 8 percent have been found consuming, and such cases constitute up to 80 percent of the in female prostitutes in Central and East Africa (D'Costa and practice of urologists in some parts of Africa (Bewes 1973). others 1985; Laga and others 1989). Genital ulcers do not Acute and chronic epididymitis may occur in I to 10 percent directly lead to mortality. Without treatment, the very painful of cases of urethritis and may be associated with long-term chancroid lesions take two to three months to heal. The time morbidity and infertility. The proportion of cases of infertility between onset of disease and the individual's presentation at in the male due to CSTDs has not been well defined, but it is a medical facility is often two to four weeks in Sub-Saharan estimated at 20 to 40 percent in the developing world (Popu- Africa (Plummer and others 1983). lation Information Program 1983). In men, the incidence of both gonococcal and chlamydial In women, uncomplicated cervical infection with Neisseria infections may be very high (up to 20 percent annually be- gonorrhoeae orChlamydia -rachomatis is usually associated with tween the ages of fifteen and forty-five years in high-risk nonspecific genital signs and symptoms, which interfere only groups). Because the associated morbidity (urethritis) is mostly minimally with daily activities. Complicated disease and its HIV Infection and Sexuallv Transmitted Diseases 459 sequelae are a significant cause of morbidity, however, and an morbidity and mortality due to the different stages of syphilis important proportion of reproductive mortality, even in the have not been documented for adults in the developing world. United States (Grimes 1986). The consequences of syphilis for pregnancy have been better Studies in Sweden have shown that 8 to 10 percent of documented and are impressive. Approximately 10 to 12 per- women with gonococcal or chlamydial infection develop pel- cent of infants born to mothers with a positive syphilis serology vic inflammatory disease (PiD [Westr6m 19801). The annual will die during the neonatal period if untreated, yielding a incidence of rID among urban women in Sub-Saharan Africa mortality as high as 1 to 3 percent among under fours (in can be estimated at I to 3 percent between the ages of fifteen populations with a prevalence of a positive test for syphilis of and forty-five, with incidence rates of 0.4 to 1.2 percent, and more than 30 percent), in addition to a 20 to 25 percent 0.4 to 1.5 percent for gonococcal and chlamydial PID, respec- stillbirth rate in the same group (Ratnam and others 1982; cDC tively (assuming that 20 to 40 percent of cases of PrD are due to 1986; Hira and Hira, 1987). Congenital syphilis is multiorga- N. gonorrhoeae and 20 to 50 percent to C. trachomatis). Half of nic and may result in severe physical and mental handicaps. these cases occur during the puerperal period. The annual Overall it occurs in 25 to 75 percent of exposed infants. mortality directly attributable to PID in women of fifteen to Trends for CSTDs and their complications between the early forty-five years would then be 0.1 to 0.5 per 1,000 (assuming a 1970s and the mid-1980s in the developing world are un- I percent case-fatality rate). These figures are probably much known. In Swaziland the prevalence rate of reactive syphilis lower in rural areas and in other parts of the developing world, serology among pregnant women remained at a level of ap- but data are lacking. proximately 30 percent between 1978 and 1987 (Ursi and The annual incidence of bilateral tubal occlusion (leading others 1981; Guinness and others 1988), and in Rwanda the to infertility) is estimated at 0.3 percent to 1.5 percent in urban annual incidence of gonorrhea continuously increased among women in Sub-Saharan Africa, with gonococci and chlamydia military recruits between 1981 and 1984 (Piot and Carael each being responsible for 20 to 40 percent of cases (assuming 1988; see figure 20-1). Both sets of data suggest that the a 15 to 40 percent risk of tubal occlusion after one episode of incidence of CSTDs in these countries remained at the same PiD [Westrd)m 1975; Westrom and others 19791). Whereas high levels during these periods. In contrast, the impressions bilateral tubal occlusion is found in 50 percent of African gained from clinic-based data are that CSTD incidence has women who are infertile, this is the case in only 14 to 20 increased recently, particularly in urban populations. percentofsuch women in Asia, Latin America, and the Middle East (Cates, Farley, and Rowe 1985). HIV INFECTION. Because .ADS is a new disease, data on mor- The annual incidence of ectopic pregnancy in urban Africa bidity and mortality in the developing world are still fairly resulting from PiD is estimated at 0.01 to 0.04 percent, with an limited. Still, a growing set of data on the prevalence and annual mortality rate of 0.001 to 0.005 percent for women incidence of Hiv infection, as well as on the incidence of AIDS between the ages of fifteen and forty-five (Urquhart 1979). cases, is becoming available. The former data are probably the Finally, maternal mortality due to gonococcal and chlamyd- more important ones, because they indicate the number of ial infection (postpartum infectious complications) may be as high as 0.04 to 0.2 percent annually in Sub-Saharan Africa Figure 20-1. Gonorrhea Prevalence among (with a maternal mortality rate of 0.5 to I percent and a 10 to Rwandan Military Recruits, by Months of Service 20 percent incidence of postpartum infections). Even in the Cumulative percentage infected United States, deaths due to STDs account for 20 percent of maternal mortality (Grimes 1986). In general, the overall 80 mortality from STDs is not well defined. It is often a hidden 70 1982 mortality and morbidity because of a long latency period 1984 between the acute infection and the complication or sequela 60 1981 leading to death. In addition, the association between CSTDs j 1983 , ' and some of these complications is not well understood 1983 (Brunham, Holmes, and Embree 1989). 40 .- , In neonates, conjunctivitis and respiratory disease are the main causes of morbidity due to N. gonorrhoeae and C. tracho- mratis infection in the mother. Their incidence depends on the 20 prevalence of these infections in pregnant women. The occur- rence of gonococcal ophthalmia neonatorum in neonates de- 10 pends on whether effective eye prophylaxis is implemented at I l birth. Disablement from gonococcal neonatal infection is due 0 12 24 36 48 60 to keratitis and blindness, and disablement from chlamydial Months of military service infection results mainly from chronic respiratory disease. Whereas prevalence rates of positive serologic tests for syph- ilis are available for numerous populations (see table 20-2), the Source: Piot and Carail. 460 Mecld Over and Peter Piot people who will develop AIDS in the next decades. For the sake increasino in several countries in the region. Finally, in Asia, of simplicity, we equate morbidity and mortality rates for AIDS, where the virus has been introduced more recently, HIV iS because no widespread effective treatment is available in the spread mainly among people with multiple sexual contacts developing world. and among intravenous (iv) drug users. The latter group There is considerable variation in the prevalence and appears to be particularly vulnerable, as shown by the rapid incidence of HIV infection throughout the world, on a spread of HIV infection among drugs users in Thailand and given continent and even within countries, making extrap- Burma (Phanuphak and others 1989). The features of these olations speculative. For instance, the male-to-female ratio of three epidemiologic patterns in the world are summarized AIDS cases is less than one in Zaire and Uganda, but two to in table 20-3. four inbothC6ted'1voire and Senegal (Piot and others 1990). The rate of transmission of HIV varies with the mode of These variations are often not well understood, and the acquisition and seems influenced by a number of parameters. elucidation of their causes may provide important clues for Thus, it is thought that the basic risk of acquiring HIV infection preventive programs. The interaction among demographic, through vaginal intercourse is 0.1 to 0.5 percent and through behavioral, and political factors probably determine how and receptive anal intercourse is probably ten times higher (Piot where HIV spreads. and others 1987; Johnson and Laga 1988). In the presence of The three modes of transmission of HIV include sexual other STDs, however, particularly those associated with genital intercourse, parenteral exposure to blood and blood prod- ulceration, both the susceptibility to HIV infection and the ucts, and vertical transmission from mother to child. For infectiousness of an HIV-infected individual are increased sev- Africa and the Caribbean, it is estimated that more than 80 eral fold (five to ten times [Piot and others 1988; Pepin and percent of those who are infected with liiv acquired their others 19891). Inaddition, the infectivityofanHlvseropositive infection heterosexually. In Latin America, homosexual person seems to increase just before or when he or she develops men account for the majority of cases (15 to 80 percent) of clinical disease (Johnson and Laga 1988). This may at least HIV infection, but heterosexual transmission appears to be partly explain why HIV has spread more quickly in the hetero- Table 20-3. Three Epidemiologic Patterns of HIV Infection Characteristic Pattern I Pattern 11 Pattern III Major group affected Homosexual and bisexual men Heterosexuals Persons with multiple sex partners and IDU Period when introduced or Mid-1970s or early 1980s Early to late 1970s Early to mid-1980s began to spread extensively Sexual transmission Predominantly homosexual. Over Predominantly heterosexual. Up Both homosexual and herero- 50 percent of homosexual men in to 25 percent of 20- to 40-year sexual transmission just now some urban areas infected. age group in some urban areas being documented. Very low Limited heterosexual infected and up to 90 percent of prevalence of HIV infection even transmission occurring, but female prostitutes. Homosexual in persons with multiple partners, expected to increase. transmission not a major factor. such as prostitutes (except in some areas of Southeast Asia and India). Parenteral transmission After homosexual transmision, Transfusion of HIV-infected blood Not a significant problem at intravenous drug abuse accounts is major public health problem. present in most countries, but for the next largest proportion of Nonsterile needles and syringes growmng problem in IDLU in HIV infections, even in Europe. account for undetermined Southeast Asia. Transmission from contaminated proportion of HIV infections. blood or blood products not a con- tinuing problem, but existing cohort of persons infected by this route before 1985. Perinatal transmission Documented primarily among Significant problem in rhose areas Currently not a problem. female IDU and women from where 5 to 15 percent of women HIV- I endemic areas are HIIV- I antibody positive. Distribution Western Europe, North America; Africa, Caribbean, some areas in Asia, the Pacific Region (except some in South Aiierica, South America. Australia and New Zealand), Australia, New Zealand. Middle East, Eastern Europe, some rural areas of South America. IDU Injecting drug users. SOurce: Piot and others 1988. HIV Infecion and Sexually Transmitted Diseases 461 Table 20-4. Estimated HIV I Sero prevalence by Residence and High-Risk Category, Developing Countries, 1990 (percent) Residence Residence Country Urban Rural High Riska Country Urban Rural High Risk' Africa Africa (continued) Angola 1.3 b - 14-2 Sudan 0.0 -160 Benin 0.1 6.7 4.5 Swaziland 0.0 - Botswana 0.8c 0 Ic 1.2c Tanzania 8.9 5.4 38.7 Burkina Faso 17d 3.1 16.9d Togo - - Burundi 17.5 - 18.5 Tunisia 0.0 - 1.9 Cameroon 1.1 0.4 8.6 Uganda 24.3 12.3 86.0 Cape Verde 0.0 - 0.0 Zaire 6.0d 3 d 37.8 Central African Republic 7.4 3.7 20.6 Zambia 24.5 13.0 54.0 Chad 0.0 0.0 -d Zimbabwe 3.2' 1.4 Congo 3.9 1.0b 34.3 Cote d'lvoire 8.5 3.3 d 23.8 Latin America Djibouti 0.3 0.0 2.7 Antigua and Barbuda - - 1.7 Egypt 0.0 - 0.2 Argentina 0.3 0.1 5.8 Equatorial Guinea 0.3 0.3 - Bahamas 0.5 - Ethiopia 2.0 0.0 18.2 Barbados 0.1 - Gabon 1.8 0.8 -b Bolivia 0.0 - 0.0 The Gambia 0.1 - 1.7 Brazil 0.3 0.0 3.0 Ghana 2.2 - 25.2 Colombia 0.1 - 14.6 Guinea 06b 0.2 d Costa Rica 0.0 - 0.0 Guinea Bissau 0.1 0.0 0.0 Cuba 0.0 - 0.0 Kenya 7.8 1.0 59.2 Dominican Republic 1.6 - 2.6 Lesotho 0.1 - -d Ecuador 0.0 - 0.0 Liberia 0.0 0.0 0.0 Guadalupe 0.2 Libya 0.0 - - Guyana - - 0.0 Madagascar 0.0 - 0.0 Haiti 4.9 3.0 41.9 Malawi 23.3 - 55.9 Jamaica 0.3 - 14.6 Mali 0.4 - 23.0 Martinique 0.5 - 38.9 Mauritania 0.06 - 0.0 Mexico 0.7 - 2.2 Mauritius 0.0 - - d Panama 0.0 - 0.0 Morocco 0.0 - 7.1 Peru 0.1 - 0.3 Mozambique 1.1 0.8 2.6 Trinidad and Tobago 0.9 - 13.0 Namibia 2.5 - - Venezuela 0.0 0.0 Niger - - 5.8 Nigeria 0.5 0.0 17cd Asia/Oceania Rwanda 30.3b 1.7i 79.8h Burma - - 1.9 Senegal 0 1, 0.0 2.3 India 0.1 - 18.1 Sierra Leone 3.6 - 2.7 Papua New Guinea 0.0 0.0 0.7 Somalia 0.0 - 0.4 Philippines 0.0 - 0.1 South Africa 0.1 - 3.2 Thailand 0.0 - 0.2 - Dara not availahle. a. Prostitutes and clients, sTo patients, or other persons with known risk factors. b. Infection with only HIV-1 and dual infection (HI.-l and HIv-2). c. Data prior to 1986. d. Data not necessarily reliable because small sample size (less than I L0). Source: U.S. Bureau of the Census 1991. sexual population in Africa than in North America and Europe; single, having a history of STDs, and having sex with prostitutes genital ulcer disease seems to be more common in several or being a prostitute. Urban populations usually have much African populations, and HIV has probably been present for a higher infection rates than rural populations (see table 20-4), longer time among heterosexuals in Africa, resulting overall in though this may change when the epidemic spreads. Lack of a higher infectiousness of the HIV-infected population (Piot circumcision has been claimed to be a risk factor for HIV and others 1988). infection in men, but this remains controversial (Van de Perre Risk factors for HiV infection are generally those of other and others 1987; Simonsen and others 1988; Bongaarts and STDS, such as having a high number of sex partners, being others 1989). 462 Mead Over and Peter Piot The efficiencies of transmission by a blood transfusion or by Finally, the effect of HIV infection on the natural history of intravenous needle sharing are probably close to 100 percent. childhood diseases, such as measles and malaria, is incom- Because of a high rate of infection among young adults in pletely understood. Preliminary data suggest that iniv infection several populations, and because of the incomplete availability in children does not influence the response to childhood of iiiv antibody tests, HIV infection through blood transfusion immunizations, though, again, it is not clear what the effect of makes up a larger proportion of cases of AIDS in pattern II iiv infection on protective immunity will be when the im- countries than in Europe or North America. This is particu- mune status of these children deteriorates (Mvula, Ryder, and larly the case in children, who are the main consumers of blood Manzila 1989). transfusions in large areas of the developing world, together In pattern 11 countries, on which we focus (mainly Sub- with pregnant or parturient women. This is due to a high Saharan Africa), the overall female-to-male ratio of people incidence of severe anemia caused by malaria or obstetrical with HIV infection usually approaches one to one, though problems (Greenberg and others 1988; Ryder and Mhalu regional variations have been reported. A marked variation in 1988). In addition, indications for blood transfusion may not sex ratio, however, has been found by age group in some always be rational, and proper blood banks, involving volun- countries (for example, Zaire), where HIV-seropositive women tary, low-risk donors, are rarely functioning. Finally, there are largely outnumbered men between fifteen and thirty years of now several reports of probable H-V transmission to blood age, whereas there were more infected men than women above donors through contaminated plasmapheresis equipment that age (Ryder and Piot 1988). Several surveys in Africa have (Laga and Piot 1988). also found that the HIV seroprevalence rate is highest in indi- Whereas it is often assumed that intravenous drug use is a viduals between twenty and forty years of age and that peak prerogative of the rich West, the AIDS epidemic has dramati- seroprevalence rates occur at a younger age in women than in cally shown that intravenous drug use is spreading rapidly in men (Ryder and Piot 1988). This age pattern for both sexes many developing countries. Thus, more than 40 percent of combined is illustrated in figure 20-2; in the figure the HIV- I Thai 1v drug users were infected by early 1989, as compared seroprevalence rates are shown by age group as found in a with I percent in 1987 (Phanuphak and others 1989). It is national serum survey in Rwanda in 1986, in which a random anticipated that similar outbreaks of HIV infection among iv cluster sampling method was used. The Rwandan data also drug users will occur in other developing countries. The role indicate a higher infection rate for urban women than urban of contaminated medical injections in the spread ofHIV infec- men (Rwandan Seroprevalence Study Group 1989). tion is probably marginal (Piot and others 1988; Van de Perre In table 20-4 we summarize seroprevalence data in various and others 1987, Berkley 1991). groups of adults in different developing countries. The highest The rate of vertical transmission from mother to child is rates are consistently found in Central Africa, but epidemic of the order of 25 to 50 percent in Africa, with women who foci start appearing in other parts of the continent as well, for are more advanced clinically being more infectious for their offspring (Lallemant and others 1989; Ryder and others Figure 20-2. Seroprevalence of HPv-1 in Rwanda, by 1989). Because of the mainly heterosexual spread of HIV in Age, Residence, and Gender, 1986 Africa, and increasingly in the Caribbean, a large and grow- ing proportion of AIDS cases are infants and children. The HIV-1 seroprevalence (percent) incidence of AIDS is probably underestimated, however, be- 35 cause of technical problems in the diagnosis of AIDS and HIV infection in infants. In addition, there is growing evidence 30 - that an as yet unknown proportion of perinatally infected "I children do not hecome ill until the age of seven to ten 25 - years. This implies that it will take at least another decade / before the full spectrum of morbidity and mortality of peri- 20 / natally acquired HIV infection is known. Although a limited number of anecdotal cases of HIV infec- 15 , tion acquired through breastfeeding have been reported 10 (Ziegler and others 1985; Colebunders and others 1988; Hira, / Kamanga, and Bhat 1989), it appears that contaminated breast 5 milk does not contribute significantly to the transmission of HIV infection from mother to child (Ryder and others 1990). 0 Furthermore, Kennedy and others (1990) have convincingly 0-5 6-15 16-25 26-40 >40 All ages Males Females argued that the health damage from bottle-feeding would probably exceed any gains from the averted transmission by Rural Urban Rural breast milk.6 Thus, there is currently no reason to modify - Urban current promotion ofbreastfeeding in HIV endemic areas in the developing world. Source: Bugingo and others 1988; Rwandan seroprevalence study group 1989. HIV Infection and Sexually Transmitted Diseases 463 instance, in Cote d'lvoire. In addition, within each country, Furthermore, dense urban populations with high rates of drug prostitutes are at a much higher risk of infection and constitute use or STDs, like those of Abidjan and Bangkok, may soon part of a "core" group of highly sexually active individuals experience rapidly spreading HIV epidemics (De Cock, Pozter, whose special role in the epidemic is described below. In the and Odehouri 1989; Phanuphak and others 1989). Latin American cities homosexuals also show the higher prev- In 1989, WHO (1989c) estimated that by the year 2000 alence rates associated with core groups. annual adult AIDS cases would rise from the present level of Annual incidence rates of 0.5 to 1.8percentforHlv infection fewer than 100,000 per year to more than 800,000 per year. among adults without particular risk factors have been docu- In the same survey it was estimated that adult HIV infections mented inCentral Africa (N'Galy andothers 1988). Still, such would rise to 13 million worldwide (Chin, Sato, and Mann rates were as high as 10 to 40 percent among highly exposed 1989). prostitutes in Kinshasa and Nairobi (Laga and others 1989; In addition to the morbidity directly caused by HIV infection, Plummer and others 1991). one should also consider the excess morbidity from other In table 20-5 we present the cumulative number of cases of diseases as a result of HIV infection. This has already been AIDS by continent reported to the World Health Organization documented for tuberculosis in the United States and several (WHO) as of August 1989. The data from Africa in particular African countries, where the incidence of tuberculosis is rising probably represent gross underestimations, and a figure of as a result of the HIV epidemic (cDc 1989; Colebunders and 250,000 AIDS cases by 1988 seems closer to reality. These others 1989; Standaert and others 1989). morbidity data ultimately also indicate the number of deaths due to HIV infection. Socioeconomic Correlates of HIV Infection Although HIV infection has been introduced recently in all populations (mid-1970s to mid-1980s), it has spread more Information on the correlation, or lack of correlation, of STD rapidly in some than in others (figure 20-3). Possible patterns incidence rates with socioeconomic indicators would be epi- of morbidity and mortality during the next twenty-five years demiologically useful in several ways. First, such information are difficult to predict because of inadequate knowledge of the could help in the effort to target interventions. Second, it natural history of the disease, of behavioral patterns in differ- could help us to understand the practices which spread the ent populations, and of the potential changes in the relative disease. And, finally, it could tell us more about the diseases' contributions of different modes of spread of HIV. Based on effects on society. current trends in some Latin American and Caribbean coun- Unfortunately very little information has been available on tries, however, it is anticipated that heterosexual transmission STDs. Because individuals have incentives to hide an STD infec- will become much more important in the Americas, and tion-and because the rich can hide these infections more perhaps also in Asia, than it is at the present time. In the near successfully-any data on STD incidence or prevalence by future, it appears that proportionately more women and more social class were known to be suspect. With HIV infection, poor people will be among those with HIV infection and AIDS. however, the situation has reversed itself. Because HIV/AIDS is Table 20-5. Reported and Estimated Cases of AIDS, by Region and Stage of Development, 1990 Reported cases Estimated cases Region Number Per million population Number Per million population Industrial countries United States and Canada 159,194 577.9 175,000 635.3 Europe 43,441 71.2 48,000 78.7 Australia 2,347 140.2 3,000 179.2 South Africa 650 17.8 1,000 27.4 Japan 294 2.4 300 2.4 New Zealand 207 61.7 200 59.6 Total 206,133 193.4 227,500 213.5 Developing countries Sub-Saharan Africa 81,833 173.0 648,000 1,369.7 Latin America and the Caribbean 31,943 71.7 275,000 617.3 Europe 2,462 13.9 52,000 293.7 North Africa and Mediterranean 686 1.5 1,000 2.2 Westem Pacific 180 0.1 2,500 1.9 Southeast Asia 141 0.1 5,000 3.7 Total 117,245 27.7 983,500 232.8 Total 323.378 61.1 1,211,000 228.9 Source: Reported cases from World Health Organization 1991a; Chin, Global Programme on AIDS, WHO; Zachariah and Vu 1988. 464 Mead Over and Peter Piot Figure 20-3. Evolution of HIV Seroprevalence Independence has not altered the ambivalence and outright in Selected Populations of Developing Countries hostility towards urban women generated during the colo- nial era.... Most East and Central African countries have High-Risk Groups taken actions to restrict urban women's activities. These Population infected (percent) range from the banning of miniskirts and other provocative 100 clothing to more threatening actions such as attempts to bus _._ -unmarried women out of town or arresting unescorted 80 - - Kiai male women found on the streets or in bars hotels and cinemas at 60 Nairobi _ .- S$D patients night (Larson 1989, p. 4). prostitutes 4 .. 40 , As a result of these powerful negative incentives, there are 20 - - Bangkok IV many fewer young women than young men in the sexually Kinshasa prostitutes , drug users most active age range in the urban centers of countries such as 0 I I I C6te d'lvoire, Kenya, Rwanda, Burundi, and South Africa. It 1981 1983 1985 1987 1989 is natural to hypothesize that in these situations the demand by the young men for prostitution services is very high. In contrast, in cities which are relatively hospitable to young Low-Risk Groups women, such as those of Zaire, Senegal, and Mali, the demand for sexual services can be expected to be smaller. Population infected (percent) But the quantity of commercially supplied sexual services is 30 determined by the supply as well as the demand. The supply Kampala pregnant will be high only if the potential suppliers have relatively few women remunerative alternatives. We hypothesize that the number of 20 women enrolled in secondary school per 100 men enrolled is Kinshasa .............. ...pregnant a good indicator of the opportunity cost of the time of urban 10 women women.8 In cities in which women are relatively well educated, we expect that, other things being equal, fewer women will 0 _ ag I I I I I become prostitutes and the HIV virus will spread more slowly. 1981 1983 1985 1987 1989 More empirical evidence for the proposition that women's education and a high ratio of females to males might both Year of observation contribute to reducing the prevalence of STDS is presented in figure 20-4. The upper-left panel of the figure displays a scatter Source: Piot and others, 1990. plot of 1987 HIV prevalence rates against the urban ratio of females to males. Note that the higher prevalence rates are known to be fatal, wealthy people are alleged to have identified associated with urban areas in which there are many fewer themselves in the hope of getting treatment. Thus, some have young adult women than men. The upper-right panel of figure argued that estimated HIV infection rates by class are biased 20-4 shows, for the same countries, the association between toward a higher infection rate for higher classes. HIV prevalence and the ratio of female to male secondary The data on this question are still scarce; we are aware of no education enrollment. l0 Here, the relationship is even more data outside Africa. In table 20-6, however, we present infor- marked, with the highest seroprevalence rates observed in mation from three studies in three different African countries, those countries with the poorest record on female education. all of which confirm the hypothesis of higher prevalence rates For the smaller number of countries for which the infection at higher income levels. These pattems might be explained by rate of female prostitutes was known in 1987, the left lower the relatively greater access of the rich to foreign travel or, panel of figure 20-4 demonstrates even more strongly the alternatively, by greater rates of sexual partner change among hypothesized effect of small female-to-male ratios-a high higher-income adults. In either case, the pattern may dissolve prevalence rate of HIV infection among prostitutes. as the epidemic spreads in a given population. Additional work A multivariate regression of urban adult HIV seroprevalence is needed not only on prevalence rates but also, and especially, rate on these two indicators of the status of women explains on incidence rates by social class so that public health officials 48 percent of the variation in prevalence rates and is statisti- can determine how best to target their control efforts. cally significant at the I percent significance level. In this It has been asserted, but not demonstrated, that the corre- sample, each of the two variables, urban adult sex ratio and lation between infection rates and higher social status is posi- female education, contributes independently and significantly tive for men but negative for women.7 Larson (1989) and to explain variation in the seroprevalence level." Caldwell, Caldwell, and Quiggin (1989) advance an image of Because this regression equation was estimated on a cross- the sociosexual role of the African woman that predicts just section of cities, in some of which the HIV prevalence rates may such a possibility. In Larson's words: still be rising, the relationship is likely to shift over time. Still, HIV Infection and Sexually Transmitted Diseases 465 Table 20-6. Relationship of Socioeconomic Status with Higher Rates of HIV Infection in Sub-Saharan Africa Samnple Countr-v Date Population Size Socioeconomic indicator HIV infection rate (percent) Rwanda 1987 National sample 1,255 Education of urban adults Primary or less 20.8 More than primary 29.6 Zaire 1987 Employees of urban 5,951 Job textile factory Worker 2.8 Foreman 4.6 Executive 5.3 Zambia 1985 Patients, blood donors, 1,078 Years of education completed personnel of urban hospital 0-4 8.0 5-9 14.7 10-14 24.1 15 or more 33.3 Source: Melbve. Nselesani, and Bavley 1986; Bugingo and others 1988; Ndilu and others 1988. the strong estimated effects of these two indicators of low of each disease to the total burden of ill health borne by a given female status on the prevalence of HIV infection are likely to population. Furthermore, these estimates indicate the total persist. An implication is that one of the most promising ways health benefit that would accrue from eradicating one of the to fight STDs over the longer run is to improve the education diseases. and increase the number of urban women.12 The example of smallpox notwithstanding, eradication is unfortunately not usually an option available to public Public Health Significance of STDs health decisionmakers. Instead they are asked to make de- cisions at the margin, allocating 10 percent more of their ThissectionpresentsestimatesoftheburdenofsTDsandofthe resources to this prevention effort, perhaps by cutting re- benefit of averting a case of CSTD and HIV infection. sources on another effort. As GHAP recognized (1981), esti- mates of total burden are not useful for these marginal Health Lost and Saved decisions. Instead decisionmakers need to know (a) how maniy disability-adjusted life-years could be saved for every Manyothercategoriesofdiseases,aswellascsTDsandHlv,have case of the disease prevented or cured and (b) the relative substantial public health significance. In order to attach a costs of preventing or curing a case of each disease. In this priority ranking to STD interventions in a specific country, it is section we present and interpret the burden measures, from necessarytoattempttoquantifythepublichealtheffectofsTDs both the static and the dynamic perspective. Then in the in comparison with other diseases in that country. Two broad next two sections we turn to the second and perhaps more methods are relevant. First, one can compute any of a variety important issue, the estimate of the health effect of averting of measures of the good health that is lost as a result of these a case of an STD, again from both a static and a dynamic diseases in comparison with others. Second, one can estimate perspective. the good health that would be saved by interventions on each category of disease. The Static Burden of STDs The computation of the amount of good health lost due to each prevalent disease in a developing country was Analysesof the public health importance of sexually transmit- pioneered by the Ghana Health Assessment Project Team, teddiseaseshavenottypicallyaccordedthemgreatimportance hereafter referred to as GHAP (1981). The method the inves- either in absolute terms or in relation to other infectious and tigators used was to multiply a measure of the disability- parasitic diseases. For example, GHAP ranked "venereal disease" adjusted life-years lost (DALYs) from a case of each disease number 53 in order of the burden it imposed on the population by the annual incidence of that disease to obtain an estimate of Ghana in the 1970s. Partly because of AIDS and partly of the average annual burden of each disease on a typical because of a new understanding of how widespread the CSTDs member of the population. 13 Expressed as a formula, the are, especially in Africa, and how damaging their sequelae, equation for each disease is: opinion has begun to shift toward a more serious appreciation of the harm done by these diseases-and of the potential DALYs = Cases/capita/year x DALY/case benefits of their alleviation. (See, for example, Curran 1980; Brown, Zacarias, and Aral 1985; Grimes 1986; Washington, By comparing estimates of DALYs per capita per year across Arno, and Brooks 1986; Washington, Johnson, and Sanders diseases, one arrives at an estimate of the relative contribution 1987; Wasserheit 1989). 466 Mead Over and Peter Piot Figure 20-4. Correlation of Urban HIV Infection with Gender and Schooling in African Countries, 1987 HIV seroprevalence in adults (percent) 20 Rwanda 20 Rwanda 15 15 C6te d'lvoire South Africa C6te d'lvoire 10 Malawi 10 Central Malawi African Central African Republic Zaire Republic Zaire ZaireZUganda Burundi Tanzania Burundi Zambia Zimbabwe Tanzania Kenya Mali Kenya Cameroon Mali eZimbabwe Mozambique Ethiopia Mozambique Mozambique Senegal Cameroon Botswan 0 Botswana Chad Mauritius Benin 0 Chad Benin Senegal Sudan Mauritius Ethiopia 60 80 100 120 140 15 40 65 90 115 HIV seroprevalence in adults (percent) 100 100 Rwanda Rwanda 80 80 Kenya Kenya 60 60 Malawi Malawi 40 Tanzania 40 Tanzania C6te d'lvoire Cote d'lvoire 20 Zaire Central African 20 Zaire Republic Central African Mali Mali Republic Ethiopia Cameroon Senegal Ethiopia Senegal 0 South Africa 0 Cameroon 60 80 100 120 140 15 40 65 90 115 Urban females per 100 males aged 20 -39 Secondary school enrollment of females per 100 males Source: U.S. Bureau of the Census 1991 HIV Infection and Sexually Transmitted Diseases 467 In view of the heterogeneity of the epidemiological picture of incidence for each STD in each of the two extreme urban and the gaps in available information, it is difficult to discuss settings described in table 20-7, a high-prevalence urban in detail the public health effect of STDs on any specific setting and a lowprevalence one. Our guesstimates appear country. Instead we define a typology of countries according in table 20-8, together with estimates for the other diseases to the estimated prevalence rates of CSTD and HIV infection in which have been derived from the GHAP study of Ghana. The the sexually active populations of their main urban centers. In estimated STD incidence rates might apply anywhere in the table 20-7 we define nine pattems of STD prevalence and world that an urban area can be characterized as in table tentativelyclassify represenitative counitries fromAfrica, Latin 20-7. The estimated incidence rates for the other diseases, America, and Asia. however, are specific to Ghana in the late 1970s and can In order to compute estimates of the "burden" of each of only indicate the rough orders of magnitude of the burdens the various diseases in model countries, we apply the of these diseases in other times and places. method originally developed by GHAP (1981) and extended Multiplying the age-specific incidence rate of a disease by Barnum (1987) and Over, Bertozzi, and Chin (1989), (from table 20-8) by the disability-adjusted life-years lost which is based on the above simple equation. In appendix per case from that disease at that age group (a calculation 20A we present calculations of the number of disability- intermediate to obtaining the figures in table 20A-3), con- adjusted life-years lost per case for the SInDs and other im- verting the result to days (by multiplying by 365), and, portant diseases, which is one of the two pieces of the finally, computing the weighted average of these products formula. The other necessary component is the estimated across all age groups. with the age-group size as the weight, incidence rate for each disease. Because incidence rates of yields the estimated number of healthy life-days lost per STDs are largely unknown, we "guesstimate" the age structure capita per year from a given disease. In table 20-9 we present the results of this calculation, and in figures 20-5 and 20-6 we portray them. Table 20-7. Urban Prevalence ofHuwvan4dCSTD, The rather surprising feature in table 20-9 is that the burden by Region of sTDs in a high-prevalence urban area is a suibstantial fraction of Prevalence of HIV infection . the entire disease burden of that population. By itself, HIV ranks CSTD Lou) or second on the discounted disability-adjusted life-days (DALD) Prevalence' unknown Intermediate High criterion and moves up to first when lost days are weighted by Low or Cambodia Angola None their relative productivities. Furthermore, the sum of the bur- unknown dens of HIV, syphilis, and chlamydia equals 85 DALDs and 63 Cape Verde Botswana productivity-weighted discounted disability-adjusted life-days China Burkina Faso (PDAl Ds) per capita, enough to place these STDs in aggregate at Eastem Europe Namnibia the top of the list in importance in an urban high-prevalence MiddleEast SierraLeone area. The eleventh ranking disease in table 20-9 and figures Togeo 20-5 and 20-6 is chlamydia, which attains its rank because of Viet Nam an extremely high incidence rate in ages fifteen to fifty and an assumption that 5 percent of cases are permanently disabled to Senegal Mali the extent of 30 percent incapacity for women (crippling Zaire pelvic inflammatory disease) and 50 percent for men (severe Zimbabwe urethral strictures). Only chancroid appears in these calcula- tions to have an effect as small as that attributed by GHAP to all High Brazil Cameroon Burundi Colombia Caribbean C6te d'lvoire STDs together, but new information on the likely links between Lesotho Nations Malawi genital ulcers and the probability of transmission of HIV infec- Madagascar Central African Rwanda tion may promote chancroid far above its place here. (See the Mauritania Republic Uganda discussion below of the effect of genital ulcers on the transmis- Mexico Ethiopia Zambia sion of STDS.) In contrast, in low-prevalence urban areas, Nigeria Gabon syphilis is the most burdensome STD, and its burden is less than Philippines Ghana that of any of the top fifteen Ghanaian diseases. Swaziland Kenya Mozambiique Tanzania The Short-Term Dynamic Burden of an STD Epidemic Thailand In order to illustrate the essential features of dynamic STD a. High:rates f gonrrhea mong exual actie aduls exced prceni epidemiology and to characterize the differences among epi- or prevalence of serological markers for syphilis exceed 10 percent in pregnant women. Intermediate: prevalence below these levels but at least I percent for demics of the different STDS, it is useful to experiment with two both of these diseases. Other countries are categorized as low or unknown." simple models of an STD epidemic. In this section of the b. Low or unknown: less than I percent. Intermediate: between I and 10 chapter, these models extend our estimates of the burden of percent. High: moire than 10 percent. ctDS by models te our e seima uses theburdnao Source: Authors. STOs by incorporating the fact that each case causes additional 468 Me ad Over and Peter Piot Table 20-8. Incidence of STDs and Other Important Diseases, by Age (per .0OO people) Age (years) Disease Prevalence' Sex 0-1 1-4 5-14 15-49 50-64 65+_ Sextally transmitted disease Chancroidl High Both 0 0 0 12.5 9 0 Low Both 0 0 0 1.25 0.9 0 Chlamydia High Male 50 0 0 50 5 0 High Female 50 0 0 37.5 0 0 Low Male 5 0 0 5 2.5 0 Low Female 5 0 0 3.75 0 0 Gonorrhea High Male 25 0 0 45 4.5 0 Hilgh Female 25 0 0 30 0 0 Low Male 2.5 0 0 5 0.45 0 Low Female 2.5 0 0 3 0 0 HIV High Both 20 0.5 2.5 15 2.5 0 Low Both 0.1 0 0 0.3 0 0 Sv'philis High. Male 25 0 0 20 2.5 0 High Female 25 0 0 20 2.5 0 Low Male 2.5 0 0.5 2 0.25 0 Low Female 2.5 0 0.5 2 0.25 0 Other diseases Birth injury n.a. Both 36 0 0 0 0 0 Cerebrovasculardisease n.a. Both 0 0 0 3 9 1 2 Cirrhosis na.a. Bothl 0 0 0 1.2 1.2 1.2 Congenital malformations n.a. Both 21 0 0 0 0 0 Gastroenteritis na. Both 800 200 5 5 5 5 Injuries' n.a. Both 5 6 7.7 9 5 5 Malaria n.a. Both 600 100 0 0 0 0 Measles n a. Both 375 150 0 0 0 0 Pneumonia, adult n.a. Both 0 0 0 12 13 15 Pneumonia, child n.a. Both 7 9 3 0 0 0 Prematurity nia. Both 213 0 0 0 0 0 Severe malnutrition 1n.a. Both 8 6 0.5 0.1 0.1 0.1 Sickle cell n.a. Both 28 0 0 0 0 0 Tetanus (neonatal) n.a. Both 11.2 0 0 0 0 0 Tuberculosis n.a. Both 0.5 0.5 1 3 3 3 n.a. Not applicable. a. For STI. urban areas of high or low prevalence in pattern fl countries, defined by rable 20-7. b. Such as accidents. Source: Authors; diseases not related to 1T0s, Ghana Health Assessment Project Team 1981 future cases. In the next section these saine models will be parameters, one set to describe the sexual behavior of the two useful in estimating the effect of alternative interventions on groups and one to describe the medical characteristics of the the course of an epidemic of each STD. disease to be modeled. The model presented first is a short-run model of the course We characterize the sexual behavior of the two groups by of an STD epidemic within the confines of an enclosed stable making the following assumptions: population. Although this miodel reveals some features of an . 1 . . . , ,, , @~~~~~~~ The core group of highly sexua lly active people increases HIV' epidemic, ItS focus Is too short-term to represent fully the important effects of such a slow-acting disease. Thus, after (btcen an women) in clu e 10ivdlonly 2 ' ~~percent of the 50,000 in the noncore group. exploring the implications of a short-term model, we turn to a presentation of a longer-term demographic model, which in- * Individuals in the core group are ten times as sexually corporates the interactions between an AIDS epidemic and the active as those in the noncore group, with the former having demographic features of a population. a new sexual partner every five days and the latter every fifty The short-run model posits just two groups of individuals, a days. 14 core and a noncore group, which have different sizes and * In choosing new sexual partners, individuals exercise no different rates of sexual activity. To predict the pattern of an preference according to group but instead select randomly epidemic from the starting prevalence rates of infection in among all individuals who are choosing new partners during these two groups, it is sufficient to specify only two sets of that time period. 9 HIV Infection and Sexually Transmitted Diseases 469 Table 20-9. Per Capita Annual Disease Burden of STDs and Other Diseases in Sub-Saharan Africa Discounted disability-adjusted life-days lost Discounted productive disability-adjusted life-days lost Disease Value Rank Value Rank Measles 68.9 1 45.1 2 Hiva 60.6 2 48.3 1 Malaria 55.1 3 35.2 3 Gastroenteritis 35.7 4 22.4 4 Syphilis' 15.9 5 9.3 5 Birth injury 12.4 6 7.7 6 Sickle cell 11.3 7 6.5 9 Prematurity 10.1 8 6.4 1 0 Pneumonia, child 9.7 9 7.1 7 Severe malnutrition 8.6 10 5.7 13 Chlamydia' 8.6 11 5.8 12 Cerebrovasculardisease 7.9 12 6.6 8 Injuries (i.e., accidents) 7.5 13 6.0 11 Tuberculosis 5.2 14 4.1 14 Pneumonia, adult 4.8 15 4.0 15 Tetanus (neonatal) 4.2 16 2.6 17 C'irrhosis 3.7 17 3.0 16 Congenital malformation 3.6 18 2.2 18 Gonorrhea- 1.9 19 1.2 19 Syphilis" 1.3 20 0.7 21 HIV 1.0 21 0.8 20 Chlamydiab 0.8 22 0.5 22 Gonorrheab 0.6 23 0.4 23 Chancroida 0.5 24 0.3 24 Chancroid" 0.05 25 0.03 25 Total nons,r)s' 303 204 Total sTis' 88 65 Total sTimsb 3.6 2.5 Note: Burdens are summed over the entire population of both genders. a. Sexually transmitted disease; high-prevalence urban area. b. Sexually transmitted disease; low-prevalence urban area. c. In addition to the listed non-SrDs, this total includes forty-one other diseases, all with values of discounted DALDs and productivity-weighted discounted DALDs lost per capita less than 4.0. Source: Ghana Health Assessment Prject Team 1981; authors. These specific values were used by Hethcote and Yorke For the purpose of this modeling exercise, only two aspects (1984, p. 38) in their gonorrhea model. For ease of refer- of eachdisease are required, the probabilityoftransmission per ence, they are presented together with some other deriva- new sexual partner and the number of days that an infected tive parameters in table 20-10. It must be emphasized that person remains infective before recoveringordying.17 In the these parameters are not considered to be representative of first four rows of table 20-11 we summarize this information African populations. Although they were originally chosen for six distinct STD epidemics to be modeled. In contrast to by Hethcote and Yorke to characterize a North American table 20-10, which is not intended to be particularly repre- gonorrhea epidemic, sexual behavior varies enormously in sentative of a specific part of the world, the estimates in the United States just as it does in Africa, and these para- table 20-11 have been adjusted to approximate as closely as meters are not known to be typical of a specific North possible the medical characteristics of these diseases in American population either. The intention here is not to Sub-Saharan Africa. Note that the probability of transmis- predict accurately every detail of an STD epidemic in Africa sion of HIV infection on a single encounter is assumed to with a single model, a hopeless and senseless task. Rather, increase by a factor of three to five in the presence of genital by presenting the results of simulations with this model, we ulcers. For this reason, we distinguish HIV in the absence of intend to develop an analytical technique for approximat- genital ulcers as a separately analyzed epidemic from HIV ing the public health effect of these diseases, assuming the infection in their presence. parameters of sexual behavior were known. Because a great The parameters which represent the average duration of deal of effort is currently being expended to determine these infectivity should be thought of as determined by the interplay quantities, before long it may be possible to substitute for of demand and supply for medical care. Factors which influ- the assumptions in table 20-10 some numbers based on encedemand includethediscomfortcausedbythediseaseand empirical data.16 socioeconomic characteristics of the infected individual. Dis- 470 Mead Over and Peter Piot Figure 20-5. Static Burden of STDS in Relation to ply considerations, such as the availability and quality of Other Diseases in a High-Prevalence African City medical services, will also affect the duration of infectivity. Because we believe that the demand factors conspire with the I l l supply factors in lengthening the duration of the average STD Measles in m adopt MM.HIlVa _ infection dn many developing countries, we adopt assumed Malaria _ _ _ _ _ _periods of infectivity which are roughly twice the durations Syphiis typically found in industrial countries. Birth injury Because our model does not distinguish between the sexes, Sickle cell Prematurity we must choose single values of the transmission probability Pneumonia, child and the duration of infectivity for each disease. Rows three and Severe malnutrition f Chlamydiaa six of table 20-11 present these parameters, which are the Cerebrovascular disease Injuries averages of the parameters for the separate sexes. Tuberculosis A full statement of the model requires the specification of Pneumonia, adult TPtan(neuonia,tad) the equations of motion that determine the flows of individuals Cirrhosis from the pool of susceptibles in a group (that is, the healthy, Congenital malformation i Gonorrheaa uninfected, nonimmune individuals) into the pool of infected Chancroida i _ in that group (that is, those individuals who suffer the conse- 0 10 20 30 40 50 60 70 quences of the disease and are capable of transmitting it). In Disability-adjusted life-days lost per capita appendix 20B we present these equations and the derivations of the assertions made in the next few paragraphs. Discounted DALYs lost Discounted productive DALY s lost Suppose the noncore group were isolated from the core. Under this assumption, row eight of table 20-11 shows the value of the reproduction rate in the noncore group for each a. Sexually transmitted disease. simulated Recall that a rate less than Source: Authors' estimates; Ghana Health Assessment Project Team 1981. unity epidhemic. tdeall fato reproduction telf, dyin unity implies that the disease will fail to reproduce itself, dying out over time. Gonorrhea and chancroid would behave this comfort is known to vary by disease, by sex, and randomly way in the noncore group, because the reproduction rates in across individuals. Socioeconomic factors which affect de- this group, in the absence of interaction with the core, would mand for medical care of an STD will include the income, be 0.825 and 0.293, respectively. In contrast, note the partic- education, geographical location, attitude toward STDS, and ularly high value of 4.32 for the reproduction rate of HfV with access to household resources of the infected individual. Sup- ulcers in the noncore group. Also, note from row seven that Figure 20-6. Share of STDs in Total Disease Burden in a High-Prevalence African City, and Proportion of STDs Share of STDS Breakdown of in total burden STD burden /as roen ers / Malaria \ / 18% HIV 69o \l ~~HIV S STDs n i n fant 22'. \ I Gonorrhea diseases \ i;;; > 2% \ 5 //\ 10%hydia u adultdiseases / Other 7-- / Syphilis 7\ diseases18% Chancroid 1% Source: Authors' estimates. HIV Infection and Sexually Transmitted Diseases 471 Table 20-10. Assumptions for Base Run of Core disease in a single isolated group. Because infected individuals and Noncore CGroups remain infected only D days, it is possible for an equilibrium Sexual activity to be established between the flows of individuals into and risk groups out of the infected pool. In appendix 20B we show that the Parameter Symbol' Core Noncore equilibrium prevalence rate for an epidemic confined to a single group with reproduction rate R greater than one is Group si:e N 1.000 50,000 simply expressed as (R - I )/R. The fourteenth and fif- New sexual partners per day a 0.2 0.02 Toa enontr pe da pe rup aN 20 .0 teenth rows of table 20-11 present these rates, which XTotal encounters peT day per group a N 200 1.u00 Ratio of group encounters to total h 0-167 0.833 would be approached asymptotically as the epidemic be- Selectivity G 0 0 comes endemic within a single group. Note the marked Mixingcoefficients Ml, 0.167 0.833 difference in equilibrium prevalence rates between the M2 0.167 0.833 core and noncore groups for each disease, which is caused Starting prevalence by group 10 0.2 0.01 by the assumed tenfold difference in the rate of partner a. See appendix 20B for the definitions of parameters not defined in the change of the two groups. The only exception to this text. Alltheseparametersaregroupspecificexcept G, which isconstantacross pattern is the epidemic of HIV with ulcers, which, as groups. distinct from HIV without ulcers, eventually becomes prev- alent within more than 75 percent of the noncore as well as the core groups. This result derives from the extremely all the diseases can sustain themselves independently in the high value of the reproduction rate for this disease, which core. itself is due to the high probability of transmission com- Another important consequence derivable from the re- bined with the long duration of infectivity for HIV-ten production rate is the equilibrium level of prevalence of a years. 8 Table 20-11. Medical Parameters and Simulation Results for STDs HrIV without HIV with Pararneter Gonorrhea Syphilis Chlamydia Chancroid ulcers ulcers Transmission probabilities Male to female 0.6 0.250 0.4 0.350 0.03 0.1 Female to male 0.4 0.200 0.3 0.300 0.01 0.05 Q2 0.5 0.225 0.35 0.325 0.02 0.075 Duration of infectitvity (days) Male 45 180 90 45 2,880 2,880 Female 120 270 240 45 2,880 2,880 D& 82.5 225 165 45 2,880 2,880 Reproductive rates' R (core) 8.25 10.1 11.55 2.93 11.5 43.2 R (noncore) 0.825 1.01 1.155 0.293 1.15 4.32 Contact numberd 2.063 2.5 2.887 0.731 2.88 10.8 Parameters of equation of moton when selectivity, G, is set at Oc C,, 0.0167 0.0075 0.0117 0.0108 0.0007 0.0025 C12 0.0833 0.0375 0.0583 0.0542 0.0033 0.0125 C22 0.0083 0.0037 0.0058 0.0054 0.0003 0.0013 C21 0.0017 0.0007 0.0012 0.0011 0.0001 0.0003 Equiliyrium prevalence rates: isolated groupse Core 0.879 0.901 0.913 0.658 0.913 0.977 Noncore 0 0.012 0.134 0 0.132 0.769 Equilibrium prevalence rates: interactng groupsc Core 0.684 0.778 0.822 0 0.822 0.972 Noncore 0.178 0.259 0.316 0 0.315 0.778 a. Average of male-to-female and female-tr-male transmission probability per sexual partner. b. Average of male and female durations of infectivity. c. See appendix 20B for definitions. d. Weighted average of R (core) and R (noncore). e. Multiply by 1,000 for prevalence per 1.000. Source: Rows 1, 2, 4, and 5 present results of a Delphi survey conducted by the atithors. Other figures are authors' calculations; Hooper and others 1978. 472 Mead Over and Peter Piot As was pointed out earlier, the interesting feature of STLm ulcer and the no ulcer scenarios lends credence to the hypoth- epidemiology is that the core and noncore groups do not esis that a putative greater prevalence of genital ulcers in remain isolated from each other but instead choose sex Africa than in Asia or Latin America accounts for the appar- partners from the other group. Assuming that partners are ently more rapid heterosexual spread of AIDS in Africa.20 selected from the two groups without prejudice in propor- How "realistic" are the simulations presented here? The tion to their availability (that is, the selectivity coefficient equilibrium prevalence rates of gonorrhea, syphilis, and is 0 percent), Hethcote and Yorke (1984) show that the chlamydia in the noncore are similar to prevalence rates esti- reproduction rate for the core-noncore model is simply the mated in some African capitals (see table 20-2). The estimated weighted average of the two individual rates.19 This joint prevalence rate increase for HIV infection in the noncore group, rate is called the "contact number" by Hethcote and Yorke from I percent to 10 percent in ten years, is remarkably similar and in table 20-11. If it is less than unity, as it is for to the epidemic's trend in some of the worst-hit African chancroid. the disease will die out, not only in the noncore capitals (see figure 20-3). group, in which the reproduction rate is less than one, but How do the results of this modeling exercise change our also in the core group, in which it is greater than one. This understanding of the relative burden of STDs as presented occurs because the disease in the core is diluted by interac- above? One observation stands out. The total burden ofan STD tion between the core and noncore, making the disease is likely to be unequally distributed between the core and the unsustainable in either group. To produce the endemic noncore groups, with the core group bearing a much larger per levels of chancroid infection observed in African cities capita burden. The observation by Rothenberg (1983) regard- today, this simple model would require modified parameters, ing the relatively greater risk of STDs of inhabitants of urban as perhaps more active sexual behavior, a higher rate of trans- opposed to adjoining rural New York State communities is mission, or a longer duration of infectivity than we have directly applicable to African countries. Table 20-4 and figure assumed. 20-2 both display the dramatic differences that have been To explore the differences among the six different diseases, found between the prevalence levels of HIV infection in urban while holdingconstant the assumptions about sexual behavior, and rural African communities. The World Health we simulate an epidemic of each disease from the same initial Organization's Global Programme on Ait)s adopts the assump- conditions. We start each epidemic at a prevalence rate of I tion that rural prevalence levels are about one-tenth those of percent among the 50,000 noncore and a prevalence rate of 20 urban areas (James Chin, in a 1988 article cited by Bongaarts percent among the 1,000 members of the core. The time paths and Way 1989)2. 1 ofallsixepidemicsoveraten-yearperiodaredisplavedinfigure Although the model-generated values on the estimated 20-7. In the graph in the left part of figure 20-7, only the core trajectories of these epidemics cannot be used to predict actual groups in the six epidemics are compared, whereas in the graph prevalence rates, the behavior of the various epidemics de- on the right side the results for the noncore groups are pre- picted in figure 20-7 does capture some features of their real- sented. In comparing the two graphs, keep in mind that 100 world counterparts. In particular the modeled gonorrhea and percent of the core is only 1,000 people, whereas 10 percent of syphilis epidemics demonstrate the lesson leamed by STD mod- the noncore, the upper limit of the vertical axis, represents elers a decade ago, that the core group plays an important role 5,000 people. Because 2 percent of the noncore equals the in maintaining the prevalence rate of an epidemic in the entire population of the core, the data in figure 20-7 reveal that noncore at levels substantially above those that would obtain the size of any growing epidemic in the noncore quickly without interaction with the core. In the next section of this exceeds the total size of the core group. chapter, we will use the simulations presented in figure 20-7 as When the model is simulated for ten years, five of the six a base against which to compare the effects of interventions. epidemics closely approach their equilibrium levels in the core The results in this section suggest that policies which target (chancroid at zero), whereas three do so in the noncore. Note interventions at the core group may be particularly promising that instead of disappearing entirely, as gonorrhea would have candidates. done in the absence of the core group, the prevalence of this disease rises to an equilibrium level of about 3 percent in the The Demographic Effect of AIDS Epidemic noncore. Similarly the prevalence of syphilis, which would have leveled off at the endemic level of 7.4 percent, instead Because the model used in the last section does not incorporate reaches 28 percent at the end of the ten-year period. Although demographic changes, it is not suitable for modeling the long- HIV spreads rapidly in the core even in the absence of ulcers, run demographic burden of the [iIV epidemic. Still, several the large increase in its rate of spread in the nolncore as a result models in the literature do incorporate an explicit model of of the ulcers is interesting indeed. Of course the simulated rate heterosexually transmitted HIV infection into a demographic of spread assumes that the entire noncore (and core) popula- model of population growth. One of the first of these, by tions suffer from genital ulcers. Because such ulcers are caused Bongaarts (1988, 1989), simulates the spread of the infection by syphilis, herpes simplex, and chancroid, a more realistic over twenty-five years in a population structured to resemble simulation for iiv might lie between the ulcers and no ulcers that of many African countries. Without the epidemic, the scenarios presented here. The dramatic difference between the model country's population would continue to grow at 3 per- HIV Infection and Sexually Trans7nitted Diseases 473 Figure 20-7. Simulated STD Epidemics in Core and Noncore Gjroups Prevalence Prevalence 1.0 1.0 0.8 rz,~ - 0.8 0.6 ~~~~~~~~~~~~~~~~~~~0.6 0.4 ~~~~~~~~~~~~~~~~~~0.4 0 2 4 6 8 10 0 2 4 6 8 10 Elapsed years Elapsed years Core group Noncore group - - - Chiamydia - Gonorrhea --- Chancroid - Syphilis - - - - - HIV withoul ulcers HIV with ulcers Source: Authors' estimates. cent per year, birth and death rates would continue to fall, and Bongaarts's model projects these increased mortality rates life expectancy would continue to rise from its current level of onto an entire national population in which the baseline forty-six by 0.3 years per annum. mortality is currently about 19 per 1,000. Instead of declining Choosing hypothetical sexual behavior parameters and from 19 to 13 per 1,000, as Bongaarts assumed would occur plausible epidemiological parameters, Bongaarts finds that, without AIDS, the death rate is predicted to rise to 26 per once introduced, the HIV epidemic reaches a stable dynamic 1,000-twice what it would otherwise have been. With the equilibrium after about twenty-five years. By that time, the birth rate's downward trend unchanged but the death rate model predicts, the proportion of HIV-infected adults will sta- increasing, Bongaarts's model projects the population growth bilize at less than 30 percent, although approximately 55 rate to slow from 3 percent to 1.9 percent per year by the end percent of core-group females will be infected. Because of the twenty-five-year period. Bongaarts's model assumes a latency period of 9.3 years, in- This conclusion is remarkably robust to changes in two fected women have almost as many children as they would principal assumptions: the initial population growth rate and have had in the absence of AIDS. As a result the birth rate is the severity of the epidemic. If the population growth rate were unaffected by the epidemic and the increased perinatal and only 2.5 percent per year (lower than in any Sub-Saharan adult mortality offset one another, producing no change in the African country except Chad and the Central African Repub- dependency ratio as a result of the epidemic. lic), the population growth rate would remain positive unless What does change, of course, is the death rate. If the HIV the seroprevalence rate attains the extremely high level of 45 prevalence rate in a given adult population is 10 percent and percent among all adults (Bongaarts 1989, figure 7). With about 5 percent of those infected die each year, then the population growth rates at a more typical level of 3.0 percent AIDs-related mortality in this population will be 5 per 1,000 or 3.5 percent, the epidemic severity required to shrink the more than it otherwise would have been. Each additional 10 population would be even higher-55 percent or 60 percent. percent in the infection rate similarly adds about 5 per 1,000 Asecondmodelcombiningepidemiologicanddemographic to the mortality rate. Because the baseline mortality rate for features is by Anderson and others (1986). It predicts that adults fifteen through forty-nine in Sub-Saharan Africa is 5 or population growth rates could eventually become negative if 6 per 1,000, infection rates of 10 percent, 20 percent, and 30 the initial incidence rate of HIV infection in the susceptible percent will double, triple, and quadruple the adult mortality population exceeds the population growth rate. However, the rates, respectively. incidence rates estimated from a few Central African cities 474 Mead Over and Peter Piot exceed national population growth rates. For this reason, in Stanecki 1991). Future work with this model will characterize this and later papers, Anderson and co-authors conclude that the likely epidemics in the harder-hit African countries and HIV infection could cause populations to stop growing within on Latin American and Asian countries, in which transmission 30 to 100 years from the beginning of the epidemic. In their patterns and the underlying population dynamics differ sub- words, "[a] wide range of parameter values, all within the stantially from Africa. bounds suggested by current empirical studies, predicted as- For similar choices of parameters, all three models produce ymptotically negative population growth rates" (Anderson similar results regarding several key indicators of the long-term and McLean 1988, p. 231). burden of the epidemic. First, both models agree that the These predictions of population shrinkage in African coun- dependency ratio, measured by tries have received a great deal of attention. The incidence rates of HIV infection cited by the authors to support the a Number of people < 15 or >64 plausibility of their assumptions, however, are all from studies Dependency Number of people 15 through 64 of sexually active urban adults, some of whom were prostitutes. In fact, African populations are typically only 15 percent would not be greatly affected by the epidemic. This result at urban, and the unknown portion ofwomen who are prostitutes first seems surprising, in view of the anecdotal evidence is unlikely to exceed a few percentage points. Even an extraor- from heavily affected areas of Tanzania and Uganda that dinarily high incidence rate among all urban adults of 20 AIDS is the "grandmother disease," because it kills adults, percent per year results in only a 3 percent incidence rate in leaving orphaned children in the care of the grandparents. the entire population if the rural 85 percent population re- The explanation for this apparent contradiction lies in the mains uninfected, and this would be too low to result in distribution of the deaths. Because parental deaths precede negative population growth rates in most countries. If the child deaths in some households and follow them in others, incidence rate among urban adult susceptibles is instead under the overall dependency ratio can remain relatively constant 2 percent per year, as it was estimated to be among the despite large increases in the number of orphans from the employees of the main hospital in Kinshasa, then negative former households. population growth would require implausibly high rates of Still, the authors of all these models emphasize that any growth of infection in the rural population. reductions in population growth rate caused by the HIV The Anderson-May-McLean models are parsimonious epidemic will be due to increased deaths primarily among in portraying the interactions among demographic and young adults. Far from benefiting the population by reduc- epidemiologic variables. The models predict negative pop- ing its growth rate, the consequent doubling or quadrupling ulation growth, but only for implausible assumptions regard- of the mortality rate in individuals age fifteen through ing the initial incidence rates of HIV infection in national forty-four would have "disastrous effects on the health care populations.22 system, the economy and the fabric of society" (Bongaarts Recently the work of a team of modelers sponsored by the 1988, p. 36). In the absence of a vaccine or effective treat- U.S. Department of State's Interagency Working Group on ment, these consequences can be prevented only by a con- AIDS Models and Methods has come to fruition with the certed educational campaign designed to change sexual release of a model called the iwgA]ns model (Stanley and behavior. We will consider the possible design of such a others 1989). Embodied in a user-friendly computer soft- campaign below. ware package, the iwgAiDS model addresses such issues as transmission between prostitutes, their clients, and the The Gender-Specific Burdens of STDs wives of their clients and choice of sexual partners by age and location (that is, urban or rural), as well as transmission Do sexually transmitted diseases burden one sex more than by blood transfusion, needle-sharing, and the vertical route another in developing countries? There are tentative an- from mother to infant. Although sufficiently flexible to swers to this question from both the static and the dynamic project negative population growth, given sufficiently pes- perspective. simistic, "worst case" assumptions, the intermediate sce- Appendix 20A and table 20-8 include separate estimates for nario predicted by early runs of this model is a growth rate men and women of the STD mortality and morbidity rates in of 1.1 percent per year (Stanley and others 1989). More both high-incidence and low-incidence urban settings. Be- recently a careful attempt by the U.S. Bureau of the Census cause the static burden calculations summarized in table 20-9 to characterize a "typical" African HIV epidemic over and figures 20-5 and 20-6 are intended to compare the effects twenty-five years with this model predicts dramatic in- of individual diseases, the effects on the two sexes are com- creases in adult mortality comparable to those predicted by bined there. To the extent that appendix 20A and table 20-8 the Bongaarts model and a fall in urban life-expectancy by differentiate between the genders, we can return to them to nineteen years compared with the base case. At the end of estimate the ratio of the burden bome by women to that borne twenty-five years, the national population growth rate is by men for each disease. still growing at 2.2 percent per year, compared with 2.8 Because the statistics are based on pattem II urban areas, percent projected without the AIDs epidemic (Way and they ignore homosexual and iv drug transmission routes for HIV HIV Infection and Sexually Transmitted Diseases 475 infection. In these pattern 11 countries, the ratio of male to Figure 20-8. Annual Gender-Specific Burden of STDs female cases is close to unity for HIV transmission (see table 20-3). Therefore the data in appendix 20A and tables 20-8 and Discounted DALYS cost per capita 20-9 are not able to discriminate between the effects of HIV 50 transmission on men and on women. The effects of chancroid on the two sexes are also difficult to distinguish. For the 40 - remaining three diseases, however, the burden borne by each of the two sexes is presented in columns 4 and 5 of table 20-12, and a comparison of the per capita burden on women with that 30 on men for each of the five ST)s is shown in figure 20-8. According to the data in table 20-12 and figure 20-8, women 20 - in a high-prevalence urban setting bear a total STD illness burden of 44.4 disability-adjusted life-days per capita per year, 10 - about 3 percent more than the burden borne by men. If human papilloma virus and its sequela, cervical cancer, had been _ included, and if our methodology permitted us to capture the Females Males earlier onset of HIV infection in women than in men (Ryder and Piot 1988), the excess burden on women over that on men would be greater. _ HIV = Gonorrhea Syphilis = Chiamydia But these static calculations ignore the interactions between the presence of this disease and the socioeconomic positions Source: Table 20-12. of women in a population. Three types of such interactions are of concern. First, what are the effects of infertility caused by first man to have intercourse with a woman after she gives STDS on the future welfare of the affected individual-and how birth. In this same culture, a husband may return an infertile do such effects differ between men and women? Second, what wife to her parents and request the return of her bride-price have been the cultural adaptations to high prevalence of STDs (Reining 1972). and how have those adaptations affected the welfare of the two Whether a given cultural practice has developed in response genders? Third, what has been the effect of these cultural to the high prevalence of STD-caused infertility would be adaptations, if any, on the spread of STDs and on consequent extremely difficult to establish, even with an extensive ethno- levels of infertility? graphic survey, because it would require longitudinal informa- To do justice to any of these questions across all human tion specific to a given culture on both STD prevalence rates cultures would require an extensive ethnographic survey, and the frequency of various cultural practices.4 In the case of which space limitations do not permit. With respect to the first the practices just mentioned, however, it is hard to resist the question, however, it is well established that a woman's value interpretation that the practices have become more en- in a poor culture often hinges crucially on her presumed or trenched as a result of the high prevalence of STDs. Whatever demonstrated fertility. Although infertility clearly harms men their cause, it is clear that these cultural adaptations have as well, some cultures in which infertility is comtnon have resulted in an unequal sharing of the burden of infertility-im- customs which protect an infertile man-but not an infertile posing it more on the women than on the men. woman-from its consequences. For example, in northwestern The next link in the chain of causation is between such Tanzania, an infertile man may acquire an heir by being the gender-biased adaptations to the prevalence of STDs and the Table 20-12. Burden of STDS in High-Prevalence Urban Areas, by Gender Average incidence Discounted disabilit'v-adjusted life-days lost Discounteddisability-adjusted (per thousand) (per capita per year) life-years saved Disease WVomen Men Total Women Men Women Men Chancroid 6.2 6.2 0.5 0.25 0.25 0.2 0.2 Chlamydia 9.5 12.4 8.6 4.8 3.8 1.3 0.8 Gonorrhea 7.3 10.8 1.9 0.9 0.9 1.0 0.7 HIV 8.5 8.5 60.6 30.3 30.3 19.5 19.5 Syphilis 5.8 5.8 15.9 8.2 7.7 3.9 3.7 Total n.a. n.a. 87.5 44.4 43.1 n.a. n.a. Average DALY saved per case selected n.a. n.a. n.a. n.a. n.a. 5.6 4.7 n.a. Not applicable. Source: Authors' calculations. 476 Mead Otter and Peter Pioi further spread of these diseases. In an exhaustive study of dently postponing conception. In Frank's words, "[a]s long as infertility in Sub-Saharan Africa, Odile Frank writes: infertility remains prominent in Africa, large numbers of indi- Marital instability causedbyifertilityndthespredviduals and some entire populations will remain thwarted in real disease caused by marital instability and sexual mobility their ambitions to bear and raise children, and even larger rea dieas casedby arial nstbilty nd exul mbilty numbers may resist both intrinsic and extrinsic pressures for can form a vicious cycle. T'he movement of abandoned or nubr ma.eitbt nrni n xrni rsue o rjcafrav ,th movementiof hasband nor fertility limitation in the face of the risk to which they see rejected b.arren women to urban prostitution has baeen noted in Niger, Uganda, and the Centra African Republic. Similar others exposed" (Frank 1983, p. 40). in many of these societies, marital and sexual mobility on the Wasserheit draws an additional and more direct link be- par-t of the women is interpreted as a desperate attempt to tween STDs and the success of family planning programs: "In partoofthegwomen isd itoerpreted as atdsprae atttiepty to a the absence of accurate diagnosis and effective education and become pregnant, and tolerance on the part of society as a thrp[frss]itsfaeserorawmnoblehr means to maximize theirchances ofdoing so.... Once venereat disease was introduced into a community with some degree vaginal discharge on her contraceptive method than to enter- sexual ormarital mobility, itsdiffusion mighthavebeenassur tain and address [other possibilities]. The net result is often bytheexistingcustoms. [Subsequently]themobilityitself[may that the woman drops her contraceptive method under the have been] intensified to m.[uercome the fertility effects (Frank misimpression that the method caused an unrelated infection" 1983, pp. 22 and 26). (Wasserheit 1989, p. 10). If the cost of a large population growth rate is borne The existence of such a vicious cycle has recently been disproportionately by women, then these arguments that argued by Judith Wasserheir (1989), whose depiction of the STDs exacerbate the population growth rate reinforce the causal links is reproduced as figure 20-9. To Wasserheit and case that STDS affect women more than men. We leave as a Frank's argument, we add the intervening link of low status of challenge to others the quantification of these interaction women to their low education and their small numbers in effects. urban areas, factors which the earlier discussion suggests cause, or create the preconditions for, urban prostitution. Further- Lowering or Postponing the Incidence of STDs more, Wasserheit and Frank both argue that the resulting high level of infertility, far from damping the African population This section will discuss the principles and benefits involved growth, exacerbates it by preventing individuals from confi- in preventing STDs. Figure 20-9. Interaction between Sociocultural and Physiological Factors in Reproductive Tract Infections in Developing Countries Factors affecting prevalence Factors affecting morbidity Lower Inadequate sex education RTls Stigma associated with seeking care and health information Lack of available and appropriate Female circumcision and care and information other practices Multiple A t a sexual Antibiotic resistance partners Poorly performed transcervical upper procedures (abortion, IUD insertion) RTIs Urbanization and other migration Divorce of infertile women Infertility Ectopic pregnancy Local customs and traditions, e.g., Adverse outcomes of postpartum/perimenstrual abstinence, pregnancy high bride prices that delay marriage Inadequate sex education and health information Decreased social and economic productivity Decreased acceptance of contraception Source: Wasserheit 1989. HIV Infection and SexuaUy Transmitted Diseases 477 Principles of Primary Prevention of STDs havior modification can be encouraged only by information, education, and communication (IEC) campaigns and by indi- Principles of primary prevention follow, grouped by the form vidual counseling. Mandatory behavior modification requires of transmission targeted. the enactment and enforcement of government laws and reg- ulations. Passive responses are the least intrusive of the three TYPES OF PRIMARY PREVENTIVE STRATEGY. Thie four different types of strategies, depending on government policies which modes of transmission of the HIV virus are sexual intercourse, change either the inherent likelihood of the risky behaviors or mother-to-child (that is, "vertical"), transfusion of blood or the risk attached to them. blood products, and needles and other skin-piercing instru- In the rest of this subsection, available information on the ments. Each of these transmission modes applies to at least one effectiveness of interventions will be described for each trans- CSTD as well as to HIV. Primary prevention strategies are de- mission mode. By reference to table 20-13, each intervention signed to disrupt these modes of transmission, but a large can be associated with a voluntary, mandatory, or passive number of different policies could contribute to this goal. means of behavior modification. In table 20-13 we classify primary prevention strategies according to whether the indicated behavioral change is vol- PREVENTION OF SEXUAL TRANSMISSION TARGETING. According untary, mandatory, or only a passive response to environ- to Hethcote and Yorke (1984, p. 32), "jiun the early 1970s, the mental changes. This classification scheme, which has been prevalent idea was . . . that everyone who was sexually active borrowed from the field of injury prevention, directs our atten- could get gonorrhea and, consequently, . . . that 'gonorrhea is tion to the degree of coercion required to implement a policy everybody's problem."' Then in the late 1970s public health and thus to the probable cost of the intervention, opinion changed markedly in the United States, to the extent Each of these three strategy categories corresponds to a that the change was described by a WHO scientific working distinctly different type of government policy. Voluntary be- group in the following words: "This assumption is no longer in Table 20-13. Interventions for Primary Prevention of STDs and HRV Infection, by Mode of Transmission and Form of Behavior Modification Mode of Transmission Transfusion of blood and Needles and other skin- Behavior modification Sexual intercourse Mother-to-infant (vertical) blood products piercing instruments Voluntary (change Encourage Offer Encourage Encourage demand by changing Use of condoms or Voluntary screening of Test of donated blood Providers to sterilize information or vinicide pregnant women Recipient choice of reusable needles and preferences) Fewer sex partners Eye prophylaxis at donors to discard disposable Partner selection birth, on request Donor deferral ones Partner notification by Public information on Consumers to request patient referral low benefits and high sterile needles Sex educatiorn in risk of transfusion Providers co exercise schoolb caLltion with blood Early treatment of s-ms Needle exchange by Mandatory (change Enforce Enforce Enforce EnfJrce extemal incentive Regulation or prohi- Screening of pregnant Provider compliance Destruction of structure by imposing bition of prostitution women with transfusion disposable needles quantity restrictions Quarantine of infected Pregnancy counseling criceria Regulation of handling individuals of infected women 13lood screening and of blood by HIV- Partner notification by Contraceptive blood banking positive patients provider referral counseling Eve prophylaxis Laws against Iv drug abuse Passive response to Increase Increase 1Prevent Protide environmental changes Subsidies for condoms Subsidies for condoms Malaria, especially Needles that self- (change external and for ssr rrearment and tor s-n treatment among children destruct after one use incentive structure Tax or regulation of Tax or regulation of Anemia through price night clubs and night clubs and restrictions) alcohol alcohol Increase Ratio of women to Ratio of women to Price of transfusions men in urban areas men in urban areas Jobs for single women Jobs for single women Education for women Edlucation for woimen Souirce: Authors consrniction. 478 Mead Over and Peter PIot vogue in [the United States], and decisions for control are now scribed above, taking maximum advantage of every available based on the concept of the core of transmitters of disease opportunity to target control activities. Instead the possibility which postulates that a relatively small proportion of the of such targeting has been resisted with language which is population is contributing to the maintenance of the epidemic reminiscent of the decades-old descriprion of gonorrhea as and that it is precisely this group of transmitters that is partic- "everybody's problem."?5 ularly important Ifor disease control]" (wi-o 1978, p. 116, as The core is an epidemiological concept, rather than a pre- quoted in Hethcote and Yorke 1984) cisely defined social group, and so identifying the core in a Motivated to a remarkable degree by quasi-cost-effective- given population and reaching its members with education or ness arguments derived from mathematical simulations, public treatment may be extremely difficult and costly. Because some health strategies against gonorrhea in the United States turned prostitutes are relatively easy to identify, they are often the away from broad screening programs of the general population only part of the core that is reached. Countries like Ghana and to focus on "targeted" control programs. Targeting strategies Rwanda are conducting an AlDS-prevention program among included partner notification in an attempt to identify and their military personnel, demonstrating that targeting this part treat both the infector and the infectee(s) of the index case, of the core group is sometimes politically feasible (Lamptey rescreening of treated cases several months later to check for and Potts 1990). Clients of prostitutes are more difficult to reinfection, and outreach activities to high-risk groups. Some reach, but some of them acquire a symptomatic STD and present examples of the arguments marshaled by mathematical models themselves for treatment at a clinic. In general, men and for such targeted activities include the following: women with an STD are by definition at risk, and they are a * Because diagnosed and treated cases are likely to be in reasonably accessible group for HJV/STP prevention activities. the core group at high risk of reinfection, rescreening these Finally, sexually active adolescents can be reached through cases several months after treatment is "approximatelyfour the school system and through formal and informal youth casesseveral nonths after treatmentis "approximaty for organizations and activities, such as sports clubs and popular times as effective per number of individuals tested as [would concerts. be thel screening [of randomly selected individuals from the By selecting some groups for special attention, a targeting general population] in reduCing total incidence" (Hethcote policy runs two risks: those targeted may be-or feel-stig- and Yorke 1984, p. 32). matized, and those not targeted may react with a false sense * Becausetheinfectorsofanidentifiedcasearemorelikely of security. Because, in contrast to CSTDs, IIIv infection is to have belonged to the core group than are infectees of that incurable, policies to target prevention efforts at core groups case, contact tracing, screening, and treatment lowerdisease or at infected individuals, for example, through contact prevalencemoreperdiscoveredcaseiftargetedattheformer tracing, can assume ominous political overtones. HlV-in- than the latter (Hethcote and Yorke 1984, p. 85). fected persons (HIPs) will lack confidence in the ability of a * Under the hypothesis that a vaccination is developed publicly operated contact-tracing program to maintain the which provides immunity to gonorrhea, but only for a short confidentiality of its records. This fear is particularly well time, "post-treatment vaccination is about five times as founded in developing countries, where the staff of the effective per number of persons immunized as random vac- public health system is known by, and accessible to, most of cination in the population" (Hethcote and Yorke 1984, the population, and traditions of btureaucratic anonymity p. 33). and confidentiality are unknown. Once publicly identified, Empirical confirina.ion ofthe.coiicept ofthe core group was HIPs may lose substantial proportions of their civil rights, as presented by Rothenberg (1983)o either official or unofficial pressures attempt to isolate them from society. The problem is exacerbated in communities in The pattern of reported gonorrhea in upstate New York which people continue to believe that AIDS might be trans- (exclusive of New York City) in the years 1975-80 is one of mitted by shaking hands, by eating out of a common bowl, intense urban concentration, with concentric circles of or by other mundane, nonsexual social acts (Wilson and diminishing incidence. The relative risk for gonorrhea in Mehryar 1991). these central core areas, compared to background state rates, One strategy that has been effective in combating stigmati- is 19.8 for men and 15.9 for women, but as high as 40 in zation has also appeared to contribute substantially to program selected census tracts.... Contact investigation data suggest success: peer counseling. Peers have been successful counselors that sexual contact tends to exhibit geographic clustering as in pilot projects with prostitutes in Kenya, Ghana, and Cam- well. These observations provide support for narrow focus- eroon (Ngugi and others 1988; Lamptey and Potts 1990), with ing of epidemiologic resources as a major disease control truck drivers in Tanzania, and with youth in Zaire.26 Peers are strategy. not only better at communicating messages about safe sex; they are also better at finding other core group members in the first There are interesting parallels and contrasts between the place. Thus their use can bring the otherwise prohibitive costs history of public health thought regarding CSTDs and AIDS. It is of finding core group members down to quite modest propor- natural to suppose that public health thinking about control tions. The section below elaborates on the cost of targeted and strategies for HIV infection would build on the insights de- untargeted programs. HIV Infection and Sexually Transmitted Diseases 479 PREVENTION OF SEXUAL TRANSMISSION INTERVENTIONS. Avoid- become aware of their risk and can protect themselves. Sec- ance of sexual exposure to pathogenic microorganisms can be ond, the infected contacts can be offered a cure. In addition, achieved in three ways: (a) avoidance of potentially infected as argued earlier, sexual contacts of an index case are more sex partners (through sexual abstinence or mutual monog- likely than is the average CSTD patient to be members of the amy); (b) protection with a barrier method during sexual core group of transmitters. intercourse (condom use); (c) practicing only nonpenetrative The distinction between "mandatory" and voluntary part- sex ("safer sex"). All three methods depend entirely on indi- ner notification is somewhat artificial because it is impossible vidual behavior, which may be modified by various kinds of to verify that an individual has named all sexual contacts. health education, targeted or not. Because of the strong effects Partner notification programs can be more or less aggressive by of culture and religion on attitudes toward sex, the effects of varying the intensity of the interview and by offering the health education on sexual behavior necessarily differ across choice between (a) provider referral, in which the provider societies or even across subgroups of a society. Methods as notifiesthecontactwithoutrevealingtheidentityoftheindex different as mass campaigns, school programs, and face-to-face case; and (b) patient referral, in which the patient notifies the counseling are being used in the primary prevention of STDs contact and advises the contact to seek counseling, testing, and HIV infection, implying widely varying personnel and and, if necessary, treatment. recurrent cost requirements. Policy prescriptions for HIV infection require a different Although the incidence of HIV infection dropped spectacu- perspective. Until recently, public health departments have larly in selected groups of homosexual men in North America not been able to offer any treatment or other incentive which and Europe (Coutinho, van Griensven, and Moss 1989), how might persuade exposed individuals to run the (real or per- muclh of this decline can be attributed to health education ceived) risk of being identified and stigmatized as seropositive programs is unclear. In the developing world, the effectiveness for HIV. Early experiments in Colorado showed that, in the ofhealtheducationonsexualbehaviorisevenlessunderstood. United States, aggressive partner notification at a public Targeted education programs in several African cities resulted health laboratory reduced the number of people asking to be in dramatically increased condom use among clients of female tested. This finding supported worries that tying partner noti- prostitutes, although this was not always associated with a fication to testing would "drive the epidemic underground." decrease in the incidence of HIV (Lamptey and others 1988; Now that zidovudine (AZT) has been found to retard the Ngugi and others 1988; Laga and others 1989).The protection progression from HIV infection to AIDS, public health facilities offered by a condom during intercourse with a person with an able to offer AZT will for the first time be able to offer an STD is unknown, but probably approaches 100 percent if prop- incentive which will help to offset the fear of stigmatization. erly used and if no breakage occurs. Furthermore, if distribution of AZT can be channeled through Because spermicides and viricides can be controlled by public health facilities with active partner notification pro- the female partner, there has been substantial interest in the grams, its availability will substantially improve the ability of few studies to evaluate their in vitro activity against gonor- these programs to trace contacts. In countries in which AZT can rhea and chlamydial and IIIv infection. There is some evi- be afforded by the public health system, we expect the effec- dence that they protect against gonorrhea and chlamydia, tiveness of partner notification programs to rise dramatically. especially when used in conjunction with a condom or Even in the absence ofa cure for HIV infection, the availability diaphragm (Cole and others 1980; Austin, Louv. and Alex- of Az r at public health facilities in a country may stimulate a ander 1984; Rosenberg, Schultz, and Burton 1986; Louv and movement of the AIDS control effort along the same path others 1988). The authors of one study found no effect of previously followed by CSTD control programs: that is, toward contraceptive sponges on HIV acquisition in highly exposed tightly targeted screening, rescreening, and partner notifica- prostitutes in Nairobi (Kreiss and others 1986). Therefore, tion programs and away from general public education it is premature to recommend them in the STD/HIV preven- (Clumeck and others 1989; Toomey and Cates, 1989). tion program. Because of their potential importance in the Unfortunately for the application of such a strategy in primary prevention of STD and HIV, further research on this developing countries, the price of AZT is extremely high. The issue is a top priority. current cost of $8,000 per patient-year may drop to $4,000 as Partner notification, also known as contact tracing, is an a result of recent findings that a smaller dose gives the same important elementofcsTDprograms. First vigorously promoted effect as a larger one. That annual cost, however, exceeds the in the United States in the 1930s, tracing the sexual contacts annual per capita gross national product (GNP) of almost all of syphilis patients only began "in earnest [in] the 1940s when developing countries. effective treatment for syphilis became available" (Bayer 1990, A remaining quandary is whether the individual who learns p. 125). Subsequently it became accepted in the United States that he or she is seropositive for HIV will reduce high-risk sexual as an essential component of the public health strategies behavior. In American and European studies, knowledge ofHIV against syphilis in the 1960s, gonorrhea in the 1970s, and infection has sometimes marginally decreased and sometimes chlamydia in the 1980s. actually increased high-risk behavior (Office of Technology Partner notification acts to prevent the spread of a CSTD Assessment 1988, p. 14; Van Griensven and others 1989). epidemic in two ways. First, the uninfected sexual contacts Until a medical treatment is available to reduce the infectivity 480 Mead Over and Peter Piot of HIV-infected persons, the principal effect of partner notifi- grade for more than four days eliminates treponemes and could cation programs on transmission may be to alert uninfected theoretically be used in countries in which no laboratory partners of their exposure and thereby motivate them to adopt testing for syphilis is available. Areas without facilities for safe sex practices. There is a pressing need for operational syphilis screening, however, do not usually have facilities for research on partner notification programs in diverse pattem 1I storing blood (Meheus and Deschryver 1989). countries. The prevention of transfusion-acquired HIV infection relies on four methods: (a) testing of all blood donations for iiv PREVENTION OFVERTICALTRANSMISSION. For over a century, eye antibodies before transfusion and discarding donations that are prophylaxis at birth has been the method of choice for the HIV positive; (b) donor selection and recruitment (also known prevention of gonococcal conjunctivitis in the neonate. The as donor "deferral") aiming at excluding donors presumably at method is simple and its effectiveness is 93 to 97 percent (Laga high risk for HIV infection or at recruiting low-risk voluntary and others 1988; Laga and others 1989). It is officially recom- donors; (c) rational use of blood transfusion by the health mended as a routine practice in most developing countries, service; (d) prevention of conditions which call for transfu- although it is not always implemented (Laga and others 1986a sion, especially severe anemia and childhood malaria. Im- and 1986b). Ocular prophylaxis at birth is not effective against provement of prenatal care and the "safe motherhood" chlamydial conjunctivitis, which is not a sight-threatening dis- initiatives can reduce the need for peripartum transfusions, ease (Datta and others 1988; Hammerschlag and others 1989). which are one of the main indications for blood transfusion in Detection and treatment of gonococcal and chlamydial the developing world. infections in pregnant women may be the optimal strategy in controlling these diseases in mother and child, because it PREVENTION OF HIV TRANSMISSION BY SKIN-PIERCING TOOLS. prevents transmission to the neonate as well as complications Outbreaks of HIV infection in the Soviet Union suggest that and sequelae in the mother. As discussed in appendix 20C, intravenous injections may play an important role in the however, diagnosis of gonorrhea and chlamydial infection in spread of HIV infection in some hospitalized populations (Pok- women is technically demanding and as yet rarely available in rovski and others 1990). In general, the risk of infection from the developing world. Such screening is more effective if contaminated equipment is larger when prevalence among targeted at women with higher prevalence rates for these patients and staff is larger (Berkley 1991). Still, in contrast to infections, though this advantage may not outweigh the stig- other modesoftransmission, the contribution ofcontaminated matizing potential and lower social acceptability of selective needles and syringes to the spread of HIV in pattem 11 countries screening. Moreover, attempts to delineate high-prevalence is probably very small. Proper sterilization of reusable needles groups of pregnant women have met with variable success and syringes and use of disposable materials have been recom- (Laga and others 1986a and 1986b; Braddick and others 1990). merded to prevent the transmission of various microorganisms The prevention of neonatal chlamydial conjunctivitis and by injection. pneumonia is based on treatment of infected women, but, Those who inject drugs are at increased risk for HIV infection again, this is rarely practiced because of technical and fiscal through needle sharing and may be a source of infection for reasons. Targeting screening activities for C. trachomatis infec- other population groups through sexual transmission. Preven- tion (that is, to women below twenty years of age) has been tion strategies for iv drug users are still largely experimental but recommended for the United States (Arnal and Holmes 1991). mainly include prevention of Iv drug use itself by health edu- Screening of pregnant women for evidence of syphilis is cation, treatnent of drug dependency, needle exchange pro- recommended in most countries of the world to prevent con- grams, promotion of use of sterile or disinfected "works," and genital syphilis. Although such screening is irregularly promotion of safer sex practices. implemented, innovative programs involving primary health care workers using a simple and rapid assay for syphilis DEVELOPMENTS IN PRIMARY PREVENTION IN THIE NEXT DECADE. A antibody (rapid plasma reagin test), have recently been initi- tremendous effort is being undertaken to develop a vaccine ated in Africa (Hira, Kamanga, and Bhat 1989). against HIV infection, though with little obvious success thus Finally, screening pregnant women for i-v infect ion is in- far. It is possible that such a vaccine will become available by creasingly used as a means of preventing perinatal HIV infec- the year 2000. Attempts todevelop vaccines against gonorrhea tion, although its application remains controversial in many and chlamydial infection have failed so far, and it is not countries and its effectiveness is unclear when the pregnancy expected that they will become available in the near future. A is not interrupted (Braddick and Kreiss 1988). vaccine against syphilis seems even less likely. Virucidal products for local genital prophylactic use will PREVENTION OF TRANSFUSION-ACQUIRED HIV. Wherever possi- probably become available, but assessing their effectiveness ble, blood donations are being screened for syphilis and hepa- will be difficult. Methods forbarrier protection of women, such titis B virus markers throughout the world-though the cost- as the female condom, may become increasingly popular. As effectiveness of screening blood donors for hepatitis B virus knowledge of people's sexual behavior and motivations in- infection in hepatitis B endemic areas has been questioned creases, health education interventions may become more (Ryder and others 1989). Storage of blood at 4 degrees centi- effective. HIV Infectiot and Sexually Transmitted Diseases 481 The Benefit of Averting a Case: Static Analysis Table 20-12 and figure 20-8 support the case that the static burden of STDs is slightly greater on women than on As explained earlier, unless a decisionmaker is considering the men in a typical high-prevalence urban setting. Another eradication of a disease, knowledge of the total burden of that way to compare the genders is to ask how many disability- disease is of less use to him or her than would be an estimate adjusted life-years would be saved on average by averting a of the benefit of averting a single case. Coupled with informa- case of STDs in each gender. According to the last columns tion on the cost of averting a single case, this benefit measure of table 20-12, the static benefit is slightly higher when a would assist the decisionmaker to allocate resources. In the case of chlamydia, gonorrhea. or syphilis is prevented or absence of interactions among different disease programs, cured in a woman than if a similar case is prevented or cured the most cost-effective health resource allocation could be in a man. Averaging these potential benefits with the equal obtained by equating the health gained per dollar across all benefits from preventing HIV or chancroid in a man or a diseases.'' The existence of these interactions prevents a woman yields an average static benefit of 5.6 disability- straightforward application of this simple principle. Informa- adjusted life-years for every case prevented in a woman, tion on potential savings of disability-adjusted life-years from which is 19 percent more than the benefit of 4.7 DALYs of each averted case can still be a useful guide, which, together averting an STD in a man. Comparing either of these figures with ancillary information, can improve the effectiveness of with the non-STDs in table 20A-3 or figure 20-10 shows that health services in developing countries. the benefit from curing an average STD ranks below tuber- Earlier we defined the disability-adjusted life-years lost per culosis and just above measles and malaria. When the un- capita per year from a given disease as a function of the quantified interactions ofsTis with a woman's socioeconomic discounted (productivity-weighted) number of disability- status are added, STDs would rise even higher in importance. adjusted life-years lost per case of that disease; table 20A-3 provided these intermediate calculations which enabled the The Benefit of Averting a Case: Dynamic Analysis calculation of disease burdens for table 20-9. But the number of DALYs lost per case is also the estimated benefit of preventing, The above static analysis ignores the fact that each case of an or of quickly curing, a single case of that disease. Thus, in order STD prevented will, because of the contagiousness of the dis- '9 to estimate the health gain from averting a case of a disease, ease, prevent additional cases.7 From the dynamic perspec- we need only turn back to table 20A-3 and reinterpret its tive, a health care intervention could achieve health benefits content from this slightly different perspective. Figure 20-10 in two distinct ways. The simplest intervention is one that displays the results from table 20A-3 with diseases ranked in cures or prevents the infection of some people once. Such a order of the disability-adjusted life-years ([ALYs) saved per "one-time" program might be implemented by a mobile clinic case averted. that visits a town once, never to return.30 By changing the Because the difference between the burden calculations infection status of some people, a one-time program would done earlier and the benefit calculations in this section is to remove the effect of incidence rates, it is not surprising that Figure 20-10. Static Benefit of Preventing a Case conditions like sickle cell anemia, neonatal tetanus, and birth of STD and Other Diseases in Sub-Saharan Africa injury are favored by this change.28 Each of these has a rela- tively low incidence compared with measles or malaria, for Sick~le cell ___-_I example, but each robs (an African) society of a great many Tetanus(nenatal) _._'_ Hiv a-- I ife-years per case because its victims are so young. Discounting Severe malnutition at 3 percent is not sufficient to offset this effect, so these Crhosis diseases appear at the top of the list of benefits per case averted. Congenital chaiormna ion Because of the hiigh fourth-place ranking of Hiv, however, STDs Tuberculosis 0 ~~~~~~~~~~~~~~~~~~Measles continue to hold a surprisingly high rank. Furthermore, note Syphilis, Fa from figure 20- I 0 that adj ustment for the relative produc t iv ity Malaria of the lost years raises HIV infection to first place among all the Injuries (i.e. accidentsr = considered diseases. Pneumonitisdl Gastroenteritis Syphilis has been displaced bv cirrhosis, cerebrovascular Chlamydia, Fa disease, tuberculosis, and several other diseases from its fifth- ChGamydia, m a Gonorrhea, Ma place ratkingkin figure 20-5. Although the other sTFs rank Chancroida a __ ______ l much lower on the scale of benefits per case averted in this 0 5 10 15 20 25 static analysis, the same caveat applies as was mentioned Disability-adjusted life-years above: to the extent that chlamydia, gonorrhea, and chancroid predispose individuals to HIV infection, a portion of the benefit D,scounted DALYs gained _ Discounted productive DALYs of averting a case of lilv should be attributed to the inter- ventions which prevent the seemingly less important a. Sexually transmitted disease. diseases. Source: Authors; Ghana Health Assessment Project Team 1981. 482 Mead Over and Peter Piot postpone ill health but would not prevent the epidemic from equilibrium in the population. Similar patterns are displayed resuming its dynamic path toward equilibrium. The health by the chlamydia, syphilis, and chancroid epidemics. benefits of a one-time program are due to the value of postpon- In contrast to the classic STDS, HIV has a much smaller ing an episode of ill health or a death. Because ultimately death probability of transmission per sexual contact (0.01 to 0.03 in can only be postponed, the benefits of such a one-time program contrast to 0.4 to 0.6 for gonorrhea) and a much longer should not be disparaged. Although temporary, they may nev- duration of infectivity (eight years in contrast to up to one year ertheless be considerable. for syphilis). These differences make an HIV epidemic much The second type of intervention is a sustained one that alters slower to reach equilibrium than any of the CSTD epidemics. As either the sexual behavior of the population or the disease- a result, as can be seen in figure 20-12, the beneficial effect on specific parameters. If maintained indefinitely, such changes an Hlv epidemic of an intervention is still apparent ten years will affect not only the epidemic's path toward equilibrium but after the intervention. the ultimate equilibrium prevalence rates themselves and The pattern changes if the HIV epidemic occurs in a popula- therefore will have a permanent effect on the burden of the tion that is saturated by genital ulcer disease (CUD). Assuming disease. Of course, a sustained intervention will typically re- that GUD increases the transmission probability by a factor of quire sustained budget support. four, the effect is greatly to speed the epidemic and therefore to speed the return of the epidemic's path to equilibrium after A CORE AS AGAINST A NONCORE STRATEGY. In order to any one-time intervention. For this accelerated HIV epidemic, examine the potential effect of a one-time intervention, we a one-time intervention results in fewer new cases each month simulate an epidemic of each of the STDs in a population for a period of about five years than would have occurred composed of two sexually active groups, the noncore and the without the intervention. After about five years, however, the core. We assume that the core comprises only 2 percent of the number of new cases each month would have begun to slow population but is five times more sexually active than the even without intervention, because so few uninfected people noncore and has a starting prevalence rate of 20 percent would have remained to be infected. As a result of the inter- infection for each STD, compared with only I percent infec- vention, there are more uninfected people in the population tion in the noncore.31 Having simulated the course of each after five years, so that the number of new cases can be slightly epidemic in the absence of intervention, we then simulate, for larger during the sixth and subsequent years than it would have each STD, two different possible "one-time" interventions. In been without intervention. Figure 20-13 shows that this pat- one of these we cure 100 individuals in the core and in the tem holds whether the intervention is in the core or the other we cure 100 individuals in the noncore. Then, by com- noncore group.32 paring each of these two alternative simulations with the base runi, we compute which postpones more ill health and thereby THE ADVANTAGEOFTARGETINGTHECORE. In order to measure best improves the health of the population. the aggregate of all the ill health averted by the act of Figures 20-11 and 20-1 2 portray the dynamic effect of pre- venting 100 cases of gonorrhea (figure 20-11) and of HIV Figure 20-11. Dynamic Benefit of Curing or Preventing (figure 20-12). Each figure contrasts the effect of preventing 100 Cams of Go Theiat the Core and the Nonore C(>ups 100 cases in the core to preventing those cases in the noncore. For example, in figure 20-11 the top curve is New cases averted (per month) calculated as the difference between the number of new cases of gonorrhea each month with no intervention and 400 the smaller number if there is an intervention, which is applied only to the core group. In the sixth month after 100 core individuals are cured of gonorrhea, there are approxi- 300 - Effect of curing mately 380 fewer new cases of gonorrhea, in the core and 1o00 core cases the noncore together, than if the intervention had not occurred. In contrast, if the 100 cures were effected in the 200 - noncore group, the number of cases averted during the sixth month in both groups would only be about 40. In table 20-14 \ we compare the simulated effects of the two types ofpreven- 100 Effectof cusng we compare 100 noncore cases x tion programs for all six epidetnics. The results of this dynamic simulation reveal several things 0 about preventing cases of an STD that are hidden by the static analysis. First, note from figure 20-11 that the one-time cure 0 1 2 3 of 100 gonorrhea cases, whether in the core or the noncore, Year of simulated epidemic has virtually no effect on the number of cases three years later. The beneficial effect of a one-time intervention is transitory, a. Cure in the initial intervention thereby preventing more infection. because of the speed with which this epidemic approaches its Source: Authors. HIV Infection and Sexually Tra77smitted Diseases 483 Figure 20-12. Dynamic Benefit of Preventing 100 effect in the presence of GUD is subsequently offset after year Cases of HIV in the Core and Noncore Groups when five as the epidemic accelerates back to its original path. No Classic STDs Are Seen in the Population SENSITIVITY OF CORE STRATEGY TO ALTERNATIVE ASSUMPIONS. New cases averted (per month) The advantage of targeting the core highlighted in the above subsection is a significant result with potentially important 20 policy ramifications. Because it is derived from a simulation based on specific parameter values, the question arises as to whether the result would hold under alternative assumptions. 15 - The exercise already varies the disease-specific assumptions across the six different simulated epidemics. We now examine Effect of preventing 100 core cases the sensitivity of these results to changed assumptions about 10 sexual behavior. We focus on two parameters from table 20-10, the selectivity coefficient (G) and the rate of sexual partner change (a). 5 - As described above, the selectivity coefficient is the param- Effect of preventing 100 noncore cases eter which captures the degree of mixing between the core and noncore. The extreme values of 0 and 100 percent represent proportionate and zero mixing, respectively. The assumption 0 -I I I I of proportionate mixing (G = 0) used to this point in the 0 5 10 chapter obviously does not represent the epidemiology of Elapsed years since intervention pattem I countries, in which homosexuals tend to be exclu- sively homosexual and iv drug users also represent a relatively (not entirely) isolated community. Proportionate mixing, however, may be a much more realistic representation of preventing 100 cases today, we subtract in each month the behavior patterns in pattern 11 countries. number of new cases despite the intervention from the number Recall that the assumption of proportionate mixing implies predicted in the base run to have occurred in the absence of that an individual chooses indiscriminately among core and the intervention. Then we discount each of these averted noncore partners according to their availability for sexual future new cases back to the time of the intervention (at 3 contact. As table 20-10 shows, there are 1,000 noncore indi- percent per annum) and sum their discounted values to express viduals available with whom to change partners each day and the result as discounted new cases averted. Table 20-14 in- only 200 core individuals. Thus the assumption of proportion- cludes these figures for each disease and for each of the two interventions. Figure 20-13. Dynamic Benefit of Preventing 100 Consider again the disease gonorrhea. Despite the absence Cases of HIV in the Core and Noncore when of any long-run effect on prevalence rate of either the core or Transmission is Increased by Genital Ulcer Disease the noncore one-time intervention, each of them does save a substantial number of case-years. Curing (or preventing) 100 New cases averted (per month) initial cases in the noncore averts a total of 426 discounted future cases of gonorrhea (composed of 402 in the noncore and 100 F 24 in the core). Suppose, however, that the 100 cases pre- vented are instead extracted from the core group. In this case the total number of cases averted rises to 4,278, or ten times as many (of which only 231 are in the core). A policy of targeting 50 Effect of preventing the one-time intervention at the core averts ten times as many \ cases as would have been averted by a policy directed at the noncore. Furthermore, examination of the rest of table 20-14 shows that this result is robust across all the analyzed diseases. 0 Effect of preventi The absolute number of discounted future cases averted by 100 noncore cases the one-time intervention is roughly similar for gonorrhea, chlamydia, and syphilis but is much smaller for chancroid (because it is less infective) and somewhat smaller for the -50 _ , , slower epidemics of Hiv with and without the exacerbation of 0 5 10 genital ulcer disease. Note that preventing 100 cases of Iinv has Elapsed years since intervention only slightly more effect when the HIV epidemic is exacerbated by GUD than when it is not. This is because the initially greater Source: Authors. 484 Mead Over and Peter Piot Table 20-14. Dynamic Effects of Preventing 100 however, that these estimates do not yet include the additiosnal STD Cases in Core Rather than Noncore Group benefits of preventing or curing one of the CSTLOs because of the Discounted new cases consequent reduction of Eilv transmission. The next subsection averted over ten years3 addresses this issue. Targeting Targeting Ratio of core Disease core group nioncore group to noncore HEALTH BENEFITS WHEN STDR AkFFECT HIV TRANSMISSION. The Chancroid 810 83 9.8 hypothesis that CSTPS affect the efficiency of Hiv transmission Chlamydia 4,096 423 9-7 is both biologically and epidemiologically appealing. When Gonorrhea 4,278 426 10.0 sexual contact occurs between an Dilv-infected individual and HIV without ulcers 1,744 180 9.7 a susceptible person, the presence of genital ulcer disease in HIV with ulcers 2,106 201 10.5 either partner could plausibly increase the probability of HIV Syphilis 4,132 422 9.8 transmission by allowing the virus easier exit from the infected a.Sum of the savin-gs in both the core and noncore group of an initial person or easier entry to the susceptible one. If STDs do increase preventive or curative policy applied to only one group. The streams of saved the efficiency of HIV transmission, the higher STD prevalence cases are discounted at an annual rate of 3 percent. rates in Sub-Saharan Africa would provide a partial epidemi- Source: Authors'calculatiom. ological explanation for the faster increase in prevalence rates among heterosexuals in Africa than has been observed among ate mixing in this model implies that any individual in the core heterosexuals in North America and Europe. or the noncore has a five times greater chance of contacting a Whether STDs actually do increase the efficiency of HIV noncore than a core person. Among the sexually active groups transmission is inherently difficult to demonstrate, because of in highly affected Sub-Saharan African cities described in the need to control properly for the frequency of sexual partner Larson (1989) and Caldwell, Caldwell and Quiggin (1989). change. Without such control, an observed correlation be- this scenario may indeed be quite realistic. tween HIV infection and a past history of STDs could actually be Still, what would happen ifselectivity were greater than zero due to the correlation of both variables with the degree of or if sexual behavior were less active? Repeated in the upper- sexual activity. The researchers in several independent studies left panel of table 20-15 are the effectiveness ratios from the in both East and West Africa, however, have found large and right column of table 20-14. In the other three panels of table statistically significant effects of STDs on the efficiency of HIV 20-15 are displayed the effects on these ratios of varying either transmission, even after controlling for past sexual activity the selectivity or the rate of partner change in the core. Note (Cameron and others 1989; Ryder and others 1990). Further- that none of the three alternative sets of sexual behavior more, investigators in a recent study in Zaire have found that parameters casts doubts on the greater effectiveness of a policy even nonulcerative cases of gonorrhea and chlamydia increase aimed at the core group. Indeed in no case does the ratio of the the transmission probability (Laga and others 1990). core to noncore effectiveness fall below 8.5, and in one case it The simulations of an HIV epidemic both with and without rises to 55.5.33 ulcers in figures 20-12 and 20-13 demonstrate that, when an HIV epidemic is accelerated by STDs, the benefits of an inter- COMBINING THE STATIC AND DYNAMIC BENEFITS. In the absence of an intervention, an epidemic causes a loss equal to the Table 20-15. Sensitivity Analysis: Advantage discounted sum of the losses from all future cases of the disease. of Preventing or Curing STD in the Core Group, If an intervention is effective, its benefit is to reduce the Based on Selectivity and Rate of Partner Change magnitude of this loss by reducing or postponing future cases. - Selectivitv ((3) In the case of a one-time intervention like the one simulated Rate of partner here, prevalence rates of the STD epidemics ultimately reach ange o Disease 0.0d 0.30 the same levels as they would have reached without the inter- vention, only later. Thus the beneficial effect of the one-time 0.02' Chancroid 9.8 20.5 intervention is to postpone these cases.' Chlamdia 9.7 9.0 In table 20-16 we combine the static estimates (from table Gonorrhea 10.0 10.5 20A-3) of benefits from preventing or treatino a case of each HIv without ulcers 9.7 80.5 STP with the dynamic estimates (from table 20-14) of the Syphilis 9.8 9.5 discounted sum of averted future cases. Because the dynamic 0.04 Chancroid estimates are based on a ten-year simulation, they somewhat Chlamydia 27.1 11.2 understate the estimated furure effect of an intervention on an Gonorrhea 17.1 12.4 extremely slow epidemic, such as HIV in the absence of CL)PI. HlV without ulcers 17.4 12.5 These estimates nevertheless show dramatically the substan- HIV with ulcers 55.5 22.0 tial health benefits from the prevention or cure of the more Syphilis 17.0 11.5 serious STDs, such as syphilis and Div infection, in contrast to a. Base value. the less serious, such as chlamydia and even gonorrhea. Recall, Source: Authors' calculations. HIV Infection and Sexually Transmitwed Diseases 485 Table 20-16. Discounted Disability-Adjusted Life-Years Saved per Case Prevented or Cured When Epidemics Independent: Core vs. Noncore Dynamic benefitb Total benefitC Disease Static benefita Core Noncore Core Noncore Chancroid 0.2 1.6 0.2 1.8 0.4 Chlamydia 1.05 43.0 4.4 44.1 5.5 Gonorrhea 0.85 36.4 3.6 37.3 4.5 HIV without ulcers 19.5 340.1 35.1 359.6 54.6 HIV with ulcers 19.5 410.7 39.2 430.2 58.7 Syphilis 3.8 157.0 16.0 160.8 19.8 a. Benefit to only the cured or protected individual fro m curing or preventing a case of STP. Because the dynamic model does not distinguish the genders, the static benefits of averting a case in the two genders are averaged. b. Benefit to meople other than the cured or protected individual from curing or preventing a single case of STD. Computed by dividing the figures from table 20-14 by 100 and multiplving by the static benefit per case averted. c. Benefit of a single cure or prevention to both the individual and to the people he or she would have infected. Computed by adding the static benefit to the dynamic benefit. Source: Authors' calculations. vention occur sooner and also decay more quickly. If the CSTDs the intervention's indirect effect on the HIV epidemic out- could be somehow removed from a population in which they weighs that resulting from its effect on the gonorrhea epidemic. have been endemic, the rate of growth of iiv prevalence would The transformation from cases averted to disability-adjusted slow dramatically, taking more than twice as long to reach its life-years saved thus reverses the apparent importance of the ceiling. But STDs cannot be magically removed; they must two effects. The discounted sum of these health benefits are either be prevented or cured one at a time. 3,646 disability-adjusted life-years from averted gonorrhea When STDs affect HIV transmission, the health benefits of cases and 8,289 from averted HIV.36 The total dynamic effect curing or preventing a case of a CSTD should be larger than in of averting 100 cases of gonorrhea in the core group is the sum the absence of such an interaction. To estimate the increased health benefit of curing or preventing a case of a CSTD in the Figure 20-14. Effect of Curing 100 Core Cases presence of an HIV epidemic, we modify the simulation model of Gonorrhea When Gonorrhea Increases HIV used above to include an interaction between the STD being Transmission modeled and HIv. Because chancroid and syphilis cause genital ulcer disease, we assume that the probability of HIV infection New cases averted (per month) when one partner has either of these diseases is five times larger than if neither is so infected. The nonulcerating diseases, 4000- gonorrhea and chlamydia, are assumed to increase infectivity by three and two times, respectively.35 300 c Figure 20-14 illustrates the effects on the gonorrhea epi- 200 - demic and also on a simultaneous HIV epidemic of an interven- \ tion which cures or prevents 100 cases of gonorrhea in the core 100 New cases of HIV averted group. In the top panel, the top curve charts the number of 0 cases of gonorrhea averted each subsequent month as a result 0 5 10 of this intervention, which totals 4,2 78, just as it does in table 20-14 as depicted in figure 20-11 above. The bottom curve in Disability-adjusted life-years saved (per month) the top panel, although barely distinguishable from the hori- zontal axis, captures the indirect effect of curing gonorrhea 400 cases on the iiiv epidemic. The discounted sum of averted future cases of HIV infection is 425.1 over the ten-year span of 300 DALYs saved from HIV the simulation. 200 The bottom panel of figure 20-14 graphs the same two DALYs saved from gonorrhea epidemic paths as the top panel with one difference: each is 100 multiplied by the static benefit per averted case to yield the 0 flow of saved disability-adjusted life-years thar result from 0 I curing 100 cases of gonorrhea. Because, according to figure 0 5 10 20-10 and table 20A-3, the static benefit per case of gonorrhea Elapsed years from intervention is only 0.85 disability-adjusted life-years (average of 0.7 for men and 1.0 for women) and is 19.5 disability-adjusted life- years per case of HIV infection, the flow of health resulting from Source: Authors. 486 Mead Over and Peter Piot Table 20-17. Dynamic Effects of an HIv Epidemic greatly increased in the presence of an HIV epidemic. These from Preventing 100 STD Cases in Core and Noncore health benefits presented in table 20-18 and figure 20-15 are Groups used later for the cost-effectiveness calculations. Dascounted new cases of HIV averted over ten yearsa BENEFITS OF A SUSTAINED INTERVENTION. Each STD is char- Targeting Targenng Ratio of core acterized in table 20-11 by a probability of transmission on a Disease core group noncore group to noncore given sexual contact and by a duration of infectivity. Interven- Chancroid 275.9 17.4 15.9 tions which reduce either of these parameters for an STD in a Chlamydia 355.8 30.3 11.7 given population will slow the epidemic and, if the interven- Gonorrhea 425.1 36.2 11.7 tion is permanent, will reduce the equilibrium incidence and Syphilis 1,207.8 109.1 11.1 prevalence rates in the population. Interventions which re- duce the risk of transmission include circumcision (Bongaarts, a. Sum of the averted cases of HIV infection in both the core and noncore groups from an initial prevention or cure of 100 cases of classic STD in only Reining, and Conant, 1989; Moses and others 1989), the one of these groups. In addition to these health benefits, saving 100 cases of use of nonpenetrative sex, and, most important, the use of classic STD also reduces future cases of that STD. condoms. Source: Authors' calculations. Our method is analogous to that used above to model the effect of 100 cures. We use the same base runs for each of these two figures, or 11,935 disability-adjusted life-years. epidemic to represent the situation without an intervention. Adding this to the 85 disability-adjusted life-years gained for Then we model two alternative interventions, one in the core the lives of the individuals whose cases of gonorrhea were and one in the noncore. The intervention consists of assuming directly treated or prevented gives a total of 12,020 disability- that 100 people randomly drawn from the chosen (core or adjusted life-years saved, or 120.2 disability-adjusted life-years noncore) group are protected for one year from either becom- saved per case of gonorrhea averted. ing infected or from infecting others. In this way the probabil- In table 20-17 we show the total cases of iiiv averted when ity that a susceptible person will acquire the transmission 100 cases of one of the CSTDs are cured or prevented in the core through contact with a person of a given group is reduced by or the noncore group. These benefits are in addition to the the probability that at least one of the two partners is thus benefits given in table 20-14. As might be expected, the "protected."37 The result is that, for a year, the rate of growth indirect effects on averting cases of HIV, like the direct effects of the epidemics is slower than it would be in the absence of in table 20-14, are much greater when the intervention is this intervention. At the end of the year the epidemic resumes targeted at the core group. its normal pace and the prevalence rate converges to the same In table 20-18 we gather together the components to give equilibrium that it attains in the base scenario. the total health benefit, measured in disability-adjusted life- By protecting 100 people of the 1,000 in the core, we reduce years saved, per case of each CSTD prevented or cured. The the effective danger of sex with a core person by 10 percent. figures in table 20-18 are shown graphically in figure 20-15. Incontrast, protecting 100 people ofthe 50,000 in the noncore Note the extraordinarily large health effect of preventing or only reduces the danger of having sex with a noncore person curing a case of syphilis in the core group. By saving almost 400 by 0.2 percent. This difference is responsible for the dramatic disability-adjusted life-years per case cured or prevented, this difference in the effectiveness of the two interventions. Notice intervention has an even greater effect on health than would that a person-year of protection of a core individual in the the direct prevention of a case of HIV infection. The beneficial presence of an HIV epidemic alone saves 56.6 DALYs, almost effects of interventions against all the other CSTDs are also three times more than are saved by protecting a noncore person Table 20-18. Discounted Disability-Adjusted Life-Years Saved per Case Prevented or Cured When STDs Affect HiV Transmission: Core vs. Noncore Static and dynamic DALYs per Dynamic D.ALYs per case from Total DALYs saved per case case from classic STD only averted HIV onlya of classic STD Disease Core Noncore Core Noncore Core Noncore Chancroid 1.8 0.4 53.8 3.4 55.6 3.8 Chlamydia 44.1 5.5 69.4 5.9 113.5 11.4 Gonorrhea 37.3 4.5 82.9 7.1 120.2 11.6 HIV without ulcers 359.6 54.6 n.a. n.a. 359.6 54.6 HIV with ulcers 430.2 58.7 n.a. n.a. 430.2 58.7 Syphilis 160.8 19.8 235.5 21.3 396.3 41.1 n.a. Not applicable. a. Additional benefits of preventing or curing a case of a classic STD resulting from indirect prevention of HIV infection. Calculated by dividing the cases of HIV averted (table 20-16) by 100 and multiplying by 19.5, the sraric benefit of each case averted. Source: Authors' calculations. HIV Infecton and Sexually Transmitted Diseases 487 Figure 20-15. Total Health Benefit of Averting multiple of seventy-four for chancroid. Unless the cost of a Case of Classic STD When STD Exacerbates HIV targeting such a sustained intervention at the core is as much Transmission: Core vs. Noncore Strategy as forty times larger per person-year of protection than target- ing it to the noncore, the data in table 20-19 lead us to predict Chancroid that targeting to the core will be the more cost-effective strategy. Chlamydia The modeling method used togenerate the numbers in table Gonorrhea 20-19 allows only one or two epidemics to be present simulta- Ili neously in the population. When there are more than two Syphilis simultaneous epidemics, reducing transmission probabilities by protecting individuals, with condoms or otherwise, would HIV _ have an even larger health benefit. Because the most import- 0 100 200 300 400 ant interaction is with HIV and because averted HIV infection Disability-adjusted life-years saved per case averted provides most of the benefits of the sustained intervention, the effect of considering only two diseases at a time will be a minor understatement of the benefits of this intervention. FromHIV: core m FromsTD core From STO: noncore FrorriHiv noncore Costs of and Expenditure on Primary Prevention Programs Source: Table 20-18. It would be useful to present cost and expenditure information on all forty-three of the interventions categorized in table for a year. If the only STD epidemic is syphilis, which is much 20-13 by the mode of transmission they are designed to inter- more infectious, protecting a core person for a year saves 141.7 rupt. Unfortunately there is almost no information on either DALYS, about twenty times more than would protecting a non- the cost of or the current expenditure on primary prevention core person for a year. If the syphilis and HIV epidemics are both of STDS in developing countries. In this section we present the in the population simultaneously, however, protecting a core fragmentary available information on the cost of a few of these person saves a total of 384.6 DALYs, forty times more than interventions designed to interrupt three of The four modes of would protecting a noncore person. Of the 384.6 DALYs saved, transmission: sexual intercourse, mother-to-infant, and blood 141.7 are from syphilis, 56.6 are from HIV in the absence of any transfusion. For lack of information on expenditure, we con- interaction with syphilis, and 186.3 are from HIV, because the clude the section by analyzing the budgeted expenditures in slower syphilis epidemic has a smaller effect on HIV transmission. the Sub-Saharan countries' medium-term plans by 1988. The model estimates of a one-time intervention curing 100 people showed in table 20-18 that the intervention would have COST OF PREVENTING SEXUAL TRANSMISSION. Of the fourteen roughly ten times greater effect when targeted to the core than interventions listed in the first column of table 20-13, we focus to the noncore. In table 20-19 it is shown that a sustained ourcost-effectiveness analysisononly one: the encouragement intervention which lowers transmission probabilities also has of condom use through social marketing. Both the costs and a larger effect when targeted at the core. But this rime the the effects of the other interventions are insufficiently well advantage ranges from a multiple of forty for syphilis to a known to permit quantitative estimates. Table 20-19. Discounted Disability-Adjusted Life-Years Saved per Person-Year of Protection when STDs Affect HIV Transmission: Core vs. Noncore Static and dynamic DALYs per case Dynamic DALYs per case from Total DALYs saved per case from classic STD only averted HIV onlya of classic STD Disease _______ Core Noncore C ore Noncore Core Noncore Chancroid 1.4 0.2 79.5 0.9 80.9 1.1 Chlamydia 52.3 2.3 144.4 1.9 196.6 4.2 Gonorrhea 64.9 2.4 210.3 2.8 275.2 5.2 HIV without ulcers 56.6 19.9 n.a. n.a. 56.6 19.9 HlVwith ulcers 156.0 21.5 n.a. n.a. 156.0 21.5 Syphilis 141.7 6.5 242.9 3.3 384.6 9.7 n.a. Not applicable. a. Obtained by multiplying the discounted sum of future cases averted over ten years as a result of the intervention, times the static benefit of each case averted, averaging the male and female values ewhere they differ. b. Additional benefits of a person-year of protection due to both the direct protection against HIV and the indirect prevention of HIV infection. The portion of this benefit due to the indirect effect through the reduced prevalence of the CSTD can be calculated by subtracting the numbers for HIV without ulcers in columns 1 and 2 from the numbers in columns 3 and 4. Source: Authors' calculations. 488 Mead Over and Peter Piot Table 20-20. Two Condom Social Marketing Programs in Sub-Saharan Africa Feature Zaire Tanzania Location Kinshasa Six truck stops and two trucking companies Start date 1989 1989 Baseline condom use 200,000 condoms sold privately per year at Half of drivers had had more than fifty sexual partners cost of $1 per condom. in their lifetime. Condoms used consistently with casual partners by 42 percent of bar girls and commercial sex workers (csw) and by 37 percent of drivers. Accomplishments 13,000,000 condoms sold in two years at 6 Each truck stop is serviced by an average of 200 csws cents per condom. Expect to sell who use 20,000 condoms per month, or an average of 8,000,000 more in third year. Condoms 100 per csw per month or four per night. now present in 7,000 of 9,000 targeted outlets. Total cost per year About $2 million $100,000 budget plus free condoms provided by National AIDS Control Program. Amortization of vehicles provided free by African Medical and Research Foundation (AMREF). Perhaps $750,000. Total condom sales per year About 8 million 2 million Price per condom 6 cents in 1990, falling to 0.6 cents in 1991 No charge to customer as result of domestic inflation and devaluation. Cost per couple-year of protection Core' $300 $456 Noncoreb $30 $45.60 Contractor Population Services International Family Health International/AIDsTEcH Note: All money in U.S. dollars. a. Assumes four contacts per night, twenty-five nights per month. Cost per condom 25 cents in Zaire, 38 cents in Tanzania. b. Assumes ten contacts per month. Cost per condom 25 cents in Zaire, 38 cents in Tanzania. Source: Personal communications from Linda Cole, Family Health International, March 1991 (Tanzania) and from Richard Frenk, Population Services Intemational, February, 1991. The use of condoms by sexually active individuals with were the net subsidy cost to the government (Bulatao 1985; multiple sexual partners affects the STD epidemics by reducing Janowitz, Bratt, and Fried 1990). Because family planning the probability of sexual transmission on a given sexual programs presume that couples are monogamous, the cost of contact. This probability, Q, is one of the four key behav- the condoms to protect a couple from conception for a year is ioral parameters in the dynamic epidemiological model pre- roughly comparable to the cost of the condoms to protect one sented earlier.38 The government-sponsored programs in of these individuals from STDs for a year, assuming that person several Sub-Saharan African countries have focused only on is in the noncore group. The difference of a factor of five to frequency of partner change, thereby failing to help and ten in average subsidy costs between the condom marketing sometimes stigmatizing individuals who find it impossible programs aimed at STD prevention in Africa and the family to remain monogamous. planning programs in Latin America and Asia is partly due In many of these same countries, private organizations have to the high overhead and substantial input of expatriate informed prostitutes and others of means to reduce the danger personnel during the start-up phase of the former. Perhaps as of each sexual contact through the proper use of a condom. the STD prevention social marketing programs mature, the The cost of such programs is poorly understood. Sketchy average annual subsidy per noncore person-year of protec- information on programs funded by the United States Agency tion will fall closer to the levels observed for the family for International Development through AIDSTECH, Family planning programs. Still, the fact that AIDS prevention Health Intemational, is summarized in table 20-20. The subsidy social marketing programs target high-risk groups may costs of these condom social marketing programs for AIDS control keep their average costs higher than the values attainable range from $30to$45 per yearofprotectionforanoncoreperson by family planning programs that target the noncore. We in 1990U.S.dollars,notincludinganymoneyfromcost-recovery hypothesize that the budgetary (or subsidy) cost of a efforts which remunerates the distribution network. condom social marketing program targeted to the non- For comparison, the estimated cost of a couple-year of core will lie between the value of $5.00 per protected protection by condoms distributed by social marketing in individual observed in the family planning programs and family planning programs in Honduras and Bangladesh was the value of $45.00 observed in the Tanzanian experi- recently estimated to be $14.77 and $6.55, respectively, in ment in peer education. Whatever the cost per individual 1988 U.S. dollars, of which $4.59 and $ 5.90, respectively, user in the noncore, we further hypothesize that the cost HlV Infection and Sexually TrarLsmitted Diseases 489 per protected individual will be ten times larger in the core a developing country which performs large numbers of such group. tests. Assuming another $1 per test for management overhead, the average total cost of such a test could be as low as $2. To COST OF PREVENTING MOTHER-TO-INFANT TRANSMISSION. A obtain this low a cost in a developing country, however, major sequela of gonorrhea is gonococcal ophthalmia neona- requires good management, well-trained and well-managed torum, which can be inexpensively prevented by the applica- technicians, and large volume. In rural areas of developing tion of an antibiotic to the eyes of the newborn immediately countries where none of these conditions obtain, an alterna- after birth. Because 3.5 percent of all live births in some tive test not requiring refrigeration is a rapid serologic test for African populations suffer from this potentially blinding dis- HIV for which the average variable cost is as much as $4 per ease, there is reason to consider a prophylactic application of test, which becomes $5 with the addition of the same dollar antibiotic to the eyes of all neonates as part of a standard for management overhead. Finally, a laboratory expert in maternal and child health package. Laga, Meheus, and Piot Sub-Saharan Africa has informed the first author in confi- (1989) have estimated the cost of applying an antibiotic dence that the average total cost of blood screening in some (either silver nitrate or tetracycline) to the eyes of all newborn Sub-Saharan African capitals can be as large as $10 per test. infants to be approximately $0.10 per newborn for silver ni- Suppose that a perfect test to determine whether a unit of trate and approximately $0.05 per newbom for tetracycline. donated blood is infected with HIV costs $2 per blood sample On the basis of clinical trials of these alternatives in Nairobi, tested. If 5 percent of donors are known to be infected, then it they estimate that silver nitrate prevents 40 of the 47 cases that would require an average of twenty tests to find a single would occur among 100 babies bom of women with a gono- infected unit of blood. By eliminating this unit of blood and coccal infection, whereas tetracycline would prevent 44 of the replacing it at no cost with an uninfected unit, a transfusion 47 cases. The calculation of the dollars per case of gonococcal service has avoided transfusing a patient with HIV-infected ophthalmia neonatorum at different prevalence rates in the blood at twenty times the cost of a blood test. In addition, population is presented in table 20-21. suppose that three-quarters of the people to receive blood With respect to HIV infection, from 30 percent to 50 percent transfusions are HIV-negative before the medical emergency of the children born to HRv-infected women will be infected and subsequently survive the medical problem that caused withHIvatbirth. lfthosewomencouldbecounseledtoprevent them to need the blood. Then the cost of averting an HIV pregnancy or abort their fetus, one HIv-infected child would be infection through blood screening is 4/3 of 20 times the cost prevented for every one to two uninfected children born. of a single test, or $53 per HIV infection prevented. The general Unfortunately, anecdotal evidence suggests that many poor equation is: women, when informed that they are HiV- infected and of the probable consequences, prefer to continue bearing children. Cost per test Our general position regarding mandatory programs-that Cost per HR' Prevalence rate they are unethical and counterproductive-seems particularly Survival rate of transmission recipients apt in the case of hypothetical programs to enforce or strongly encourage abortion of the fetuses of Hl v-infected women. Thus A more complete model of the cost-effectiveness of blood we conclude that the only cost-effective way to prevent verti- screening would relax many of the assumptions made in the cal transmission of HIV is to prevent the infection of the mother above analysis. For example, the tests currently available are in the first place. not perfect, especially under field conditions. They generate false positives and false negatives. Furthermore, infected units COST OF PREVENTING TRANSMISSION BY BLOOD TRANSFUSION. of blood cannot be replaced at no cost but cost as much as $5 The average variable cost of an enzyme-linked immuno- each to replace. Elaboration of these more complete models sorbent assay (ELISA test) for the presence of HIV antibodies with plausible values for the complicating parameters, how- in the blood can be as low as $1 in a well-run laboratory in ever, raises the estimated cost per averted case of HIV infection by only a few percentage points. Therefore, for ease of exposi- Table 20-21. Cost per Case Averted of Eye tion, we present here only the results of this extremely simple Infection from Gonorrhea, by Prevalence of model.9 Gonorrhea among Pregnant Women Shown in figure 20-16 on a logarithmic scale is the relation (1989 U.S. dotlars) between the cost per averted case of lilv infection and the Prevalence of gonorrhea among prevalence rate graphed for two different values of the cost per pregnant women test, $2 and $10. At a prevalence rate of 5 percent and a test Effec- cost of $2, the cost of averting a case of HIV infection is $53 as Therapy Cost Otveness 0.001 0.01 0.10 0.25 in the above example. Note the dramatic effect of prevalence rate on cost. At one extreme, when the prevalence rate among Tetracycline 0.05 0.94 55.56 5.32 0. 13 0.21 donors is as high as 40 percent, as it might be for the donors that an urban prostitute would recruit from among her cowork- Source: Laga, Meheus, and E'iot 1989. ers in a high-prevalence African city, screening can avert a 490 Mead Over and Peter Piot case of HIV infection for only $7. At the other extreme, in areas Figure 20-16. Cost per HIV Infection Averted of low prevalence, blood screening is a very expensive way to by Blood Screening, as a Function of Prevalence avert HIV infection, costing more than $6,000 per case averted. and Test Cost The left-most data point on the graph is at a seroprevalence rate of 4 per 10,000, or 0.04 percent.4 All these costs are costper HIVinfection averted (U.s. dollars) multiplied by five if the cost per test is $10 rather than $2. 1oo,ooo These higher costs are related to prevalence rates in the upper $33,k33 line in figure 20-16. 10,000 $6,667 $0sopertest In any given country the possible sites for blood transfusion 2,667 all have differetnt prevalence rates among their donors and 1 333 different laboratory conditionis leading to different test costs. 1,000 667 The lesson of figure 20-16 is that the cost-effectiveness of blood $2 per test 267 $267 screening for averting cases of Dilv infection will vary a great 100 deal from one of these sites to another. Screening should begin at those sites where screening is most cost-effective and only I0 7 be developed in the least cost-effective sites if the country is unable to avert cases of HIV or to save disability-adjusted 1 I III I II II I I I I V II life-years more cheaply in other ways. 0.01 0.1 1 10 100 BUD:ET FOR PRIMARY PREVENTION OF HIV INFECrION. In table Proportion of donors infected with HIV 20-22 we present the first-year budget of the sixteen Sub- (percent) Saharan African countries which were first to adopt such medium-term plans for AIDS prevention and control. With Source: Authors. only partial minor exceptions, the first-year proposed expenditure has been fully funded, typically out of grants from term plan (NMTP) will augment central government health ex- bilateral donors, both directly and through the World Health penditure by 6 percent. In Zaire and Rwanda the MtTP repre- Organization's Global Programme on AIDS (WHO/oPA). sents a 20 percent increment to central government health In the thirdcolumnoftable 20-22 is the total (recurrentplus expenditure, and in Uganda, Zambia, Tanzania, and Burundi capital) central government health expenditure for those the increment is about 10 percent. countries from which figures are available. The figures in the One way to gauge the magnitude of these MTP budgets is to fifth column show that, in the average country, the medium- relate them to the estimated number of HIV-infected persons. Table 20-22. Donor-Funded AIDS Prevention and Control Budgets in African Countries Central MfTI' as percent of First-year government government health Estimated persons MTP budget per CountrN Year of MIT1U MlTp budget' health expenditures expenditures HIV positive HIV-positive person Uganda 1987 1,508 14,695 10.3 894.3 1.69 Zambia 1988 3,101 34,094 9.1 205.2 15 Zaire 1988 4,363 22,269 19.6 281.8 15 C(entral African Republic 1987-88 1,599 - - 54.3 29 Rwanda 1987 2,922 13,680 21.4 81.5 36 Congo 1988 1,727 - - 45.5 38 Tanzania 1987 3,945 51,250 7.7 96.6 41 Mozambique 1988 1,788 - - 43.5 41 Cameroon 1988 1,841 86.593 2.1 33.2 55 Kenya 1987 2,940 120,335 2.4 44.5 66 Burundi 1988 1,719 14,925 11.5 15.0 115 Zimbabwe 1988-89 3,799 128,323 3.0 30.9 123 Botswana 1987 150 32,369 0.5 0.9 167 Senegal 1988 1,443 - - 2.9 498 MauritILus 1988 242 26,045 0.9 0.1 2,420 Ethiopia 1988 3,137 45,480 6.9 0.1 31,370 Mean 2,264 36,879 6.0 114.4 20 Note Budget and expendirures in thousands of 1987 U.S. dollars. - Not available. a. From first meditim-rerm planiing documents prepared by individual countries with WVHO/( ,PA assistance. S0ouTrce Bongaarts and Way 1989; World Banki 1989. See also note a, above. HIV Infecuon and Sexually Transmitted Diseases 491 The fifth column of the table contains the estimated number of trained staffhave been spread too thinly trying todo all these of Fi-rs in each country in 1987-88. From the data in the second activities at once. We hope that analyses such as the present and fifth columns, we arrive at that in the sixth column, the one will help guide the prioritization of activities and expen- MTF budget per HIP, which ranges from less than $2 in Uganda diture so that national AIDS programs can achieve greater to more than $31,000 in Ethiopia. health benefits (that is, save more disability-adjusted life- The appropriate budget allocation to a national AIDS control years) for their allotted budget. program depends on the cost of preventing a case of HIV infection in that country and on the alternative possible uses Cost-Effective STD Prevention Strategies for the same budget resources. For an international agency such as WHO/GPA, the alternative expenditure for grant resources not Primary prevention of STDs requ ires the interruption of one or spent in country X is expenditure on preventing mIIV infection more of the modes of STD transmission displayed as column in country Y. Such an agency's goal might reasonably he to headings in table 20-13. Resources are not so plentiful, how- minimize the number of cases of HIV infection worldwide by ever, that countries can afford to press forward on all possible maximizing the number of prevented cases each year. With a interventions simultaneously.41 Instead, choices are required. fixed global grant budget, this goal could be achieved only by In this section, we review the available primary prevention equalizing the number of cases prevented per grant dollar interventions and recommend some as likely to be more cost- across couLntries. effective than others. We retain from previous sections of this The cost per case of HIV prevented will vary across countries chapter the lesson that preventing a case OfSTD in the core will and within a country across interventions. As long as HIV save many more subsequent cases than would preventing a case prevalence rates are increasing, however, the number of people in the noncore. at risk of new infection will be roughly proportional to the In the first row of table 20-13, we present the voluntary number of people capable of infecting others in that country, behavioral changes, which can be stimulated by appropriately that is, the numberofHIPs. Thus a rational allocation ofa global targeted information, education, and communication pro- grant budget for AIDS control should approximately equate the grams. These IEC programs typically affect the individual de- MTP dollars per HIP actoss countries. Some variation in budget mand for a protective voluntary behavior, such as use of a per IttI would remain because of the different institutional condom or avoidance of a transfusion. They do so either by structures-and, therefore, the different costs of interven- providing information or by changing preferences; sometimes tions-across countries. Still, the large variations in MTP they do both. budget per HIP revealed by the sixth column of table 20-22 In contrast to voluntary behavior modification through IEC, demonstrate that the allocation of resources across these Sub- mandatory and passive behavior modification typically occur Saharan African countries is quite different from the preven- through changes in the availability or price of a protective tion-maximizing allocation. Even ignoring the two extreme behavior or product, that is, on the supply side ofa market. For countries, Uganda and Ethiopia, the MTh budget per HIP is more example, either an enforced prohibition of prostitution or the than thirty times larger in Mauritius and Senegal than in provision of training and jobs for urban single women would Zambia and Zaire, a multiple too large to explain by differences reduce the supply, raise the price, and therefore reduce the in absorptive capacity. It is hard to escape the conclusion that frequency of commercial sex. the MTP budgeting process was driven by random political and Although both mandatory and passive behavioral changes other factors which bore little relation to a rational allocation typically operate on the supply side, their mechanisms and of resources across countries. effects are clearly distinct. Mandatory programs (such as the There is a prior question, however-whether these large prohibition ofprostitution) involve the enactment of laws and incremental expenditures can be effectively absorbed by the regulations which require the health-promoting behavior on countries in questioni. Because the health expenditure of cen- penalty of fine or imprisonment. Examples outside the area of tral governments rarely fluctuates by more than 2 or 3 percent- STD prevention include laws on speed limits and regulations age points in any given year, and then the fluctuation is usually concerning pollution control. To be effective, such laws and spread over all the many preventive and curative programs in regulations must be enforced, which requires expenditure on the ministry, it is not surprising that governments have had government regulators. Furthermore, the regulators must be substantial difficulty launching such large incremental expen- well paid and closely supervised to the point that their subom- diture programs, amounting to as much as 10 to 20 percent of ing is more costly to the would-be transgressor than is compli- their health budgets, for a single disease. Although the key ance with the laws. All these difficulties imply that mandatory resource constraint over the short run has proved to be trained behavior modification is best suited for short-term emergency manpower, an additional limit to the absorptive capacity of control of easily observed, politically unpopular behavior. national AIDS programs has been a lack of clear guidance on Passive programs, in contrast, are typically slow to start but prioritizing interventions. The medium-term plan has typi- can be profound in their long-run influence on private deci- cally enumerated a long list of desirable activities, including sions. If the government policies which support the passive most of those listed in table 20-13, without ranking them by program (such as improved job opportunities for women) are effectiveness or urgency. The result is that the small numbers perceived to be permanent, they can induce individuals to 492 Mead Over and Peter Piot choose different residential locations, careers, and family sizes demonstrated earlier that African cities with low female-to- and thus can profoundly affect the details of their lives. male secondary enrollment ratios also had high rates of STD The choice of a cost-effective package of primary prevention infection. To the extent that primary infertility caused by STDs interventions must be conducted in two dimensions: choices prevents woomen from marrying or remaining married, pro- are required both within modes of transmission (the columns grams to prevent or quickly cure STDS will help women avoid of table 20-13) and between modes of transmission (the rows the stigmatization and rejection caused by infertility-and of the table). We begin with the sexual transmission mode, thus avoid recourse to prostitution. Education is a long-run which is the sole mode of transmission for most CSTDs and strategy that will increase the woman's contribution to the accounts for more than 90 percent of all adult cases of Iiiv development process, induce her to have fewer, healthier infection in pattern I countries.42 children, and improve her bargaining position in sexual rela- tionships. Such a policy is an appropriate one to counter a slow, long-run epidemic like HIV infection that is exacerbated by low SEXUJAL TRANSMISSION AND MAND)ATORY' PROG'RAMS. The pri- femaleeuain vacy of sexual practices and the primacy of the sex drive make femewhation. sexual behavior one of the least-suited activities for mandatory A somewhat shorter-mn policy that has so far received little behavior modification. Duting World War II the United States or no attention in national AIDS control programs is the cre- conducted an excellent example of a failed mandatory policy ation of jobs for urban women. In a sampling of prostitutes in one Nigerian town, 67 of them reported that they would cease when it attempted to enforce the prohibition against prostitu- prostitution (an average of 4.4 paying clients per day) if they tion in order to halt the spread Of STDS. could find a job paying the equivalent of $15 a month (Wil- [After] thousands of women were institutionalized or de- liams, Hearst, and Udofia 1989). A possible example of an tained [and] jails became overcrowded, Ness arranged for inadvertently beneficial effect on STD transmission of job cre- the creation of some 30 "civilian conservation camps" for ation for women is the "enterprise zones" of northem Mexico, young women as a means of relieving some of the pressure where many of the young female employees are said to be on existing facilities.43 [But] despite the incarceration of grateful for these low-wage jobs as a welcome alternative to thousands ofprostitutes, it soon became clear that this could their former lives as prostitutes (personal communication from not in itself solve the venereal disease problem. Indeed, the Sally Stansfield 1989). effect of closing the red-light districts was sometimes disap- Although the immediate effect of alternative female em- pointing to military officials. Increasingly, army physicians ployment in urban areas of pattem 11 countries would be to reported that prostitutes constituted only a minority of the reduce the supply and thereby raise the price of commercial soldiers' sexual contacts. In the Third Service Command, sex for men, the longer-run effect would be to attract addi- for example, only 19 percent of the infections could be tional women to these urban centers. Still, on the basis of the attributed to prostitutes; in other communities even fewer statistical evidence presented earlier, we believe that this, too, infections could be so traced (Brandt 1987, p. 167).4 would contribute to the control of STDS. Thus, on practical grounds alone, we recommend against man- datoa pogrms wichrelon egalproibiionsFurherore SEXUAL TRANSMISSION AND VOLUNTARY PARTNER NOTIFICATION. datory programs which relyon legal prohibitions. Furthermiore, Although passive behavioral change seems to us the most the consequences of such programs for civil liberties are likely to b unccetabl todemcratc scieies45 promising long-run strategy by which to control sexual trans- to be unacceptable to democratic societies, mission, certain voluntary strategies, when tightly targeted at the core group, may also be cost-effective. As the first among SEXUJAL TRANSMISSION AND CHANGED "PRICES" Setting aside these we designate voluntary partner notification, by both mandatory programs as unlikely to be effective, we tum to patient and provider referral. This method has proven increas- policy choices categorized as either voluntary or passive. We ingly effective in the control of CSTDS, especially after the believe that too little attention has been given to passive availability first of salvarsan (in 1909) and then of penicillin methods of modifying behavior in relation to that given to (in 1943) made it possible to offer a cure to the sexual contacts voluntary and mandatory ones. Among those listed in table of the index case. Some European countries, however, oppose 20-13, the potential effectiveness of subsidies to condom partnernotificationprogramsfortheCsTmsontheground that, distribution through private channels has been demon- even when voluntary, they invade privacy. strated by a social marketing effort in Kinshasa, Zaire, which Recently, some American states have been experimenting increased the sales of condoms in 110 pharmacies from with partner notification in cases ofHIV infection. Toomey and 19,000 per month to 300,000 per month in only ten months Cates' concise overview of U.S. practices points out that (Lamptey and Goodridge 1991). Still, to our knowledge, the partner notification is far more effective than HIV screening of possibility of taxing alcohol served in public places in order a population of STD patients in identifying new cases of HIV to increase the cost of sexual partner change has not been infection (1989). For example, compared with seroprevalence studied. rates for IDIV of 2.7 percent and 6.4 percent among STD patients More promising still are longer-run programs to improve the in Virginia and Florida, seroprevalence rates among notified marriage, education, and job opportunities for women. It was partners were 13.5 percent and 25.1 percent, respectively. The HIV Infection and Sexually Transmitted Diseases 493 cost per index case in Virginia was $50 if the patient was asked Africa are rather disappointing (Temmerman and others to refer, and $64 if the public health staff conducted provider 1990). For all these reasons, we do not recommend partner referral. In Colorado, researchers in a similar study using dif- notification as a primary strategy for iiiv control at the mo- ferent methods found costs of $50 per index case without ment, but we believe that pilot projects should be supported to provider referral and $33 for each partner notified by the address the issues discussed. providers.46 Because these figures, $33 to $65 per index case, consist largely of the cost of labor, they can be deflated to SEXUAL TRANSMISSION AND IEC PROGRAMS. The most cost- figures that would apply in poor pattern 11 countries by express- effective IEC program to inform the general public about STDs ing them as 0.5 percent of GNP per capita.47 If partner notifica- is targeted not at individuals at risk but at journalists. Using tion programs could be operated in pattern 11 countries at even STF S as an excuse to sell newspapers by openly discussing sexual 1 or 2 percent of local GNP per capita for each index case, the behavior, journalists and their editors have become willing resources currently available for AIDS control could mount collaborators with the public health authorities in STD control significant partner notification programs. at least since President Franklin D. Roosevelt's surgeon gen- These American successes with partner notification were eral, Thomas Parran, broke taboos in the United States by achieved before it was possible to offer any treatment for publishing an explicit article on STDS in the Reader's Digest in nonsymptomatic HIV infection. Now that AZT has been found 1936. More than 125 newspapers, many other magazines, and to retard the development of the disease, these American even the staid monthly the Ladies' Home Journal picked up the programs will achieve even greater success. Although the high campaign, providing a wealth of free IEC (Brandt 1987, p. 141). price of AZT will prevent its immediate wholesale provision to In 1988 the British campaign against AIDS obtained 8 million asymptomatic infected people of the pattern 1I countries, this pounds of free publicity in one month alone while spending technological advance nevertheless holds out hope for im- only 3 million on its own campaign (United Kingdom, Na- proved partner notification programs in either of two distinct tional Health Service 1988). scenarios. Any visitor to or resident of a developing country with an First, suppose that the annual cost of a temporarily effective HiV epidemic is aware of how enthusiastically journalists there antiviral medication remains as high as it is now, several have picked up and reported any information about HIV thousand dollars per year. In this case, a public health program infection, whether it is true or false. With respect to the media, in a poor developing country could afford to buy only a few the challenge for the national AIDS control programs in these such treatments next year. If the program could convince the countries is threefold: to prevent the publication of misinfor- public that these few treatments are to be allocated fairly to a mation; to continue to provide new, interesting press releases few of those who are actively cooperating with prevention expressed in lay language and accompanied by vivid graphics; efforts, then the availability of even a small chance for a and to broaden the discussion beyond HIV and AIDS to other SIDs. temporarily effective treatment could serve as a powerful in- In our opinion, efforts to address publicly financed IEC cam- ducement for behavioral change.48 paigns to the entire population or even to all sexually active But the price of antivirals is unlikely to remain constant. If people are likely to be markedly less cost-effective in slowing it falls in price as fast as penicillin did after its discovery, STD epidemics than would the passive and voluntary programs antivirals will be 100 times cheaper by the mid-1990s (Brandt described above. Evidence from a variety of national settings 1987, p. 170). At $40 per annual treatment, antivirals cannot confirms that such IEC campaigns may improve knowledge and be denied to the poor, severely affected pattern II countries. If change attitudes but usually produce little if any behavioral they are tightly controlled by the government so that access to change (Warner 1983; Hornik 1988). The IEC programs that them is only through a public health clinic which aggressively are targeted at a specific subset of the core group may be a useful pursues partner notification, then antivirals can substantially addition to the efforts described above. increase the effectiveness of partner notification efforts. The For discussion of IEC campaigns, it is useful to cross-classify incidence of all STDs, including HIV, may be strikingly reduced. the population in two dimensions by "target group" and "access Because partner notification may have significant ethical group." We define target groups as groups of individuals with and practical problems and undesirable social and biological homogeneous risk behavior, whose behavior change is the side effects, it may not be a feasible or recommendable ap- object of some STD intervention listed in table 20-13. Access proach to HIV control in many societies. Some issues which groups are defined according to the means or channel by which have to be resolved before initiating such a program include we can gain access to them for the communication of lEC and possible stigmatization of those notified, violation of privacy, training messages. In table 20-23 we present a classification of potential for misuse by index cases, rejection of notified target groups as rows and access groups as columns. women by their partners (sometimes leading to prostitution), The target group for behavioral change can be either the the need for a large corps of counselors and for medical facilities group at risk of infection or the group that controls the risk of capable ofmonitoring the biological side effects ofantiretrovi- infection. Information, education, and communication pro- ral therapy, and an assurance of long-term sustainability and grams would be addressed to the former and training to the continuity. Preliminary results from counseling and partner latter. In the rows of table 20-23, we identify three subgroups notification efforts with HIV-infected pregnant women in among those at risk: sexually active persons, health care con- 494 Mead Over and Peter Piot sumers (at risk from transfusion or contaminated needles), and or "low." As an example of how to read the table, consider an health care providers (at risk from accidental sticking by IEC program which attempts to change the behavior of prosti- contaminated needles, and so on). In other rows of the table, tutes (the target group) by addressing all the clients of bars (the we classify those controlling the risk of infection as health care access group). Because such a program is likely to reach a high providers or govemment officials. These latter groups admin- proportion of all prostitutes, it scores an "H" for "high" on ister and enforce mandatory or passive interventions. The coverage in the appropriate cell of table 20-23. A program target groups are, of course, of very different sizes. Although addressed to the access group "saloon clients," however, will the relative sizes of the groups will vary from one urban setting necessarily be diluted in its effect on prostitutes because of their to another, it is useful to note typical or "notional" relative sizes small share in the population of such clients. Therefore this for the groups in table 20-23.49 program scores an "L" for "low" on concentration in the same In order to change the behavior of either the group at risk cell. If the goal of a program is to change the behavior of or the group controlling the risk, an EC' or training interven- prostitutes, consideration should be given to reaching them tion must communicate, either through the mass media or through an STD clinic, especially one which might have a face-to-face. The columns of table 20-23 identify subsets of the higher-than-average concentration of prostitutes, or in their population reached by each of a selected set of access routes. residential neighborhood, in which we estimate both the cov- Like the target groups, the access groups differ in size. Because erage and the concentration to be "medium." If we assume that the size of a specific access group can be varied, however (that the goal of the IEC program is to slow the epidemics, preventing is, by calling a meeting of unmarried workers in a given factory, as many case-years of illness as possible per dollar of the IEC or by targeting a media campaign to adolescents), it is not budget, the candidates for programs which hold the most useful to associate even a national relative magnitude to each promise are those which are targeted at part of the core group, access group at this level of aggregation. have high coverage of and concentration on the targeted An [EC or training program aimed at a given access group group, and provide a message which can be expected to interest will typically reach only some of the members of any target and persuade many in the target group. group but will also reach other individuals outside the target Consider, first, prostitutes who are part of the core. If we group. Define the "coverage" of a specific target group by an have guessed correctly on the entries in the first row of table intervention as the proportion of the target group reached. 20-23, none of the access groups listed will provide both high Clearly, an intervention is to be preferred, other things being coverage of and high concentration on this group. Further- equal, if it reaches a large rather than a small proportion of the more, experience with female prostitutes in Nairobi, Kenya target group.50 It is possible roughly to rank the coverage that (Ngugi and others 1988); male bar workers in Bangkok, Thai- messages aimed at each access group would achieve for each land (Sittitrai 1990); and other groups has shown that even target group. For example, IEC efforts at bars or STD clinics may face-to-face contact fails to change the frequency of safe sex achieve relatively high coverage of female prostitutes, whereas enough to slow the spread of the HIV virus. The most promising IEC efforts at barracks, schools, and offices would do much less results from prostitutes came instead from peer training pro- well on this measure. grams, in which trusted fellow prostitutes are the source of the Still, even if a message covers all of a target group, its effect persuasive message. A program like this in Cameroon in- on that group will be reduced if it also reaches large numbers creased the reported use of condoms (at least half the time) outside the target. For example, a message on the desirability from 28 percent to 72 percent over a twelve-month period of using condoms with multiple sexual partners, when routed (Monny-Lobe and others 1989). Previous similarsuccesses had through the mass media, will lose credibility and plausibility been reported in Ghana (Lamptey and others 1988) and Mex- to the prostitute, because she and her peers are a small percent- ico. The medium concentration that may be achieved in some age of those receiving the message. Define the "concentration" red-light districts argues that these residential neighborhoods of a message on a target group as the percentage of all message be the venue for these meetings. recipients who are in the given target group. Again, the The data in table 20-23 indicate that the clients of prosti- concentration of a message routed through a given access tutes are also likely to be hard to reach with high concentra- group at a given target group can be roughly ranked. In the tion, because they do not form a large percentage of any of above example of a mass media campaign, we argue that the these access groups. Because clients are a largely clandestine small concentration makes the campaign less effective, other group in many countries, peer training programs are also likely things being equal. By focusing a message aimed at prostitutes to fail. Hence, we recommend against targeting them. via an access group defined by the clients of an STD clinic or by Like prostitutes, sexually active adolescents may respond a red-light residential neighborhood, however, the message's relatively well to peer counseling and might otherwise be hard concentration can be substantially increased. Thus concentra- to reach. Because that portion of adolescents who are attend- tion must be added to coverage as a ciesirable attribute of an ing secondary school or a university are a particular critical IEC or training program. national resource in pattern 11 countries, we recommend that In the cells of table 20-23 we have tentatively classified the they be targeted with particular vigor. use of each access group to reach each target group as having In contrast to the above groups, a message aimed at bar a coverage and a concentration which are "high," "medium," patrons is predicted to cover a large proportion of adults with Table 20-23. Coverage and Concentration of IEC and Training Programs, by Target and Access 5roup Target group Access group Prisoners, soldiers, students, or workers Clients and workers Audience of in prisons, barracks, Sexually active Patients at general at family planning Residents of a Group Size mass media schools, or work sites adults at bars health care clinics and STD clinics neighborhood Groups at risk From sex Prostitutes 20 Low/low Low/low High/low Medium/low High/low Medium/medium Clients of prostitutes 100 Medium/low High/low High/low Medium/low High/low Low/low Sexually active adolescents 2,000 Low/low Medium/low Low/low Medium/low Medium/low Low/low Adults with multiple partners 2,000 High/medium High/medium High/high Medium/low Medium/high High/low From transfusions 10,000 Medium/high Medium/high Low/medium High/high Low/medium High/high From needles 5 High/low Low/low Medium/low High/medium High/high Low/low Group controlling risk Health care providers 5 High/low Low/low Medium/low High/medium High/high Low/low Governmentofficials 5 High/low High/medium Medium/low Low/low Low/low Medium/low Note: Coverage is defined as the proportion of the target group reached by a message. Concentration is the proportion of those reached who are in the target group. "High" is defined as greater than two-thirds; "medium" is defined as one-third to two-thirds; "low" is defined as less than one-third. The estimated coverage precedes the estimated concentration; thus, the notation "high/low" means that using the indicated access group to target the indicated target group will have a high coverage but a low concentration. Source: Authors' development of Homik 1989a and 1989b. 496 Mead Over and Peter Piot multiple partners and to do so with high concentration. We Table 20-24. Cost per Discounted Disability- recommend targeted lEC at this core group. Adjusted Life-Year Saved for a Condomn Subsidy We explicitly assign a low priority to publicly financed IEC Intervention: Sensitivity to Cost per Person Year campaigns designed to change the sexual behavior of noncore of Protection and Core versus Noncore Strategy groups, such as the health care consumers of table 20-23. (1990 U.S. dollars) Although some people in these other, lower-risk, populations Cost per year of protection may become infected with HIV through sexual Intercourse, Disability-adjusted $50 in core $450 in core public programs are unlikely to deliver more effective messages life-years saved group group than the privately financed media, and each public dollarspent perperson-year $5 in noncore $45 in noncore here could be spent with greater effect on any of the programs Disease (group) of protection group group discussed above. Chancroid It is extremely difficult to quantify all the above considera- Core group 80.9 0.62 5.56 tions. Still, the epidemiological model presented earlier can be Noncore group 1.1 4.5 5 40.91 used to arrive at a rough estimate of the relative cost-effect'ive- Chlamydia ness of a particular kind of IEC program, one that is designed to Core group 196.6 0.25 2.29 increase the frequency of condom use. We assume that con- Noncore group 4.2 1.19 10.71 doms are distributed with the assistance of a condom social Gonorrhea marketing campaign which subsidizes [EC efforts that vary, Coregroup 275.2 0.18 1.64 depending on the campaign's target group. For the noncore, Noncore group 5.2 0.96 8.65 the social marketing resources would be spent on the public media. To target the core, the campaign would employ peer Nocv rithou p 56.6 0.88 7.95 counselors from among the high-risk population. In table Core group 19.9 2.51 22.61 20-24 we present the estimated cost per disability-adjusted life-year saved by such an intervention depending on which CIor Witk uk2ers population, the Core group 156.0 0.32 2.88 STDs are present in the population, on the cost per person-year Noncore grokup 21.5 0.23 2.09 of the social marketing campaign, and on whether the program is targeted at the core or the noncore. Syphilis As predicted in the discussion of table 20-19, the immense Core group 384.6 0.13 1.17 . . . . . ' . . . ~~~~~Noncr ru . 0.52 4.64 increase in effectiveness attained by protecting core individu- ncore group 9.7 als rather than noncore is large enough to offset our assumption Source: First column from table 20-19; cost per person-year of protection that the cost per person-year of protection will be ten times from table 20-20; janowitz, Brant, and Fried 1990. larger in the core. In table 20-24 we show costs per disability- adjusted life-year below $1 for interventions targeted to the In table 20-21 we presented calculations of the cost per core group in the presence of certain epidemics. These costs case of gonococcal ophthalmia neonatorum averted through compare favorably with those of other interventions. the preventive application of silver nitrate or tetracycline. When the prevalence rate of gonorrhea among pregnant PREVENTION OF TRANSMISSION FROM MOTHER TO INFANT. The women is above I percent, a case of gonococcal ophthalmia World Health Organization estimates mother-to-infant trans- neonatorum can be averted for less than $6, if silver nitrate mission as the second most important transmission mode for is chosen. Assuming that each averted case saves the af- HIV infection in pattern 11 countries, accounting for up to 11 fected child one disability-adjusted life-year, this preventive percent of all cases. Obviously, prevention of Hlv-infection in intervention buys disability-adjusted life-years for less than women also prevents these women from infecting their infants $6 when prevalence rates are high. Therefore, we recom- at no additional program cost. The question is whether addi- mend universal eye prophylaxis, without prior screening of tional resources should be reallocated to prevent infected the mother, in all areas in which gonorrhea prevalence rates women from conceiving or from giving birth. are above 1 percent. As part of the program of notifying sexual partners, we Finally, to prevent congenital syphilis, we recommend recommend that Hiv-infected women, especially if they are universal screening ofpregnant women for serological mark- pregnant, receive extra, gender-specific, voluntary counseling ers of syphilis, followed by effective treatment of infected which focuses on the risks of having seropositive babies and women. 6 the costs of their care.54 Recent experience from Kenya and Zaire suggests that the fertility rate of women who know they PREVENTION OF TRANSMISSION BY BLOOD AND BLOOD PRODUCTS. are infected with HIV is at least as great as that of uninfected Because WHo estimates that only about 6 percent of all cases women, despite counseling about their infection status and the of HIV infection in pattern 11 countries are caused by transfused risk of perinatal infection (Temmerman and others 1990). In blood or blood products, the case for diverting resources from addition, the illegality of abortion in many countries is a other prevention programs to this one must be based on sound serious impediment to such a strategy. cost-effectiveness analysis. We consider first mandatory then HIV Infection and Sexually Transmitted Diseases 497 passive and voluntary policy interventions. Then we estimate usling simple rapid tests should be considered (Laleman and the cost-effectiveness of blood screening Linder various others 1992). The choice between these institutional alterna- circumstances. tives depends on the organization and capability of the local The high incidence and severe consequences of malaria health system. and anemia in many developing countries are discussed in The second priority for the prevention of HIV infection from other chapters of this collection. Blood transfusion has often blood transfusion is to develop a corps of low-risk, voluntary been administered to these patients with little curative donors who repeatedly donate blood in a given community. effect. Now, in pattern II countries, sLich patients incur the Such a policy will be more costly to administer than the risk of HIV infection in addition to the other risks of these laissez-faire ad hoc policies which required the patient's family diseases. to recruit a donor, but they will still be less costly and more In thearea of blood transfusion, wedepart from ouraversion cost-effective than would be the universal screening of all to mandatory policies to recommend one: a prohibition of the transfused blood. transfusion of unscreened blood except as a life-saving mea- Any effort to reduce the incidence of illnesses requiring sure. According to Fleminlg (1988), such indications include blood transfusion will in a pattern 11 country also reduce the the following: incidence of HIV infection. When this benefit is added to the intrinsic merits of these other programs, some of them will * Profound anemia (hematocrit less than 4 grams per assume even higher priorities in national health promotion deciliter) with incipient cardiac failure strategies than they would on their own merits. Such health * Severe neonatal jaundice (sernim bilirubinl greater than problems include trauma (especially from road accidents), 300 micromol/L malaria, anemia, and adverse birth outcomes. * Blood loss of more than 25 percent of total volume when A quantitative analysis of the cost-effectiveness of blood the blood pressure and oxygen cannot be maintained by screening should consider the cost per disability-adjusted life- plasma expanders. year saved of donor deferral, donor recruitment, and training physicians to use more conservative criteria for transfusing. Each pattern 11 country should develop and enforce its own set Unfortunately, costs of these varied strategies are not avail- of transfusion guidelinies. Becatise the necessary enforcement able. It is possible, however, to calculate the cost per DALY saved effort is focused at secondary care institutionMs, it can be im- by blood screening under the simple assumptions used to plemented in each such facility by a committee of senior construct figure 20-16. From table 20-16 we know that avert- physicians. The enforcement of this bani should be the top ing a case of HIV infection saves about 359.6 DALYs if the person priority program for the prevention of infection by blood is in the core group and only 54.6 DALYs if the person is in the transfusion. noncore. Table 20-25 is constructed to combine these figures If additional public resources for transfusion remain after the with the costs per case of Hiv averted in figure 20-16, producing above two programs have been implemented. the additional costs per DALY saved. Note that the argument for HIV screening security which comes from a blood-screening program can be in low-prevalence populations of poor countries is weak. considered. When efficiently performed at a blood bank, blood tests now cost less than a dollar in Africa, whichi is less than PREVENTION OF TRANSMISSION BY SKIN-PIERCING INSTRUMENTS. 25 percent otf the cost of a unit of blood in most African Fewer than 3 percent of all cases of fiIv infection have been capitals. Furthermore, some laboratories are experimenting attributed to needles and skin-piercing instruments in pattern with the use of a single test to screen a vial of blood in which 11 countries. Furthermore, the costs of significant reform in this the samples of several individuals are pooled. With some of the area seem high. We recommend that efforts here be restricted available blood tests, a negative test result on the pool of five to a modest IEC campaign designed to stimulate health care separate samples ensures that all individual samples are nega- consumers to insist on brand new or properly sterilized needles. tive with a probability above .95. In areas of poor countries In addition, health care providers should be educated about where blood bank facilities are unavailable or undependable the importance of properly sterilized needles and, for their own and prospective donors must be screened on the spot, only the protection, about safe methods of discarding them. more expensive ($2.50-$4.00) rapid tests are useful for screen- ing blood, and pooling is impractical (Laleman and others Case Management and Secondary Prevention 1992). In this case, the screening will double the cost of blood. These considerations lead us to recommend that, once trans- By treating CSTDS, further spread of the infection is prevented, fusion guidelines and donor recruitment programs are in place, and the risk of complications and sequelae in patients with an a blood-screening program should be largely self-financing STD is reduced. through patient fees, with appropriate sliding fee schedules to accommodate the indigent. Because user charges are already Goals of Case Management the de facto policy for blood screening in many African capi- tals, implementing such a policy will not be difficult. Both a Treatment of STDs benefits both the infected individual and, centralized blood bank system and a decentralized strategy by reducing the reservoir of infected persons, the community 498 Mead Over and Peter Piot Table 20-25. Cost per Discounted Disability- may be a critical element in the prevention of complications Adjusted Life-Years Saved of Blood and sequelae through early diagnosis and treatment, as well as Screening in the reduction of secondary cases ofSTD by the patient. Delays Prevalence rate of HR' infection among by men of several weeks and by women of several months blood donors before seeking treatment for an STD are not unusual. Health- Sexual acnvit group seeking behavior is not only a function of attitudes toward of proposed blood Cost of disease and sex but also of the accessibility and quality of health recipient blood test 0.001 0.01 0. 05 0.25 care facilities dealing with STDs. These are often of poor quality. Core group $2 7.42 0.74 0.15 0.03 understaffed, and lacking even the most essential diagnostic $10 37.08 3.71 0.74 0.15 tools and drugs. Whereas STDs can usually be managed at public primary health care facilities, patients often prefer to go to Noncore groL4p $2 48.84 4.88 0.98 0.20 more expensive private physicians who for the most part are _______244___20___24___ 42__ 4__ 88__ 0__98_ not offering a better standard of management. Training of Note: See figure 16 and discussion. Costs include estimated overhead cost health care workers in STD is also grossly inadequate in most of managing blood screening service. From Table 20-26 we assume that each medical schools. case of HIV averted in the core group saves 359.6 1ALYs while each case axverted in the noncore saves 54.6 I)ALYS. Source: Authors' calculations. DIAGNOSIS. A core problem in STD case management is the difficult etiological diagnosis of most syndromes, particularly of uninfected people. Traditionally, early diagnosis and treat- in women. Thus, both gonococcal and chlamydial infection in ment (secondary prevention) have been the cornerstones of the female are diagnosed by isolation of the bacterial agent, programs for the control of bacterial STDs, including syphilis, the simpler microscopic examinations not being adequate. gonorrhea, and genital chlamydial infection. As effective anti- Culture techniques are expensive and technically demanding viral chemotherapy becomes available against HIV infection and are beyond the competence or fiscal possibilities (reagents and genital herpes, secondary prevention may become an must be paid for in hard currency) of most laboratories in increasingly important aspect of the control of viral STDs. In developing countries. Culture-independent techniques (en- addition, it will be possible to increase greatly the life expec- zyme immunoassay, immunofluorescent assays, DNA hybridiza- tancy of patients with HIV infection. The overall goals of tion) for the diagnosis ofbacterial and viral STDs have become treatment of STDs are the following: available recently but are expensive and often still lack sensi- tivity and specificity. Clinical methods of diagnosing STDs by * To cure the actual disease means of simple algorithms are being increasingly used (see * To prevent complications and sequelae appendix 20C). Thus far, however, these have failed effec- * To prevent transmission of the treated disease tively to diagnose gonococcal and chlamydial infections in * To reduce the efficiency of HIV transmission women. Principles of Case Management CASE FINDING AND SCREENING. Case finding and screening have traditionally been an important component of STD con- The STDs caused by bacterial agents are all fully treatable by trol programs. Their objective is to identify individuals who specific antibiotics. Still, case management of STDs is often of are infected, but are not symptomatic, in order to treat them poor quality and ineffective in many, if not most, countries of before they develop complications and sequelae. As weapons the world. Furthermore, even the best treatment of viral STDS for actively combating STDS, these two strategies must be com- remains purely symptomatic or marginally effective and very pared with that of simply treating persons without diagnosis. expensive (for example, herpes). Problems in case manage- In appendix 20C we analyze these two options for all the STDs ment and secondary prevention of STDs are listed below. considered in this chapter and find that, for cost ranges rele- vant to developing countries, screening is rarely more cost- * Health-seekingbehavior(delayofdiagnosisandtreatment) effective than treating without a test. Furthermore in those * Accessibility and quality of health care facilities cases in which screening is cost-effective, clinical diagnosis is * Etiological diagnosis of syndromes (inadequate labora- almost always more cost-effective than laboratory tests (see tories, lack of simple, inexpensive diagnostic tests) appendix 20C for details.) * Antimicrobial resistance (gonorrhea, chancroid) One use of screening not analyzed in appendix 20C is for * Partner referral the prevention of congenital syphilis by screening pregnant women for serological markers for syphilis. By treating infected Current case management guidelines are summarized in women, congenital syphilis is prevented in the newborn. Case appendix 20C. finding is also used for gonorrhea control in populations where the prevalence of this infection is reasonably high. It has been EARLY TREATMENT. Improving health-seeking behavior is a used with success in STD control programs in prostitutes in much neglected aspect of management strategies for STDs but various parts of the world (Tuliza and others 1991). HIV Infection and Sexually Transmitred Diseases 499 EFFECTIVE THERAPY. Because of the development of anti- the foreign resources include the cost of drugs and of diagnostic microbial resistance of N. gonorrhoeaestrains, mainly in South- tnaterials. Domestic resources include the personnel who de- east Asia and Sub-Saharan Africa, treatment of gonorrhea has liver the services and the buildings in which they work. In become more complicated and more expensive. Treatment appendix 20C we assemble the available information on the guidelinesforsTDs and adequate trainingofhealth care workers cost of STrD treatment in order to arrive at an estimate of the in STI) management are not available in most developing cost per effectively treated case of STD. In table 20C-3 we countries. present sensitivity analysis of the cost per effectively treated Although not as such part of individual therapy, treatment case with respect to two key parameters, the cost per clinic- of primary (the source contact) and secondary sexual contacts hour and the prevalence rate of the STD in the population. The (individuals exposed to the patient) through partner notifica- first of these parameters varies with the GNP of the country- tion is an essential part of case management, aiming at reduc- relatively rich countries will have more highly paid medical ing the reinfection rate in the patient, limiting the spread of workers and more expensive rental rates for their buildings. the infection in the popLilation, and decreasing the rate of Contrarily, a high prevalence rate decreases the cost per effec- complications and sequelae in these contacts. This requires tively treated case by ensuring that few resources are wasted on considerable resources in time and personnel and is heavily people who are not really sick. influenced by the cultural-behavioral environment. Its cost- Two strategies can be defined. The first is the one usually effectiveness remains to be explored in a cdeveloping country recommended by medical experts: the health care provider context. applies a diagnostic procedure. He or she might take a speci- men and examine it with a microscope or culture it in a (COUNSELING. Finally, because sTi) patients may put them- laboratory in an attempt to diagnose the etiologic agent accu- selves again at risk for STD-and because a small group of core rately. Or the provider might simply examine the patient, take transmitters is directly and indirectly responsible for the ma- a medical history, and apply a decision rule or "health care jority of cases ofSTD--counseling aiming at behavioral change algorithm" to decide whether, and how, to treat. Each of these should also be part of the STD case management. three diagnostic procedures has a different degree of precision and a different cost. We call a strategy that applies one of these HIV INFECTION. Case management of patients with HIV infec- procedures a "test-before-treatment" strategy. tion includes treatnent of the associated opportunistic infec- The second strategy is an extreme form of presumptive tions in patients with AIDS, therapy for HIV infection itself, treatment. In this method, the provider prescribes a broad- and appropriate social and psychological support. Whereas spectrum antibiotic to everyone in some defined population, several opportunistic infections can be reasonably effec- without taking the time todo a careful examination or history. tively treated (such as tuberculosis, candidiasis, herpes sim- Such a method might be especially worth considering at small plex virus infections), others are either difficult to diagnose health posts, in which the health provider is not trained to (such as cerebral toxoplasmosis, cytomegalovirus infec- follow accurately a diagnostic algorithm. The population tion), requiring sophisticated imaging or laboratory tech- treated might consist of all patients who present to the clinic nology, or very difficult to treat (such as crytosporidiosis, complaining OfsTD symptoms. A more radical strategy would infection with atypical mycobacteria). Relapses after the he to provide this broad-spectrum treatment to everyone in end of therapy are frequent for all opportunistic infections, a community, whether or not they are currently experienc- and treatment is often purely palliative. Estimates otf the ing STD symptoms or have presented for treatment. The costs of treatinig opportunistic infections in AIDS patients in cost-effectiveness of this "treat everyone" strategy compared Zambia are shown in appendix 20C. with a test-before-treatment strategy will vary, depending Treatment of AIDS patients with AZT results in an average on the cost of drugs and diagnostic tests, the prevalence rate prolongation of life of at least two years and a considerable of STDs in the reference population, and the sensitivity and improvement in the quality of life. Still, side effects (mainly specificity of the diagnostic procedure being considered.5' hematological, necessitating blood transfusions and interrup- tion of therapy) are common. Resistant strains of HIV appear to COST OF AIDS TREATMENT IN DEVELOPING COUNTRIES. Estimates emerge under therapy, and the drug costs as muchI as $750- of the treatment costs of persons with AIDS in developing $1,000 per month. countries have been constructed by applying use patterns (based, in the absence of data, on expert opinion) to im- Costs of Case Management perfectly known inpatient and outpatient average costs (Over and others 1988). Such AIDS treatment cost estimates, there- This section considers issues in the cost of case management fore, should be considered preliminary. Nevertheless, the re- of both HIV and STI. sults provided in table 20-26 are indicative of the range of values to be expected in developing countries. COSTS OF AN STD TREATMENT PROGRAM. Operation of an STD A principal finding of the studies which generated these treatnent program in a developing country involves both estimates is that the cost per patient varies considerably, both domestic and foreign resources. In most developing countries, across countries and within a country. Most cross-country 500 Mead Over and Peter Piot Table 20-26. Treatment Costs of AIDS in Selected Developing Countries (U S. dollars) Treatment cost as percentage Treatment cost of GNP per capita Countr'v GNP?' per capita Lowu High Low High Brazil 2,160 6,000 12,000 278 556 Mexico 2,080 3,286 7,344 158 353 Tanzania 290 104 631 36 218 Zaire 170 132 1,585 78 932 Note: Braili, estimates are 1988 U.S. dollars; Mexico, 1985 U.S. dollars; Tanzania and Zaire. 1986 U.S. dollars. All estimnates include both inpatient and outpatient treatment costs. The low and high estimates correspond, respectively, to the most miodest and the most comprehensive health care options available in the countrv. The average cost will rp;tcally be closer to the lowx than tt the high end of this range. Source: 0vet and others 1988; Tapia and Martin 1990; authors' calcularionm. variation in costs is caused by differences in wage rates paid to mates, which reveal the increased cost-effectiveness of CSTD providers, which tend to vary with levels of per capita UNP. treatment where an HIV epidemic exists. Treatment costs per case exhibit a range within a country for Because they are derived from the estimates in tables two principal reasons: variation in the clinical symptoms 20-16, 20-18, and 20C-3, the cost-effectiveness estimates in which manifest themselves and variation in the socioeco- tables 20-27 and 20-28 share a sensitivity to the level of nomic characteristics of the patienr and the medical and sexual activity of the treated person, the prevalence rate of institutional characteristics of the available health care op- the STD, and the average cost per clinic-hour. The magni- tions (Over and Kutzin 1990). On a percentage basis, the tude of the difference in cost-effectiveness strongly suggests poorest countries tend to exhibit greater cost variation because targeting treatment programs at the highly sexually active only a small proportion of all illness episodes are treated in a core group. In a region experiencing an HIV epidemic, for relatively high-cost hospital setting (Scitovsky and Over example. curing or preventing a CSTD where it has a preva- 1988). Cost variation exists in indtistrial countries but to a lence rate of 25 percent and the cost per clinic-hour is $2.00 lesser degree becatise widespread insurance coverage provides will buy disability-adjusted life-years for between $0.02 and 58-huwa 1.0a better access to hospital care for a greater proportion of the $0.11 each. If the cost per clinic-hour was $10.00, a population and standard treatmenit protocols are used on a disability-adjusted life-year could be saved for between wider basis. $0.04 and $0.25 by curing a CSTD. After interventions targeted at the core have been successfully implemented, Cost-Effective Case Management Strategies a country can purchase additional life-years with pro- grams aimed at noncore groups with low prevalence rates As various options exist for case management, it is important for less than $30.00 each (assuming $2.00 per clinic- to examine the cost-effectiveness of each approach to optimize hour) to less than $87.00 each (assuming $10.00 per resouirce allocation. clinic-hour). This higher range still compares favorably with the cost of saving life-years among adults with other COST-EFFECTIVE CSTD TREATMENT AND SECONDARY PREBENTION. health care interventions. How cost-effective is STI) control in a developing country in the presence of an HIV epidemic? First, consider the cost-effec- COST-EFFECTIVE M.ANAGEMENT OF AIDS CASES. Case manage- tiveness of CSTD treatment in the absence of H-V. In this ment of HJVor.ADS through the prophylactic administration of scenario, the measure of effectiveness would be the static and an antiviral agent like AZT is clearly not a cost-effective option dynamic DALYS saved per case from the averted CSTD only, for purchasing DALYs in developing countries. This situation which are presented in the last two columns of table 20-16. could change dramatically, however, if the price of antiviral Dividing the minimum cost of an effectively treated case of a therapy drops dramatically. In the absence of antivirals, treat- CSTD from table 20C-3 by these effects yields estimates of the ment of the opportunistic illnesses of an AIDS patient can buy cost-effectiveness of CSTD treatment in the absence of HIV. disability-adjusted life-years at the substantial but feasible cost These estimates are presented in table 20-27. of $235 to $384 when clinic time costs $10 per hour. This sum In the presence of an liiv epidemic, the effective treatment is approximately equal to the N?NP per capita of many of the of a CSTD has the dynamic effect of averting cases of Filv heavily affected countries and is substantially less than the infection. The total DALYS saved in this scenario are presented annual income of prime age urban adults in those countries. in the last two columns of table 20-18 and depicted graphically There is a strong argument to ensure the provision of the basic in figure 20-15. By again dividing the minimum effective drugs required to manage these opportunistic infections in treatment cost estimates by these greater effects, cost-effec- order to buy an extra year or two of life for the person with AIDS tiveness estimates of CSTD treatment in the presence of an liiv and to protect the drug supplies needed to treat other patients epidemic are generated. In table 20-28 we present these esti- who are not infected with HIV. Table 20-27. Cost per Discounted Disability-Adjusted Life-Year Saved by STD Treatment in Absence of an HIV Epidemic: Sensitivity to Prevalence Rate and Core vs. Noncore Strategy (1990 U.S. dollars) $2 per clinic hour $10 per clinic hour $30 per clinic hour Disability-adjusted life-years saved per effectively 1 percent 5 percent 25 percent I percent 5 percent 25 percent I percent 5 percent 25 percent Disease treated case prevalence prevalence prevalence prevalence prevalence prevalence prevalence prevalence prevalence Chancroid Minimum treatment cost n.a. 73 15 3 333 67 14 983 199 42 Coregroup 1.8 40.56 8.33 1.67 185.00 37.22 7.78 546.11 110.56 23.33 Noncore group 0.4 182.50 37.50 7.50 832.50 167.50 35.00 2,457.50 497.50 105.00 Chlamydia, female Minimum treatment cost n.a. 322 64 13 544 109 22 1,100 220 44 Core group 44.1 7.30 1.45 0.29 12.34 2.47 0.50 24.94 4.99 1.00 Noncore group 5.5 58.55 11.64 2.36 98.91 19.82 4.00 200.00 40.00 8.00 Chlamydia, male Minimum treatment cost n.a. 63 13 3 286 59 13 822 164 33 Core group 44.1 1.43 0.29 0.07 6.49 1.34 0.29 18.64 3.72 0.75 Noncore group 5.5 11.45 2.36 0.55 52.00 10.73 2.36 149.45 29.82 6.00 Gonorrhea, female Minimum treatment cost n.a. 295 59 12 463 93 19 884 177 35 Core group 37.3 7.91 1.58 0.32 12.41 2.49 0.51 23.70 4.75 0.94 Noncoregroup 4.5 65.56 13.11 2.67 102.89 20.67 4.22 196.44 39.33 7.78 Gonorrhea, male Minimumtreatmentcost n.a. 62 13 4 245 51 12 678 141 31 Coregroup 37.3 1.66 0.35 0.11 6.57 1.37 0.32 18.18 3.78 0.83 Noncoregroup 4.5 13.78 2.89 0.89 54.44 11.33 2.67 150.67 31.33 6.89 Syphilis Minimum treatment cost n.a. 185 38 9 269 56 14 477 102 27 Coregroup 160.8 1.15 0.24 0.06 1.67 0.35 0.09 2.97 0.63 0.17 Noncoregroup 19.8 9.34 1.92 0.45 13.59 2.83 0.71 24.09 5.15 1.36 n.a. Not applicable. Source: Authors. Table 20-28. Cost per Discounted Disability-Adjusted Life-Year Saved by STD Treatment in Presence of an HIV Epidemic: Sensitivity to Prevalence Rate and Core vs. Noncore Strategy (1990 U.S. dollars) $2 per clinic hour $10 per clinic hour $30 per clinic hour Disability-adjusted life-years saved per effectvely I percent 5 percent 25 percent I percent 5 percent 25 percent I percent 5 percent 25 percent Disease treated case prevalence prrevalence prevalence prevalence prevalence prevalence prevalence prevalence prevalence Chancroid Minimumtreatmentcost n.a. 73 15 3 333 67 14 983 199 42 Core group 55.6 1.31 0.27 0.05 5.99 1.21 0.25 17.68 3.58 0.76 Noncore group 3.8 19.21 3.95 0.79 87.63 17.63 3.68 258.68 52.37 11.05 Chlamydia, female Minimum treatment cost n.a. 322 64 13 544 109 22 1,100 220 44 Coregroup 113.5 2.84 0.56 0.11 4.79 0.96 0.19 9.69 1.94 0.39 Noncoregroup 11.4 28.25 5.61 1.14 47.72 9.56 1.93 96.49 19.30 3.86 Chlamydia, male Minimum treatment cost n.a. 63 13 3 286 59 13 822 164 33 Coregroup 113.5 0.56 0.11 0.03 2.52 0.52 0.11 7.24 1.44 0.29 Noncoregroup 11.4 5.53 1.14 0.26 25.09 5.18 1.14 72.11 14.39 2.89 Gonorrhea, female Minimum treatment cost n.a. 295 59 12 463 93 19 884 177 35 Coregroup 120.2 2.45 0.49 0.10 3.85 0.77 0.16 7.35 1.47 0.29 Noncore group 11.6 25.43 5.09 1.03 39.91 8.02 1.64 76.21 15.26 3.02 Gonorrhea, male Minimumtreatmentcost n.a. 62 13 4 245 51 12 678 141 31 Coregroup 120.2 0.52 0.11 0.03 2.04 0.42 0.10 5.64 1.17 0.26 Noncoregtoup 11.6 5.34 1.12 0.34 21.12 4.40 1.03 58.45 12.16 2.67 Syphilis Minimum treatment cost n.a. 185 38 9 269 56 14 477 102 27 Core group 396.3 0.47 0.10 0.02 0.68 0.14 0.04 1.20 0.26 0.07 Noncoregroup 41.1 4.50 0.92 0.22 6.55 1.36 0.34 11.61 2.48 0.66 n.a. Not applicable. Source: Authors. HIV Infection and SexuaUy Transmitted Diseases 503 Greater expenditure per AIDS case does not improve the veloping countries while keeping prices higher in the indus- probability of survival. Palliative care in the home and com- trial countries. Care must be taken, however, on three fronts. munity may not be as effective at prolonging life as is the use First, WHO must be protected from the fate that has of antivirals, but the data in table 20C-I suggest that palliative befallen most regulatory bodies throughout the history of care is almost certainly more cost-effective. Assuming that regulation-capture by the regulated industry. Only if mech- clinic time costs $10 per hour, palliative care alone (second anisms can be arranged to ensure that WHO can remain an row in the table) buys one year of healthy life, and antivirals independent, flexible regulatory body responsive to the (first row in the table) buy two years of healthy life, a disabil- health needs of the poorer countries, should it be mandated ity-adjusted life-year is purchased much more inexpensively to play that role. with palliative care ($235 forone yearby treatingAIDSwithout Second, there must be allowance for competition among AZT) than with antiviral treatment ($1,200 for each of two private firms, which implies that several produce the same years by treating AIDS with AZT). More evidence of the oppor- product. A regulatory decision to allocate the production of tunity for considerable improvement in the cost-effectiveness each drug to only one firm would remove the benefits of of AIDS treatment is provided in table 20-26, where the ranges competition and prevent the lower prices and higher quality of estimates for each country suggest the feasibility of reducing that competition would eventually entail. the cost per case. Finally, despite the currently high prices of antivirals, devel- oping countries should prepare now for the day when their Developments in Case Management in the Next Decade prices will fall. We predict that, as a combined result of patent expiration, technical change, increased competition, and po- It is anticipated that major advances will be seen in case litical pressure on pharmaceutical companies, prices for anti- management, particularly of AIDS, in the near future. virals will fall rapidly-perhaps by a factor of 100 in the next three years. How much AZT should Uganda buy and how should DIAGNOSTIcs. The most promising area for innovation incase it be distributed if a year's dose costs $700? How much if $70? management is probably the development of simple diagnostic How much if $7? Rather than responding to this price drop in tests for most STDs. The basic technology (mainly enzyme an ad hoc manner, govemments of developing countries immunoassays) is already available, and research is currently should prepare now by developing guidelines for the purchase focusing on improving test performance. This will allow on- and equitable allocation of antivirals under every possible set the-spot simple and rapid specific diagnosis of CSTDs such as of future prices. gonorrhea, chlamydial infection, and chancroid-an import- ant, if not essential, element in the control of gonococcal and Priorities chlamydial infections in women. The cost of such tests is presently still high ($4 to $7), but it is expected that prices will It is clear from the above that AIDS and STD programs have not decrease because many companies are becoming active in this prioritized enough, and also that rational prioritization is pos- field. It is not clear, however, how and if these new diagnostic sible on the basis of available data and modeling exercises. tools will be used in developing countries. Priorities for Resource Allocation DRUGS. New antibiotics are being continuously developed, including agents active against bacterial STDs. Inexpensive As extensively pointed out above, STDs in general and HIV oral antibiotics able to cure gonorrhea caused by multiresis- infection, chlamydial infection, and syphilis in particular are tant strains when given as a single dose are urgently needed a considerable source of morbidity and mortality in many parts and may become available. Numerous antiviral compounds of the developing world, ranking them first among the top have recently been developed and are being or will be fifteen causes of disability-adjusted life-years lost in the most evaluated for their clinical effectiveness in HIV infection, heavily affected urban populations. Even in low-prevalence both in asymptomatic carriers of HIV and in AIDS patients. It urban populations, HIV infection and CSTDs rank eleventh is expected that more effective and less toxic therapy for HIV among the causes of health lost, ahead of tuberculosis, adult infection will become available in the 1990s. It will proba- pneumonia, and neonatal tetanus, for example. The growing bly consist of lifelong treatment with a combination of urbanization and the increasing population share of young antiviral drugs. This would have a significant effect on the adults in most parts of the developing world can only make prevention of HIV infection, because secondary prevention things worse. would then become an important additional strategy for the Still, our analysis has shown that prevention of HIV infec- control of HIV infection. tion, and to a lesser extent of some CSTDs, can result in consid- For these new pharmaceutical developments to be relevant erable health gain, as compared with other common health to developing countries, prices must he low. Yet low prices problems in the developing world. In addition, sex education, remove the incentive for continued research and development use of condoms, and prompt treatment of STDs all contribute by private firms. Intemational political collaboration, perhaps to reproductive health, which includes the ability to bear and coordinated by wHo, could help to negotiate lower prices for de- raise healthy children. 504 Mead Over and Peter Piot It is truly remarkable that this high rank both in burden and and a variety of treatment options. The results reflect the potential health gain is not reflected in higher specific expen- sensitivity of altemative interventions to assumptions regarding diture for the control of HIV infection and CSTDs. Neglected targeting strategy, CSTD prevalence, the presence of an HiV epi- training, poor diagnostic and therapeutic capabilities, high demic, and clinic costs. Extreme assumptions are used to define rates of quasi-irreversible sequelae, and insufficient research the unfavorable and favorable cost-effectiveness scenarios. and development efforts (at least for CSTDs) are all symptoms An important conclusion that can be drawn from table of this inadequate response. We can only guess why this 20-29 is that certain STD interventions are extraordinarily situation has arisen. cost-effective under favorable assumptions. If the highly sexu- Given the complex mosaic of areas of high and low preva- ally active population is targeted, STD prevalence is high, and lence, even within a single county, rational allocation of inexpensive strategies are used, blood screening, condom sub- resources for care and prevention of HIV infection and CSTDs is sidies, and IEC interventions can save a DALY for less than $0.15. even more difficult to plan in the developing world than in Using similar assumptions in the presence of an HIV epidemic, North America or Europe. This is also the case, however, for we find that STD treatment is also remarkably cost-effective. other health problems that are increasingly important in urban Even under unfavorable assumptions, however, some of the populations of the developing world, such as cardiovascular interventions, such as condom subsidies and STD treatment, diseases. The continuing strong urbanization in all developing remain cost-effective in relation to other adult health inter- cotntries and the growing proportion of the population in the ventions or to the level of per capita GNP. But some interven- sexually active age range indicate that the global population tions, such as the use of antivirals to treat persons with AIDS or potentially at risk for sexually acquired infections will continue blood screening of the general population in an environment to increase. of low STD prevalence, should be pursued only after other health investments are made. Priorities for the Control of STDs anid HIV Infection TARGETING. A general finding made earlier in the chapter is An argument for allocation away from other diseases and that the cost-effectiveness of programs to prevent and control toward STD control must be based on the costs and the effects STDs can be extremely high when the program, whether pre- of relevant options. In this chapter we have made a strong case ventive or curative, is tightly targeted. The cost per discounted for the important health benefits both statically and dynami- disability-adjusted life-year saved can be as low as $0.15 for a cally of preventing STDs. Although we have not been able to blood-screening program and $0.56 for a CSTD treatment pro- assign costs to all the interventions that would produce these gram, when such programs are aimed at a high-prevalence core effects, altematives for which we have quantified results in group of transmitters. Blood screening or case management in table 20-29 include three important preventive interventions the rural noncore or in a segment of the population in which Table 20-29. Cost per Discounted Disability-Adjusted Life-Year Saved for Alternative STD Interventions Intervention _ Parameter Unfavorable assumptions Favorable assumptioros Prevention Condom subsidies and IEC Cost High Low Target group Noncore Core Targetdisease Chancroid and HIV Syphilis and HIV Cost per DALY $40.91 $0.13 Blood screening Cost of test Expensive Inexpensive Target group Noncore Core Prevalence < 0.1 percent > 5 percent Cost per DALY > $244 $0.15 Gonococcal ophthalmia neonatorum Prevalence < 0.1 percent > 1 percent Cost per DALY >$111 <$5.32 Treatment CSTDS Hourly clinic cost $10 $2 Target group Noncore Core HIV epidemic No Yes Prevalence < I percent > 5 percent Cost per DALY > $50 < $0.56 AIDS Hourly clinic cost $10 $2 Treatment Antivirals Palliative and home care only DALYs gained 2 1 Cost per DALY $1,200 $75 Source: Authors' calculations. HIV Infection und Sexually Transmitted Diseases 505 there is no HIV epidemic, however, can be a much more tion programs first exhaust the possibilities for campaigns expensive way to save IDALYs. Blood screening can cost $300 or targeted at high-risk groups via an access group which provides more per DALY saved, and treatment of chlamydia in the both high coverage of rhe risk group and high concentration noncore when there is no HIV epidemic saves DALYs at a cost of the message as defined in table 20-23 and the accompanying of $2,457 each. Hence, our main recommendation is that discussion above. Only later and after thorough analysis should all national health programs should at a minimum include those campaigns be extended to noncore groups. a few STD clinics and control programs targeted at urban core groups. Furthermore, in view of the fact that much of the EMPLOYMENT OPPORTUNITIES FOR URBAN WOMEN. In view of benefit of these programs will accrue to individuals other the findings reported in earlier sections of the chapter, we are than those directly contacted, the economic theory of persuaded that STDS play a particularly noxious role in the lives externalities argues that these services to the core group of women indevelopingcountries. Especially incountries with should be highly subsidized. low female-to-male ratios in urban areas, STDs are both the The degree of extension of STD treatment and control be- cause and the consequence of the entrapment of women in a yond the core groups and the most cost-effective disease inter- position of low socioeconomic status. We recommend that ventions should vary across countries according to their STD) "women-in-development" (WID) programs join forces with STD epidemiology and their access to resources to fund these pro- control programs to break this vicious circle. More especially, grams. Assume that countries might seek to equate the cost- in order to maximize this development effect in countries with effectiveness of interventions on all diseases to approximately small female-to-male ratios in urban areas, WID programs their level of per capita GNP. Then a country with a per capita should request help from STD control programs in targeting job GNP of $300 and a cost per clinic-hour of $10 would be guided training and employment opportunity programs. Resources for to consider STD treatment of syphilis, gonorrhea, chlamydia, WID programs targeted at women at high risk of STDs should and AIDS opportunistic illnesses in the core groups. Blood- come from both wIn and STD programs, to serve both their screening and safe-sex programs would also be conducted goals. there. The only case management option which can save DALYs at less than $300 in the noncore group, however, is syphilis INTEGRATION OF STD CONTROL INTO EXISTING STRUCTURE. The treatment, which would be robust even if the resource cost of existing health care structure should be strengthened in its the program per patient contact (the cost per clinic-hour) components for the diagnosis and treatment of CSTDs and AIDS triples to $30 per hour. and for health promotion. This requires manpower training; In contrast, a middle-income country might have a GNP per availability of diagnosis, drugs, and educational materials; and capita of $4,000 and a cost per clinic-hour of $30. Such a programmatic coordination and supervision. Improving the country should implement all the programs described above. access of women to health services is particularly important for In addition, syphilis, chlamydia, and gonorrhea could be at- STID control. In areas of high or medium prevalence, STD and tacked in the noncore for less than $4,000 per DALY saved. HIV services should be integrated into primary health care, Provided antiviral therapy can be obtained for as low as $2,000 mother and child health services, antenatal and, especially, per year, even AZT treatment of AIDS and DIiv-infected people family planning clinics. would be cost-effective in both the core and the noncore Because of fundamentally identical strategies, STD and AIDS in such a country. Universal blood screening would be cost- control programs should be developed in close coordination. effective provided the prevalence rate of infection is greater National medium-term plans with clear and achievable objec- than one in 76,800 in the core or greater than one in 6,800 in tives should also be formulated for STD control. the noncore. Becauseourestimatesofthe cost-effectiveniessofprevention PATIENT CARE AND SUPPORT. More emphasis should be given programs are less solidly based than are our estimates of treat- to support activities as part of AIDS control programs. Such ment costs, our recommendations regarding the allocation of activities include not only etiologic and palliative therapy but resources between prevention and treatment programs are less also psychological and social support of the patient and his or certain. The qualitative conclusions arrived at earlier, how- her family. Where resources are scarce, such support should be ever, point in the same direction as the more quantitative ones targeted to the families of high-risk individuals so that the described above: targeted programs will typically be more support efforts have the side benefit of minimizing subsequent cost-effective. Our estimates in table 20-24, summarized in the infection. Secondary prevention of STD complications through first row of table 20-29, suggest that focusing an IEC program early diagnosis and treatment remains a cornerstone in the on the core group will be from four to eight times more control of bacterial STDs. cost-effective in saving DALYs than if the program is tar- geted at the noncore. Of course, countries should include Research and Developnent the costs of preventing or reducing any undesirable social and epidemiological side effects when they estimate the While writing this document, we became aware that many costs of targeted programs. Assuming such side effects can essential data on CSTDs and HIV infection are lacking and that be avoided or reduced, we recommend that AIDS preven- both operational and basic research on STDs continue to be 506 Mead Over and Peter Piot neglected. This is undoubtedly a handicap for control and the limitations of interventions directed only at the level of prevention programs. individuals' risk behaviors. TECHNIC'AL RESEARCH AND DEVELOPMENT. Simple, rapid, and INTERVENTIONS. There is an urgent need to develop and inexpensive diagnostic methods are a prerequisite for the suc- evaluate innovative behavioral and medical interventions cessful implementation of both individual case management against HIV infection and CSTD in different societies. Such and screening and case detection programs for treatable STDs. development and evaluation would include trials and feasibil- Such tests are not available, however, particularly for the ity studies, demonstration projects, and community-wide in- detection of genital infections in women. Priority diseases terventions. This research is relevant for both primary and include gonorrhea, chlamydial infection, congenital syphilis, secondary prevention. Examples of such research include the and chancroid. In addition, further development of simple effect and sustainability of campaigns for safe sex and condom serological tests for HIV antibody is necessary, ideally leading use in various groups (adolescents, prostitutes, and so on); to a cheap way of confirmatory testing. screening of pregnant women for syphilis to prevent congenital The necessary technology is available, and important dev- syphilis; eye prophylaxis at birth by traditional birth attenders elopments are anticipated in the near future. Special consid- to prevent gonococcal ophthalmia neonatorum; evaluation of eration should be given to make these tests affordable for syndrome-oriented algorithms for STDs to prevent complica- developing countries and to make them available through the tions; evaluation of various mechanical and chemical barrier health care system. methods for the prevention of IIv infection; use of rapid tests Priority should be given to prevention technologies that are for the screening of blood donations; evaluation of the effec- fully controllable by women, such as mechanical and chemical tiveness and cost benefit of partner notification; and different barriers. Products not traditionally used as contraceptives methods of counseling. The identification of appropriate tar- should be screened for both bactericidal and virucidal activity get populations for interventions has traditionally been a prob- against HiV and the full range of STD agents. Innocuous and lem and deserves more research. More attention should be acceptable vehicles for these products should be developed and given to the development and evaluation of methods for the evaluated. The possibility of the production of reusable con- evaluation of interventions, with emphasis on simple indica- doms should be investigated. tors usable in developing countries. The availability of vaccines against CSTDs and IHIV infection would obviously revolutionize the control of these diseases. For HEALTH SERVICES RESEARCH AND IEC. Service delivery plays infections such as gonorrhea and genital chlamydial infection an important role in CSTD and AIDS control. Yet both the costs in women, even a vaccine which would not completely and the effects ofaltemativeSTDprevention and case manage- prevent infection but would prevent the development of ment are almost completely unknown. Crucial questions are complications and sequelae may be acceptable. Insufficient how, whether, and at what cost STD control can be integrated knowledge of the immunobiology of many STDs, complex mech- into existing health systems, including primary health care anisms by microorganisms to escape the immune response in structures, family planning services, mother and child pro- humans, and poor commercial interest (at least for the CSTDs) grams, and drug abuse programs. This issue involves not only have all been significant obstacles to vaccine development. case detection and management, but also some of the weakest Guidelines and methods should be developed for phase Ill components of the health care system, such as information, vaccine trials (which evaluate protective efficacy). counseling, and education. Possibilities of involvement of the community, for instance, through home care, should be ex- EPIDEMIOLOGICAL AND BEHAVIORAL RESEARCH. Epidemiologi- plored and evaluated. cal research priorities include the collection of baseline data on STDs and their complications; the development of methods CASE MANAGEMENT. Because of the increasing number of for disease surveillance; further investigations on the natural patients with AIDS and AIDS-related complex, there is an urgent history and risk factors for STDs; the effect of HIV infection on need to develop simple and inexpensive strategies of case the natural history and response to treatment of other diseases; management for adults and children, making use of essential the dynamics of core groups; and factors determining diverse drugs, home remedies, and community members. Effective epidemiologic pattems. The relative and population-attribu- antiviral therapy will probably become widely available in the table risks for transmission of HIV should be better quantified. near future, but it is not clear how this will affect case manage- Behavioral sciences have been particularly neglected in STD ment in developing countries. Individual countries should research, though their importance in the prevention of AIDS commission studies to determine the recommended treatment and in the assessment of its effect is increasingly recognized. protocols for AIDS under today's set of prices. These studies Data on sexual, health, and substance use behaviors, with should also recommend criteria for the govemment to use in emphasis on risk behaviors, should be collected in various determining at what price it will begin to buy and allocate societies and groups to lead to a rational strategy of prevention antiviral drugs. Clinical trials of high priority include the and control. The study of societal patterns as determinants of effectiveness of syndrome-oriented algorithms for the manage- STDs and of the social effect of STDs should be helpful to define ment ofSTDs; innovative treatments ofresistant gonorrhea and HIV Infection and Sexuallv Transmitted Diseases 507 of acute l'ID to reduce postinfectious infertility; and evaluation not appear significantly to affect the spread of gonococcal of the validity of simple tests for the diagnosis of HIV, gonor- infection, probably because of a multitude of antigenic types rhea, chlamydial infection, genital ulcer disease, and PrID and because most infections may not induce protective anti- body when they are limited to the genital mucosa. The risk of ECONOMlIC EFFECT ON HOUSEHOLDS. Information on the mag- acquiring N. gonorThoeae during heterosexual vaginal inter- nitude of the effect on the surviving household members of course is 30 to 40 percent for the uninfected male partner and fatal illness from AIDS and other causes would serve three 50 to 80 percent for the uninfected female partner (Hooper important purposes. First and most immediately, such informa- and others 1978). tion could guide the design of carefLlly targeted programs to Gonorrhea is the main cause of urethritis among male clinic assist temporarily certain surviving household members after attenders in the developing world (Meheus and others 1980; an AIDS death. Although government-financed life-insurance Antal 1987). Urethral stricture is the most severe complica- policies will clearly be beyond the financial reach of the most tion of gonococcal urethritis in males and may make up the severely affected African countries for some time, many Afri- majority of cases seen by urologists in some parts of Africa can countries are currently considering the implementation of (Bewes 1973). Still, it is in women that gonococcal infection poverty alleviation programs. To the extent that research leads most often to severe complications and sequelae. could discover indicators which predict which surviving Women with gonorrhea mostly have genital manifestations, households were most likely to be plunged into poverty by the although these may be nonspecific (McCormack and others AIDS death, poverty alleviation programs would be able to add 1977). If untreated, between 5 and 10 percent of women with these households to their beneficiaries, thereby mitigating gonococcal infection develop salpingitis-a potentially life- some of the worst effects of the AIDS epidemic. threatening condition if peritonitis develops. The risk of in- Second, information on the relative effects of STDs and other voluntary infertility is about 15 percent after one episode of diseases would guide policy choices on the allocation of re- salpingitis, 30 percent after two episodes, and over 50 percent sources among alternative disease programs. By an extension after three or more episodes (Westrom 1980). of the logic of this collection, a disease is important not only The proportion of women who are infertile because of N. for its effect on the infected individual, but also for its effect gonorrhoeae has not been defined, but in Uganda there was an on other household members. If it is determined, for example, inverse correlation between the fertility rate and the incidence that an adult with an STD has a more negative effect on the of gonorrhea by district (Arya, Taber, and Nsanze 1980). In health of other family members than an adult sick with other addition, the risk of ectopic pregnancy-one of the leading diseases, then this would strengthen the argument for reallo- causes of maternal death-is increased tenfold after one epi- cating resources to STDS. sode of PiD (Westrom 1980). Third, and finally, information on the magnitude of the N. gonorrhoeae is also an important cause of morbidity in economic effects of STDs on households would move the allo- mother and neonate. Matemal gonococcal infection is a risk cation of resources away from other sectors and toward the factor for premature delivery, and it may be a cause of chorio- health sector. amnionitis. Furthermore, it is a significant cause of postpartum endometritis and salpingitis, a complication that occurred in up to 20 percent of all parturient women in studies in eastern Appendix 20A. The Medical Consequences Africa (Plummer, Laga, and others 1987; Temmerman and of Sexually Transmitted Diseases others 1988). The risk of transmission of N. gonorThoeae from an infected mother to her infant's eyes is 30 to 40 percent if First we describe the medical consequences of each sexually ophthalmic prophylaxis at birth is not used (Galega, Heyman, transmitted disease discussed in the text, and then we sum- and Nash 1984; Lagaandothers 1986aand 1986b). Depending marize this information in quantitative form. We end the on the prevalence ofgonococcal infection in pregnant women, appendix with the presentation of estimated discounted the incidence of gonococcal ophthalmia neonatorum is up to disability-adjusted life-years lost from a single typical case 3.5 percent ofall live births in some African populations(Laga, of each STD and a comparison of these figures with those for Meheus, and Piot 1989). Gonococcal ophthalmia is associated other important diseases. As explained in the text, these with keratitis in 10 to 20 percent of cases (Fransen and others figures can be interpreted as the benefit of averting a case of 1986), and an unknown but probably small proportion of cases each disease: for example, of the disability-adjusted life-years will become blind. N. gonorrhoeae is also an important cause saved per case prevented or cured. of keratoconjunctivitis in adults in the tropics (Kesteleyn, Bogaert, and Meheus 1987). Gonorrhea Genital Infection with Chlamydia trachomatis Gonorrhea is caused by NeisseTiagonorrhoeae, a fastidious gram- negative diplococcus, which displays antigenic variation. Chlamydia trachomatis is an intracellular parasitic bacterium Strain type specific, temporary protective immunity has been with a complex replication cycle that takes forty-eight to documented. In general, however, protective immunity does seventy-two hours. It is susceptible to several groups of anti- 508 Meal Over and Peter PTot microbial agents, including the tetracyclines and macrolides. syphilis in two cohort studies in Norway and the United States Fourteenserotypeshavebeendescribed.OftheseserotypesLI, (Sparling 1990). The case-fatality rate in these studies was L2, and L3, whiclh have distinct hiologic features, cause lym- approxinmately 20 percent. Pregnant women with untreated phogranuloma venereum, a fairly uncommon cause of inguinal syphilis of under two years' duration transmit the infection to and femoral tymphadenitis in the tropics. Three serotypes (A, theirfetus in almost all cases. The proportion of affected fetLses B, C) are mainly, but not exclusively, associated with tra- decreases in women who have had syphilis longer than two choma, a potentially blinding eye disease endemic in many years. Approximately 50 percent of the pregnancies in mothers developing countries. The remaininig types cause basically the with primary or secondary syphilis result in abortion, stillbirth, same clinical syndromes as N. gonorrhoeae (table 20-1). Gen- perinatal death, orprematuredelivery. Clinical manifestations erally, genital chlamydial iniections and their complications, usually appear between two and eight weeks in infected neo- such as PiD, are associated with milder, and even subclinical, nates and resemble those of secondary and tertiary syphilis disease-although their complications and sequelae may be (Hira and Hira 1987). Irreversible sequelae and death due to equally severe. This implies that many infections go unnoticed syphilis occur in 50 to 75 percent of the infants. or do not come to medical attention. The risk of heterosexual transmission of C. trachomatis is Genital IJlcer Disease probably somewhat lower than for N. gonorrhoeae, but the risk of developing PID for women with cervical chlamydial infection Genital ulcer disease has a diverse etiology, including primary is also of the order of 5 to 10 percent (Westrom 1980). The syphilis, chancroid, genital herpes, donovanosis, and lympho- agent has been identified in 15 to 20 percent of women with granuloma venereum (Piot and Meheus 1986). It is relatively acute salpingitis in Africa, and there is strong serological more common in many parts of the developing world than in evidence of a role played by C. trachomatis in ectopic preg- Europe or North America. There is growing evidence that nancy and infertility, particularly in women with evidence of genital ulcers increase both the susceptibility to HIV during tubal disease (Meheus, Remeis, and Collet 1986; Plummer and sexual intercourse with an HIV-infected partner and the infec- others 1987; De Muylder and others 1990). tiousness of an HRV-infected person for uninfected partners C. trachomatis is the main identifiable cause of ophthalmia (WHO 1989a; Piot and others 1988). Only chancroid will be neonatorum in Africa (Maybey and Whittle 1982; Meheus discussed here, because it is the most common cause of GUD in and others 1982; Laga and others 1986a and 1986b; Buisman the tropics. Chancroid is caused by Haemophilusducreyi,a small and others 1988). It is also a cause of neonatal pneumonia, but gram-negative rod with fastidious growth requirements. There it is unknown what proportion of cases in the developing world is geographic variation in antigenic properties, and protective is due to C. trachomatis. The negative effect of C. trachomatis immunity has not been reported. H. ducreyi strains show in- on pregnancy outcome is controversial, though it seems plau- creasing resistance to antimicrobial agents. sible that the agent is a significant cause of postpartum endo- The hallmark of chancroid is multiple, painful, purulent metritis in the tropics (Gravett and others 1986; Plummer and genital ulcers, accompanied in more than half of the cases by others 1987; Temmerman and others 1989). a painful inguinal lymphadenopathy. The incubation period is There is as yet no evidence for protective immunity in short, three to ten days. Unlike syphilis, chancroid yields no genital chlamydial infections. In two studies in Africa, genital long-term or systemic complications. Without effective ther- chlamydial infection was found to be a risk factor for the apy, lesions last for an average of two months. acquisition of HIV in female prostitutes (Plummer and others In Sub-Saharan African and Southeast Asia, H. ducrevi 1991; Laga and others 1989). can be isolated from 20 to 60 percent of patients with genital ulcerations. In most countries, patients with GUD belong Syphilis to low socioeconomic strata. Chancroid seems to be asso- ciated with prostitutes in several parts of the world, includ- Syphilis is caused by a fastidious, slowly replicating spirochete, ing North America, and is more common in uncircumcised Treponemna pallidum, which is still not cultivable in vitro. T. men. pallidum is highly sensitive to penicillin, and no in vitro resis- tance to this antibiotic has as yet been reported. The risk of AIDS and HIV Infection acquiring syphilis through heterosexual intercourse is thought to be less than 30 percent, but receptive anal intercourse Human immunodeficiency virus is a retrovirus that preferen- significantly increases this risk (Sparling 1990). tially infects CD4 bearing cells, including T lymphocytes and The disease is characterized by distinct clinical phases and macrophages. Though other cell types may also become in- a long latency period between the initial manifestations (pri- fected, there are at least two types of HIV, named HRV- 1 and mary chancre and secondary syphilis) and the severe systemic HiV-2, which share some antigenic properties (mainly at the complications of tertiary syphilis, including neurosyphilis and level of core proteins) but which are clearly distinct in their cardiovascular syphilis, which occur five to twenty years after genome. In addition, individual HIV isolates exhibit significant infection. Primary syphilis in 20 to 40 percent of those who genetic variability (Clavel 1988). Once infected, an individual did not have therapy progressed to symptomatic tertiary remains infected (and infectious) throughout his or her life. HIV Infecton and Sexually Transmitted Diseases 509 At six to sixteen weeks after infection, approximately one- In the United States, individuals with HIV infection progress third of those infected develop a benign acute viral syndrome to AIDS at an average rate of 3 to 6 percent per year, and the which resolves spontaneously after a few weeks. Subsequently, median time from infection to progression to AIDS has been infected individuals go through an asymptomatic latent phase estimated at seven to ten years (Moss and Bacchetti 1989). which may last for ten years or longer, after which they develop There is evidence that the rate of progression to AIDS is related AIDS and AlDs-related complex (ARC). By definition, AIDS is directly to age, at least among hemophiliacs residing in North characterized by the occurrence of life-threatening opportunis- America (Goedert and others 1989). In addition, HIV-seropos- tic infections and tumors, whereas ARC may be considered as a itive persons who have not progressed to AIDS show steadily non-life-threatening symptomatic IDIV disease. Approximately increasing impairment, with two-thirds of the subjects having one-third of the patients suffer from subacute encephalopathy, some clinical problem after three years of infection. It appears characterized by progressive behavioral changes associated that in the absence of treatment, most, if not all, infected with dementia (the AIDS dementia complex). The case-fatality persons will progress to AIDS. Epidemiologists hypothesize that rate of AIDS is virtually 100 percent, with an average time of the rate of progression to AIDS is faster in a developing country two to three years between diagnosis and death. Because ot because of greater stress on the immune system due to frequent inadequate means of diagnosis and treatment, this period is exposure to infectious diseases, but no data to test this hypoth- probably much shorter in developing countries. esis are yet available. Table 20A 1. Outcome Probabilities for STD and Other Major Diseases Case-fatality rates, bty age (per 1,000) Probability of permanent Disease Prevalencea Sex 0-1 1-4 5-14 15-49 50-64 65+ disablement Sexually transmitted disease Chancroid High Both ... ... ... ... ... ... 0.5 Low Both ... ... ... ... ... ... 0.5 C'hlamydia High Female 0.05 ... ... I ... ... 5 High Male 0.05 ... ... 0.05 ... ... 5 Low Female ... ... ... ... ... ... 5 Low Male ... ... ... ... ... ... 5 Gonorrhea High Female ... ... 0.1 0.2 ... ... 5 High Male ... ... ... ... ... ... 5 Low Female ... ... 0.1 0.2 ... ... 5 Low Male ... ... ... ... ... ... 5 HIV High Both 100 100 100 100 100 100 n.a. Low Both 100 100 100 100 100 100 n.a. Syphilis High Female 60 ... ... I ... ... 1.5 High Male 60 ... ... 1 ... ... 1.5 Low Female 25 ... ... 0.35 ... ... 1.5 Low Male 25 ... ... 0.35 ... ... 1.5 Other diseases Birth injury n.a. Both 67 ... ... ... ... ... 33 Cerebrovasculardisease n.a. Both ... ... ... 35 35 35 35 Cirrhosis n.a. Both 80 80 80 80 80 80 20 Congenitalmalformations n.a. Both 15 15 15 15 15 15 85 Gastroenteritis n.a. Both 7 3 0.2 0.2 0.2 0.4 ... Injuries (i.e., accidents) n.a. Both 10 10 10 10 10 10 5 Malaria n.a. Both 14 6 1 1 1 1 86 Measles n.a. Both 18 17 1 ... ... ... ... Pneumonia, adult n.a. Both n.a. n.a. n.a. 10 10 10 ... Pneumonia, child n.a. Both 40 40 40 n.a. n.a. n.a. ... Prematurity n.a. Both 10.2 n.a. n.a. n.a. n.a. n.a. ... Severe malnutrition n.a. Both 70 65 10 10 10 20 ... Sickle cell n.a. Both 80 ... ... ... ... ... 20 Tetanus (neonatal) n.a. Both 80 n.a. n.a. n.a. n.a. n.a. ... Tuberculosis n.a. Bothi 35 35 35 35 35 35 ... n.a. Not applicable. .. Negligible. a. Urban areas of high or low prevalence, as dtfined by table 20-7. Source: Authors; Ghana Health Assessment Prolect Team 1981. 510 Mead Over and Peter Piot Quantification of the STD Sequelae for Comparison for each of these diseases, ranked in order of the discounted disability-adjusted life-years saved. We assumed that indi- The method introduced in the main text of this chapter for viduals would have otherwise lived to the age of sixty-five quantifying the static burden of disease requires that the se- years, and we applied a discount rate of 3 percent to all quelae be described in terms that can be roughly compared future years. The second column of the table presents the across diseases. For simplicity the method divides the possible results of first weighing each lost year by a productivity outcomes of contracting a case of each disease into three weight before discounting and adding the years to arrive at classes: death, permanent disablement, and recovery. In table the discounted productivity-weighted disability-adjusted 2OA-1 we summarize the assumptions made regarding the life-years saved per case averted. As in Barnum 1987; Over probability of each of these outcomes for each of the diseases and others 1988; and Over, Bertozzi, and Chin 1989, pro- included in the analysis. (Note that, for simplicity, only the ductivity weights are attached to future disability-adjusted case-fatality rate is assumed to vary across age groups.) In table life-years before they are discounted to the time the disease 20A-2 we summarize our assumptions regarding the duration is contracted. Age ranges and the weights attached to the and degree of disablement due to the sequelae of the STDs and years that would have been lived at those ages are: ages 0-1, the other main diseases. 0; ages 1-5, 0; ages 5-15, 0.2; ages 15-50, 1.0; ages 50-65, Finally, in table 20A-3 we present estimated disability- 0.85; age 65+, 0.25. These weights roughly follow the age adjusted life-years lost per case, or saved per case averted, profile of hourly wages in a developing country (for exam- Table 20A-2. Death and Disablement frot Sequelae of STDs and Other Major Diseases Years Disablement Chronic Disablement Days until Disease Prevalencea Sex until death until deathb disablementb until recoveryb recovery Sexually transmitted disease Chancroid High Both n.a. n.a. 20 20 73 Low Both n.a. n.a. 20 20 73 Chlamydia High Female 1 30 30 20 292 High Male 5 10 50 10 91 Low Female 1 30 30 20 292 Low Male 5 10 50 10 91 Gonorrhea High Female 1 50 50 20 146 High Male n.a. n.a. 50 10 55 Low Female 1 50 50 20 146 Low Male n.a. n.a. 50 10 55 -IV High Both 8 18 n.a. n.a. n.a. Low Both 8 18 n.a. n.a. n.a. Syphilis High Female 20 50 50 0 730 High Male 20 50 50 0 730 Low Female 20 50 50 0 730 Low Male 20 50 50 0 730 Other diseases Birth injury n.a. Both 0 ... 20 Cerebrovascular disease n.a. Both 0 ... 75 ... 120 Cirrhosis n.a. Both 5 50 25 Congenital malformations n.a. Both 0 25 Gastroenteritis n.a. Both 0 ... n.a. ... 14 Injuries (i.e., accidents) n.a. Both 0 ... 25 ... 30 Malaria n.a. Both 0 ... 2 Measles n.a. Both 0 ... ... ... 21 Pneumonia, adult n.a. Both 0 ... ... ... 30 Pneumonia, child n.a. Both 0 ... ... ... 30 Prematurity n.a. Both 0 ... ... ... ... Severe malnutrition n.a. Both 0 ... ... ... 180 Sickle cell n.a. Both 5 50 30 Tetanus (neonatal) n.a. Both 0 ...... ... ... Tuberculosis n.a. Both 5 25 200 n.a. Not applicable. Not significant. a. Urban areas of high or low prevalence, as defined in table 20-7. b. Disablement expressed as a percentage of full health. A year at I OL percent disablement is weighted the sanie as a year lost to death. Source: Authors; Ghana Health Assessment Project Team 1981. HIV Infection and Sexually Transmitted Diseases 511 Table 20A-3. Health Gains From Preventing One Case of STD or Other Diseases Discounted disability-adjusted life-years saved Discounted productive disability-adjusted life-years saved Disease Years Rank Years Rank Sickle cell 24.5 1 14.2 3 Tetanus (neonatal) 22.7 2 14.2 2 Birth injury 20.9 3 13.1 4 Hiva 19.5 4 15.5 1 Severe malnutrition 17.0 5 11.2 6 Cirrhosis 15.6 6 12.8 5 Pneumonia, child 11.2 7 8.1 7 Congenital malformations 10.3 8 6.4 9 Cerebrovascular disease 9.4 9 7.8 8 Tuberculosis 7.1 10 5.6 10 Measles 5.0 11 3.3 11 Syphilis,' femaleb 3.9 12 2.3 13 Syphilis,a male' 3.7 13 2.1 14 Malaria 3.7 14 2.4 12 Prematurity 2.9 15 1.8 16 Injuries (i.e., accidents) 2.7 16 2.1 15 Pneumonia, adult 2.0 17 1.6 17 Gastroenteritis 1.4 18 0.9 18 Chlamydia,'femaleb 1.3 19 0.9 19 Gonorrhea,1 femaleb 1.0 20 0.7 20 Chlamydia,' malec 0.8 21 0.6 21 Gonorrhea,' male' 0.7 22 0.5 22 Chancroid' 0.2 23 0.1 23 a. Sexually transmitted disease. b. Benefit of a program targeted at females, per case averted or prevented. c. Benefit of a program targeted at males, per case averted or prevented. Source: Ghana Health Assessment Project Team 1981 and authors' calculations. pie, Lucas 1985). The results are discussed in the main part of I. The equation can be written as the difference between the text. the newly infected people and the newly cured people dur- ing the day, or Appendix 20B. A Simulation Model of an STD ( _1) dN1, [Newly infected 1 [Newly cured Epidemic d t L group i j L groupi j The basic model used here is drawn from Hethcote and The rate of new cures is simply given by the total number of Yorke 1984 (for example, their equation 3.1). It consists of infected people divided by the number ofdays each is infected two differential equations, one for the core group (group 1) or and one for the noncore group (group 2). In a given group, the equation describes the net effect of two flows, one from the pool of susceptibles into the pool of infectives (the (2) F Newly cured 1 I1Ni incidence of new infections) and one in the reverse direc- L group i ; D tion as infectives are cured of their disease and again become susceptibles. Note that the model ignores the possibility where D is the average duration of infectivity in days. that an individual's sexual behavior might change so that The rate of new infection is more complex and requires he or she moves from the core group to the noncore group, assumptions about the following: a, = the probability of a new or vice versa. Such behavioral changes must be imposed on sexual contact on a given day by a group i person; G = the model from the outside. selectivity of partner choice; Q = probability of infection on a Let 1, represent the proportion of group i that is infected on single contact. From these assumptions can be computed:59 a given day, where (i = I) is the core group and (i = 2) is the noncore group. Suppose that the number of individuals in the reproductive rate for group i: group i is N. Then the sizes of the pools of infectives and susceptibles on any day are, respectively, iN, and (1 - I,)N,. R, = Q LD a, The basic differential equation for group i is an equation for the rate of change of 1,N,, or, because N, is constant, for the proportion of new partners in group i: 512 Mead Over and Peter Plot the "mixing" coefficients:60 equations tends toward equilibrium values of the two preva- lence rates under general conditions. N, a, The simulations in the text constitute sensitivity analysis N1 a, + N2 ax ' with respect to the disease parameters, because those vary substantially across the six simulated diseases. Also reported in (I (I-G)hbi+G if i =jthe text and in table 20-15 is a sensitivity analysis of the main M,j l (I - G) b) if i # j result on targeting with respect to two dimensions of sexual behavior, selectivity and rate of partner change. Because selec- tivity is a particularly interesting parameter of sexual activity, Finally, a set of parameters, Cy, are defined as follows as and because it could itself be the object of policy, we report functions of those defined above: here on sensitivity analysis of the main results with respect to the entire range of possible selectivity coefficients. (3) C1J = M, 'R = M,, Q a,. Consider the effects of increasing the selectivity coeffi- D cient above the value of zero assumed in the text. It is clear (and confirmed by the simulations) that there is a mono- These parameters allow us to write the expression for the rate tonic increasing relationship between the value of the se- of new infections per day as: lectivity coefficient and the speed with which the core group attains its equilibrium prevalence rate. Furthermore, Newly the level of the core's equilibrium prevalence rate increases (4[) infected ]= [CII + C1j 1j] (I - i) N, V i j as G increases. By symmetry, one might expect that both group i the speed with which the noncore approaches equilibrium and the level of that equilibrium would decrease monoton- Thusthetwoequationsofmotionforoursimulationmodelare: ically with increases in G. Only the second of these two expectations is confirmed. dli it It is shown in figure 20B-1 that, although the long-run (5) c1t =(Cii Ii +Ci? 12) (I - l) -D ; equilibrium prevalence in the noncore is lower at G = 30 percent than at G = 0 percent, the equilibrium rate is ap- proached more quickly. How could this be the case if the relatively uninfected noncore (starting at a prevalence rate of (6) d = (C2 + C21 I I2) - only I percent) is now selecting a smaller proportion of its Figure 20B-1. Effect on Evolution of Epidemic in Noncore Group from Increasing Selectivity: Sensitivity wilth Respect to Selectivity iGonorrhea and Syphilis One special case of the model is worth examining. Suppose Proportion infected that members of each group are so selective in their choice of partner that they always choose members of their own group, so that G 1.0. Then by the definitions of M, and Cy, the two equations of motion simplify to: 0.20 - d 11 Ri 1 0.15 - By setting this equation to zero we solve for the equilibrium value (denoted by an asterisk) of the prevalence rate 1, 05 I (8) I= *(Ri-I) 0 RIl 1990 1991 1992 When G is not equal to 100 percent, no analytical solution --- --Gonorrhea:Opercent ---- Syphilis:0 percent exists for the two simultaneous quadratic equations in the - - - - Gonorrhea: 3 percent Syphilis: 30 percent unknowns I I and I '. Lajmanovich and Yorke (1976), how- ever, have proved that the numerical solution of the two Source: Authors. HIV Infecuon and SexuallY Transmitted Disecases 513 Figure 20B-2. Effect of Selectivity Coefficient the worst selectivity value for syphilis and HIV with ulcers is 20 on Present Value of Averted Case-Years in percent. Selectivity must be increased to 40 percent before the Noncore Group three-year burden on the noncore of these two diseases would be reduced. Discounted years of illness (thousands) 2 - Interaction of STD and HIV Epidemics The model presented in equations (1) through (6) above 1 represents the dynamic of a single STD, in the absence of any risk factors other than sexual activity. As reviewed in the text, however, there is substantial evidence that infection with o - - - - - - - - - - - several of the CSTDs increases the transmission probability of Oi_v infection. When both diseases exist in a sexually active ._- :::: ~~~~~------------- population, preventing or curing an STD will avert cases of HIV infection even in the absence ofan intervention aimed directly -t _ at Hiv. To estimate the magnitude of this effect requires that the Iuiv epidemic be simulated simultaneously with the epi- -2 _______l_l__I__l_l_ldemic of one of the CSTDs. In such a simulation, the probability 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 that a contact between an HIV-infected person and an FIIV- susceptible person will result in an infection will depend on Selectivity coefficient the probability that one or both of these individuals is suffering from the STD on the day of the contact. - -Chancroid --------Chlamydia We assume that i-nv infection has no effect on the simulta- - - - - Gonorrhea ---- HIV with ulcers HIV without ulcers Syphilis neous epidemic of a CST). Thus, simultaneous modeling of the two epidemics affects the value of Q in equation (3) and therefore the values of the C,) in equations (4) through (6) for Source: Authors. the HIV epidemic only. Let VI be the instantaneous prevalence rate of the STD in the core group and V2 be the rate in the partners from the more heavily infected core (starting preva- noncore group. Then, the probability of a new HIV infection lence of 20 percent)? The explanation is in the dynamic behavior of the sero- Figure 20B-3. Effect of Selectivity Coefficient prevalence rate in the core group. In the early stages of a on Present Value of Averted Case-Years in gonorrhea or syphilis epidemic, a higher selectivity makes the Core Group prevalence rate of the core group increase much more rapidly. As a result, even though fewer persons in the noncore are contacting core individuals, this effect is more than offset by the fact that more of the core individuals are infected. Thus 30 the prevalence rate of the noncore climbs faster than ir would if G were 0 percent. ,' Figures 20B-2 and 20B-3 display the effect on the core and 20 - noncore, respectively, of alternative selectivity coefficients from 0 to 100 percent. Both figures display on the vertical axis 1 5 the present value of the case-years of illness that would be averted in each group by changing the selectivity coefficient 10 - - at the beginning of the epidemic from 0 percent to the level 5 - - - indicated on the horizontal axis. Note from figure 20B-2 that -:-:: .. increasing G increases the burden on the core group for all I - ; x diseases, the effect being greater at higher selectivity coeffi- 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 cients. Now examine figure 20B-3. Select;vity coefficient For small increases in selectivity, the burden of illness of four of the diseases is greater on the noncore than it would be with --------- Chlamydia ---- HIV with ulcers a zero selectivity. Because of the more rapid early increase in - - - - Gonorrhea HIV without ulcers the epidemic demonstrated in figure 2QB-1, the three-year Syphilis -- Chancroid burden on the noncore actually gets worse when selectivity is slightly increased above zero. Within this three-year horizon Source. Authors. 514 Mead Over and Peter Piot on a given sexual contact, Hy, varies over time with the STD maintain the link between the CSTD epidemic and the HIV prevalence rate according to: epidemic by maintaining the substitution of equation (10) for the C, parameters. The result is that an intervention which (9) H (I - V) (I - Vj) h protects 100 individuals has three beneficial effects: (a) it directly prevents cases of the CSTD being modeled, (b) it + (I -V,) VJ a h directly prevents transmission of the concomitant HIv epi- demic, and (c) it indirectly prevents HIV infection through the + V, (1 - V,) a h mechanism modeled in equations (9) and (10) above. + V V' h, Appendix 20C. Management of Selected Classic where h is the probability ofHIV infection without concomitant Syndromes STD and a is the multiple by which h is increased when either Effective therapy is available for all bacterial CSTDs. The main of the two partners is infected. Note that we assume a multiple issues are discussed below.61 of 2 when both partners have the STD. Using equation (9), the coefficients of the equations of Management of Gonorrhea and Chlamydial Infection motion for the Hiv epidemic are modified to become: Isolation of the etiologic agent is the optimal method of (10) C1 = M,, HI) a,. diagnosis but is technically demanding and expensive. Though nonculture methods for the diagnosis of both genital infections Substitution of equation (10) into equations (5) and (6) yields are now available, they are as yet rarely used in the developing equations of motion for HIV infection which are sensitive to a world. A decrease in their cost may increase their use and have simultaneouis STD epidemic. a significant effect on the diagnostic capacity for STDs. Oral tetracycline taken for seven to fourteen days is the Reduced Transmission Probability treatment of choice for chlamydial infection and has a cure rate of85 to 90 percent. Oral erythromycin at the appropriate dosage In the main part of the text, we presented sinmulation results shoulid be given to pregnant women and infants and children. for an intervention which consists of assuring that all sex In those areas in which gonococci are still susceptible to contacts by 100 individuals in one of the two groups are penicillin, procaine penicillin as a single intramuscular injec- protected for a year. To model this intervention, let UI be the tion together with oral probenecid has a cure rate of 90 to 95 proportion of individuals who are unprotected in the core percent, as has oral amoxicillin. In most areas in Sub-Saharan group and UL- be the proportion unprotected in the noncore. Africa and in Southeast Asia, however, more than 50 percent Then the modified versions ofequations (5) and (6) above can of gonococcal strains are highly resistant to penicillin, as well be written: as to the tetracyclines in 20 to 30 percent of cases. Single-dose intramuscular injections with third-generation cephalospo- (11) i= I l 1 i+C12 U?Ij. U (I-1t) - _1 rins, sLch as ceftriaxone sodium or cefotaxime sodium, or oral d t V - - C D therapy with the new quinolones, such as norfloxacin and ciprofloxacin, have virtually a 100 percent cure rate. Spectino- (12) d12 22 L2 + U I 1 U2 (l 12) -12 mycin cures 95 percent ofcases ofgonorrhea, including those d t C ) D caused by penicillin-resistant strains. These are presently the recommended drugs for treatment of gonorrhea, but less- The base simulation is run with both Ul and U2 set to 1.0. To effective alternatives (cure rate below 95 percent, depend- run the simulation of 100 protected individuals in the core, we ing on the antimicrobial susceptibility of local strains) are set U] to 0.9 and maintain the value of U2 at 1.0. To run the being used because they are less expensive, including thiam- simulation of 100 protected individuals in the noncore, we phenicol, sulfamethoxazole trimethoprim, and kanamycin return Ul to 1.0 and set U2 to 0.998. In both cases we use (WHO 1989h). equations (11 ) and (12) to track the epidemic for one year and then continue the simulation for nine more years with both U Urethritis in Men parameters reset to 1.0. We model each CSTD jointly with HIV infection by using Because basically only two etiological entities have to be equation ( 10) to modify the transmission probability of the considered for urethral discharge in men, a simplified method simultaneous Hiv epidemic. We assumne that the protected of management has heen used widely. In figure 20C-1 we individuals are simultaneously protected from both concomi- present such an algorithm and include figures on its effective- tant epidemics by introducing the UL, and U2 parameters in the ness (WHO 1991). The selection ofthe antibiotic should ideally same way in the equations for both epidemics. In addition, we be based on the sensitivity of local strains of N. gonorrhoeae, HIV Infection and Sexually Transmitted Diseases 515 Figure 20C-1. Algorithm: Urethral Discharge A simplified clinical algorithm using a clinical examination (in the Absence of Laboratory Support) with visualization of cervix but no microscopy is shown in figure 20C-2. Several other algorithms have been proposed, but evaluations of them have not been published and their Urethral discharge 1 effectiveness is unknown (WHO 1991b). confirmed by clinician I Pelvic Inflarinatory Disease Treatment for gonorrhea and The objectives of PID management are twofold: cure of PID chamydial infection Health and prevention of tubal infertility and ectopic pregnancy. Examine and treat partner(s) a Pelvic inflammatory disease has a polymicrobial etiology I and its clinical expression includes a variety of fairly non- Follow-up, seven to fourteen b days after treatment Figure 20C-2. Algorithm Vagind Discharge (Speculum Examination Possible, but no Laboratory Support) CinCW mmw Discharge persists Vaginal dischargea Speculum examination Assess: treatment compliance Assess: treatment compliance good and reinfection unlikely bad and/or reinfection likely Profuse, runny or Whi rd likeb Inspect and clean Reler | a 8bRrt pWI -osh l malodorous vaginal e, cud-ige cervix with large I ~~~~~~~~~~~dischargeb cotton swab a. Notification and treatment of female partners of men with urethritis are of the highest priority as one of the best ways of identifying women at high risk of having asymptomatic gonococcal and chlamydial infections. Treatment for Treatment for Mucopus in cervix b. Patient may be advised to return only if symptoms persist. trichomoniasis/ catm Source: WHo 1991b. bacterial vaginosis cand idcastios Health education Hat dcto and counselling and counselling Treatment for gonorrhea and besides such other considerations as cost, availability, and chlamydial infection mode of administration. In addition, the relative frequency of Examine and treat gonococcal and nongonococcal urethritis in the patient pop- Health education ulation should be taken into account. Microscopy of a Gram's and counselling stained smear of the discharge represents the minimal standard Follow-up seven to fourteen of laboratory examination. days after treatmentd Vaginal Discharge C c Because the etiology, and consequently the therapy, of cervico- D lrs vaginal discharge is complex, clinical algorithms have not been successful for the management of this syndrome. The IAssess: treatment compliance Assess: treatment compliance primary objective of case management of this problem should I good and reinfection unlikely good and reinfection unlikely be the diagnosis and treatment of gonococcal and chlamydial cervical infection and the identification of women with an |Refer Starl grotocol again associated PID. Though a simple "swab test" had an acceptable sensitivity and specificity for the diagnosis of mucopurulent cervicitis-mainly caused by C. trachomatis and N. gonorrhoeae-in one study in Seattle, Washington, its validity a. If vaginal discharge is accompanied by lower abdominal pain or pain on limitedin fied studes in frica wth a snsitivty and moving the cervix, use the appropriate 'lower abdominal pain' algorithm. was limited in field studies in Africa with a sensitivity and b. In addition, the pH paper test can be used: if pH lower than 4.5, treat for candidiasis; if pH higher than 4.5. treat for trichomoniasis/bacterial vaginosis. specificity for chlamydial and gonococcal infection combined c In the absence of a confirmed diagnosis, the decision to notify partner(s) of less than 50 percent, respectively (Brunham and others should take into account local cultural and epidemiological factors. d. Patent may be advised to return only it symptomatic. 1984; Braddick an-d others 1990). Source: WHO lgglb. 516 Mead Over and Peter Piot specific signs and symptoms, such as lower abdominal pain and include tetracycline ointment for ten days plus kanamycin or tenderness, malaise, fever, and adnexal tenderness. When cefotaxime as a single injection, the cure rates of which are 90 validated by laparoscopy, the sensitivity and specificity of a to 95 percent (Fransen and others 1984; WHO 1985a and clinical diagnosis of PID, particularly cases caused by C. 1985b).Aregimenoftetracyclineointmentpluserythromycin trachomatis, are found to be in the range of 65 to 85 percent syrup for ten days is the recommended treatment of nongono- (Jacobson and Westrbm 1969; De Muylder 1986). Valid coccal ophthalmia neonatorum, but in practice only topical diagnostic criteria for postpartum endometritis and PID re- therapy is given. As for other respiratory infections, the diag- main to be developed. Because of the serious outcome, nosis of chlamydial infant pneumonia is extremely complex sequelae, and mortality, early recognition and treatment are and is possible only in sophisticated medical centers. essential, and sensitive criteria for diagnosis should be set even if overtreatment is the result. Infertility and Ectopic Pregnancy Treatment should be directed against infection with N. gonorrhoeae, C. trachonatis, and anaerobes and may consist of Tubal infertility is an irreversible sequela of PID and may be a regimen of spectinomycin single-dose plus tetracycline and repaired with microsurgery at a low (<10 percent) rate of metronidazole taken for two weeks. Cheaper regimens, such as success. In vitro fertilization (rate of success 10 to 20 percent) thiamphenicol, have also been used (De Muylder 1986). The is also used to achieve pregnancy. Both techniques are expen- cure rate with these regimens is not precisely known, but it is sive and rarely available in developing countries. Hospitaliza- presently estimated that 15 percent of women with acute PID tion with surgery, often with removal of the tuba, is virtually fail to respond to initial antimicrobial treatment and 20 per- always required for ectopic pregnancy. cent have at least one recurrence (Brunham 1984). The effect of antimicrobial therapy on long-term tubal function is un- clear, but several studies have shown that women treated within two days of the onset of symptoms have a lower inci- Though the etiology of genital ulcer disease is diverse, clinical dence of tubal occlusion than do women treated later in the algorithms for management have been used with success in disease (Westrom and others 1979). Hospitalization is required in a substantial but ill-defined proportion of women with PID in developing countries. Up to Figure 20C-3. Algorithm: Genital Ulcer 25 percent of PID patients were hospitalized in Zimbabwe in (without Laboratory Support) 1986 (X. De Muylder, personal communication 1989). Perito- nitis, a pelvic mass, and a tubo ovarian abscess are the main reasons for hospitalization and require intravenous therapy and Genital ulcer often surgery as well. i l t Neonatal Infection with N. gonorrhoeae and C. trachomatis [Vesicular lesion(s) present or _ history of vesicular lesions, Ye Hs manaement Ioften recurrent? The objectives of the management of ophthalmia neonatorum are the identification of cases of gonococcal infection; the No clinical cure ofconjunctivitis; the prevention ofvisual impair- I and chancroida ment and blindness (a complication of infection with N. Examine and treat partner(s)b gonorrhoeae and, to a lesser extent, C. trachornais); and the Health education and treatment of STDs in the parent. The proportion of cases of counselling neonatal conjunctivitis caused by N. gonorrhoeae depends mainly on the use of effective eye prophylaxis at birth. Treat- Follow-up after seven days ment for both gonococcal and chlamydial (nongonococcal) ophthalmia should include a systemic antibiotic, because ex- traocular infection and disease (pneumonia) are common. Improving? Yer The accuracy of a stained conjunctival smear is high for the No differentiation between gonococcal and nongonococcal oph- l I thalmia (Fransen and others 1986), and this method should be Rferfl used for every case of purulent conjunctivitis in the first week of life, because these have a high probability of being gonococ- cal. Single-dose ceftriaxone administered intramuscularly ($5 for 125 milligrams) has a 100 percent cure rate for gonococcal a. Combined treatment for both syphilis and chancroid is recommended, conjunctivitis but is not available everywhere (Laga and others except in areas where chancroid is very uncommon. Where granuloma inguinale occurs, treatment for this condition should also be considered. 1986a and 1986b). Other recommended treatment regimens b. In the absence of a confirmed diagnosis, the decision to notify partner(s) should take into account local cultural and epidemiological factors. for this indication in areas with penicillin-resistant strains Source: WHO 1991b. HIV Infection and Sexually Transmitted Diseases 517 several settings. The basic objectives of these algorithms are Cost-Effective Treatment Strategies cure of chancroid, the most common cause of GUD; and treat- ment of syphilis, the most severe cause with respect to sys- In order to devise recommendations for a cost-effective temic complications and sequelae. Though genital herpes is approach to controlling STDs, it is necessary to estimate the a common cause of GUD throughout the world, specific therapy approximate average cost and the approximate effectiveness is not considered in these algorithms because the high cost of of treating a case of each sTD (Washington, Browner, and antiviral treatment with acyclovir makes it rarely available. Korenbrot 1987). In a developing country, where diagnostic Early diagnosis and treatment of GUD has become critical in materials are difficult to obtain and the diagnostic test may populations with HIV infection, because genital ulcers have cost five to fifty times the cost of the drug treatment, part been shown to increase the rate of transmission of HIV (WHO of the choice of treatment is the decision whether or not to 1989a). condition treatment on a positive test for the disease. Here A management algorithm for GUD is shown in figure we consider two alternatives. The first alternative is simply 20C-3 and includes probabilities for the different outcomes. to treat everyone in a given population regardless of The simultaneous presence of HIV infection, however, may whether they have symptoms and without attempting to significantly decrease the cure rate of both chancroid and diagnose the disease. The second alternative is to screen the primary syphilis when the recommended single-dose treat- population and then treat only those who test positive. For ment regimens are used (Lukehart and others 1988; Cam- some diseases, the option of screening only by clinical eron and others 1989). examination is compared with the option of using micros- copy or of using a culture. Syphilis In table 20C-1 we present basic comparative cost data on the various diseases. In the costs for treating a given disease we include the drug cost per treatment episode and the number of Whereas the treatment of the various stages of syphilis is fairly simple and well standardized and has cure rates approaching minutes of clinic time per treatment episode. For the classic 100 percent for primary, secondary, and latent syphilis, its STDs, the latter cost typically is the duration of a single encoun- diagnosis is more problematic, requiring the use of laboratory ter between patient and medical practice. For AIDS and Hiiv digoi is mor prbeai, reurn the us oflbrtr nfection, however, the number of minutes is an estimate of tests. Nevertheless, assays such as the rapid plasma reagin tests the total number of minutes in all encounters with a medical are inexpensive (+ $0.40 per test) and easy to perform, and tie inta er of tel mnths. Intels threeicol they should be available as a minimum laboratory support for practice in a period of twelve months. In the last three col- the management of STD and the screening of pregnant umns, we give three different estimates of the cost per treat- thmen. mang e enstiitD and tet ,sce ing of , ment for each of the classic STDs. If we denote the drug cost by women. Thneir sensitivity for infectious syphilis beyond x h ubro iue yT n h rprino ue the primary stage (where the algorithm for genital ulcer Rx, the number of minutes by T, and the proportion of cures disease can be applied in most instances) is virtually 100 thate fecthe relathip percent. ~~~~~~~~~~~~calculated from the relationship: percent . Treatment of syphilis of less than two years' duration and of T neonates bom to mothers with a positive serological test for Rx + H, syphilis consists of one or two injections of benzathine peni- (13) C, 60__ cillin G. Therapy of neurosyphilis and cardiovascular syphilis E consists of a course of penicillin, in addition to palliative measures often requiring hospitalization. where H, is the assumed cost per hour of clinic time, which varies from $2 to $10 to $30 across the three columns. Table 20C- 1. Average Cost per Treatment of Case of STD by Disease and Cost per Clinic Hour Treatmnent by cost per clinic hour Drug cost Clinic time Disease Sex (U.S. dollars per episode) (minutes per episode) $2 $10 $30 AIDS with AZT Both 2,000 2,400 2,080.00 2,400.00 3,200.00 AIDS without AZT Both 35 1,200 75.00 235.00 635.00 Chancroid Both 0.15 10 0.48 1.82 5.15 Chlamydia Both 0.25 15 0.75 2.75 7.75 Gonorrhea Female 1.00 12 1.40 3.00 7.00 Male 1.20 10 1.53 2.87 6.20 miv with AZT Both 2,000 120 2,004.00 2,020.00 2,060.00 Syphilis Both 0.80 15 1.30 3.30 8.30 Source: Estimates by authors. Table 20C-2. Alternative Diagnostic Approaches for sTDs: Costs, Time, and Positive Predictive Value Posinve predictive value Diagnostic Cost of test Lab or clinic time I percent 5 percent 25 percent Disease Sex method (ULS. dollars per test) (minutes per test) Sensitivity Specificity prevalence prevalence prevalence AIS with AZT Both Serology 2.00 10 98 98 0.33 0.72 0.94 AIDS without AZT Both Serology 2.00 10 98 98 0.33 0.72 0.94 Chancroid Both Clinical 0.00 10 80 60 0.02 0.10 0.40 Both Culture 5.00 5 70 100 1.00 1.00 1.00 Chtamydia Female Clinical 0.00 10 20 20 0.00 0.01 0.08 Female Culture 12.00 5 80 99 0.45 0.81 0.96 Female Antigen 5.00 5 70 95 0.12 0.42 0.82 Male Clinical 0.00 10 80 80 0.04 0.17 0.57 Male Culture 12.00 5 85 99 0.46 0.82 0.97 Gonorrhea Female Clinical 0.00 10 25 25 0.00 0.02 0.10 Female Culture 5.00 5 85 100 1.00 1.00 1.00 Female Microscopy 1.00 10 50 80 0.02 0.12 0.45 Male Clinical 0.00 10 85 90 0.08 0.31 0.74 Male Culture 5.00 5 95 100 1.00 1.00 1.00 Male Microscopy 1.00 10 95 99 0.49 0.83 0.97 HIV with AZT Both Serology 2.00 10 98 98 0.33 0.72 0.94 Syphilis Female Serology 1.50 5 95 95 0.16 0.50 0.86 Male Serology 1.50 5 95 95 0.16 0.50 0.86 Average - - n.a. n.a. 0.34 0.54 0.73 - Data not available. n.a. Nor applicable. Source: Authors' estimates. Table 20C-3. Minimum Cost per Effectively Treated Case of STD: Sensitivity to Prevalence and Cost per Clinic Hour $2 per clinic hour $10 per clinic hour $30 per clinic hour I percent 5 percent 25 percent I percent 5 percent 25 percent I percent 5 percent 25 percent Disease Sex Diagnostic method prevalence prevalence prevalence prevalence prevalence prevalence prevalence prevalence prevalence AIDS with AZT Both Serology 6,521 2,934 2,217 7,623 3,405 2,562 10,380 4,584 3,424 AIDS without AZT Both Serology 465 152 89 1,084 401 264 2,632 1,024 702 Chancroid Both Clinical 73 15 3 333 67 14 983 199 42 Both Culture 821 165 33 928 187 39 1,196 244 53 Both Presumptive treatment' 304 61 12 452 90 18 822 164 33 Chlamydia Female Clinical 516 101 18 2,139 420 77 6,196 1,219 223 Female Culture 1,692 339 68 1,789 360 74 2,033 413 89 Female Antigen 827 166 34 951 192 41 1,260 258 58 Female Presumptive treatment' 322 64 13 544 109 22 1,100 220 44 Male Clinical 63 13 3 286 59 13 845 172 38 Male Culture 1,592 319 64 1,682 338 69 1,908 386 82 Male Presumptive treatment' 304 61 12 452 90 18 822 164 33 Gonorrhea Female Clinical 580 114 20 1,643 324 60 4,301 848 158 Female Culture 641 129 27 726 148 32 936 193 45 Female Microscopy 341 69 14 690 139 29 1,562 316 67 Female Presumptivetreatamenta 295 59 12 463 93 19 884 177 35 Male Clinical 62 13 4 245 51 12 702 145 34 Male Culture 574 116 25 649 132 29 838 173 40 Male Microscopy 151 31 8 302 63 15 678 141 33 Male Presumptive treatmenta 288 50 12 428 86 17 779 156 31 HIV with AZT Both Serology 6,291 2,829 2,136 6,475 2,878 2,159 6,936 3,002 2,215 Syphilis Both Serology 185 38 9 269 56 14 477 102 27 Both Presumptive treatmenta 293 59 12 495 99 20 1,000 200 40 Note: Cost-effectiveness values are those that would obtain if 100 percent of patients had the specific disease in question. Treatment effectiveness is 99 percent for syphilis, 95 percent for gonorrhea, and 90 percent for chanchroid and chlamydia. For AIDS treatment is assumed 100 percent effective at prolonging life one year without AZTor two years with AZT. a. Treatment of all patients with the drugs for all four STDs, at an estimated cost of $2.40 per visit. Source: Authors' estimates. 520 Mead Over and Peter Plot In table 20C-2 we present the estimated cost of the labora- strategy will dominate the entire domain of reasonable param- tory test for each disease, the estimated sensitivity and speci- eter values. ficity of each test and, in the last three columns, the positive In table 20C-3 we present the lowest cost solution for each predictive value of the given test for three different prevalence disease and, for those diseases which vary by sex, for each sex. Two rates. Let x represent the sensitivity of the test (that is, the remarkable findings stand out from this table. First, over a large proportion of truly positive cases which the test finds positive) range of parameter values the "treat without testing" option and y represent the specificity of the test (that is, the propor- dominates the "screen, then treat" option. Second, in those few tion of truly negative cases which the test finds negative). situations in which screening is more cost-effective than treating Then the positive predictive value of a test is defined for everyone, the clinician's judgment produces a lower cost per prevalence rate I to be: effectively treated case than do the alternatives, which in the case of gonorrhea include microscopy and a culture. (14) PPV! x I (x P + (1 -Y) (1 - I )] Notes To compare the "treat all" strategy with the "screen" strat- egy, we calculate the estimated cost per cure using each strat- All dollars in this chapter are 1989 U.S. dollars. egy for each of the classic STDS. For AIDs and HI-V infection we l.In his seminal work on the henefits of a syphilis control program, Klarman (1965) was unable to estiniare these dynamic benefits. simply calculate the cost of treatment. LetN be the number of 2.HethcoteandYorke(1984.pp. 16andl7)usetheterms"contactnumber" individuals in a given population and I be the proportion of and "infectee number" to refer to different versions of this concept. individuals infected at a given moment, that is, the point 3. The concept can be extended further by including terriary, quarernary, prevalence rate of the infection. Then the cost of treating all and other cases. This extension presents problems because some of the indi- N individuals will be C . N. Because a proportion I is infected, viduals not infected by the prevented case will instead be infected by a case that has not been prevented. The greater the number of rounds considered, and the treatment is effective on a proportion E of these, the the more serious is this problem, which can be avoided only by addressing the number cured will be E - I . N. Hence, the cost per cure for the problem more generally in the context of an explicit mathematical epidemi- treat-all strategy is C/(E - 1). ological model. Prevented cases that would have occurred more than twelve The cost of testing all N individuals and treating those who months in the future must be discounted by an appropriate rate. . . . ~~~~~~~~~~~~~~~~~~4. The new equation for Rbecomes R =QD Im +s 2/m]l where mis the mean are positive on the test is the sum of these two cost compo- ofTee,qainoReoeR=Dms/J hrmstema of the a, and s is their variance. Note that as the heterogeneity disappears, s nents. At a cost of T dollars per test, the testing cost for N approacheszeroandthisequationapproachestheonegivenearlier.Thisresult individuals is T - N. Let T be defined analogously to C depends on the assumption of proportionate mixing of sexual partners. See except that the cost per hour of laboratory or clinic time appendix 20B for a discussion of altemative models of mixing. is set to half the value used to compute the treatment cost. 5. The "prevalence" of an infection refers to a point in time and is defined The test will be positive for x proportion of the I N who as the proportion of a given population infected at that moment. The "inci- dence" of infection refers to a specific period of time and is defined as the are truly positive and for (I - y) proportion of the (I - I) N numberof new casesof infectionduringthat period. Economistswill note that who are truly negative, so the total cost of treatment will the former is a "stock" and the latter a "flow." be C [x I N + ( 1 - y) ( 1 - I) N]. Of the x I . N-infected 6. See Heymann (1990) for sensitivity analyses with respect to several people who are treated, a proportion E will be cured. Thus parameters not considered by Kennedy and others (1990). the cost per cure from the screen strategy is given by: 7. None of the three studies reported in table 20-6 breaks out the results by sex. This is more of a problem for the Rwandan and Zambian studies than for the Zairian one, because the latter is based on a group of employees which is Cost per cure C [x . I + (I - 4) (1 - 1)] + T almost entirely male. (15) of the screen = 8. Rural women are much less likely to attend secondary school in Africa strategy E x I than are urban women. 9. Data on urban sex ratios of adults age twenty through thirty-nine are available only for these eighteen Sub-Saharan African countries in Larson (1989, table I). C T 10. The number of females enrolled per 100 enrolled males is drawn from E - PPV E . x - I World Bank 1989 (table 32). Secondaryeducation is likelytobeconcentrated in urban women. A comparison of the cost per cure of the two strategies then I1. The regression equation is: revealsthepreferredstrategyforanygivendiseaseandforgiven HIVE% = -4.3 -0.10 FENR + 1,240 IRH assumptions regarding the hourly cost of clinic time, H, and (-0.8) (-2.3) (3.1) the prevalence rate in the population in question, 1. Other 2 things being equal, a higher cost of clinic time in relation to n = 18 R = 0.48 F2.15 = 7.03 Prob > F 0.007 testing cost, T, increases the relative cost advantage of the testing cost, T. increases the relative cost advantage of the where FENR is the number of girls enrolled in secondary school for every 100 screen strategy. Conversely a higher prevalence rate, other boys, FURB is the number of females age twenty through thirty-nine resident things being equal, reduces the relative cost advantage of in urban areas for every 100 males. Numbers in parentheses are t-statistics. screening. The specific assumptions determine whether one Dropping Rwanda from the regression reduces the significance of the coeffi- HIV linfection and Sexually Transmitted Diseases 52) cienits but does not chiange their sign oir magnitude. (Grossbaird-Schechtman, possibilities include discounted disability-adjusted life-days saved, discounted DuCharme, and Loomin ( 1989) also obtain significant coefficients on ihe productive disability-adjusted life-days saved, and money-valued discounted young adult femrale-to-mnale ratio in their study of gonorrhea and syphilis disability-adjusted life-day-s saved. prevalence in a cross-section of thirty-seven U.S. cities. They did not attempt 28. ForTmost diseases, we use relative incidence rates to average the effects to control for female educatioin but did c.ontrol for the prevalence of male overthe six age groups. FollowingG.HAP(1981 ), wedistinguish aduttfromchild homosexuality, pneumonia. At the expense of greater complexity, it would be possible to 12. The magnitudes of the estimated coefficients SuIggeSt that a tive-point perform a separate analysis for each age group. reduction in the aduLt prevalence OfHiV could he achiieved either by increasing 29. In rthe absence of dynamic analyses of the health benefits of preventing PLR1n by 35 per I100 mnen or by increasingtFENl by 50 girls per 1 00 boys. theothercommunicable diseases in fgure 20-lIfland table 20A-3l(forexample, 13. Somnetimes it will be inore convenient to refer to a disability-adjUsted nieasles, pneumronia), dynamic benefits of aTtDscanniothbecompared with those life-day (PALP) than a disability-adjusted life-year. There are 365 H)ALDs in one oif other diseases. isALY. ( ., I .'i i, used rhe rermn healthy life-year, whichi we have changed 30. In 1981 the country of Burkina Faso introduced the activist expression for conformnity wAith uisage in thiis volumie, "commando prograni" for a one-time vaccination campaign. 1 4. This is the parameter referred to as it previously and in table 20- 10 below. 3 1. For ai full description of the model and sensitivity analyses with respect 15. This is referred to in the literature as the assumiption of proportioniate to its key parameters, see appendix 20B. miixing and is defined bv setting a "selectivity parameter" equial to zero. When 32. The net effect of the Intervention, however, remains strongly positive, this parameter, G, is set to 100 percent. eaCh grouIp prefers members of its own becautse the averted cases early in the period more thani outweigh the smiall grouip toI the coimplete exclusion of membiers of thie other, so that: there is no nUmberiofaddirional cases later. The effectoffuture additional cases is reduced interaction between thie two grouips. Although proportional mixing probably eveni funther because they are discounted at 3 perCent. does not exactly describe actual behavioral patterns (AnderSon1 1989), 33. Effectiveness is noit the whole story. Cost-effectiveness is discussed in Rothenberg and R,itterat ( 1988, table 3) provide evidence fronti a Colorado the rest of the chapter. Springs cohort oif a remarkably large degree oifmixing, especially by coire men, 34. An Intervention which produces sustained behavioral change would fuot whom only 74 of 251 sexual contacts were with core women. The actual yield additional benefits froim the permanent reduction in the prevalence degree of miixinig in pattern) 11 coutjitries is currently unknownl. rate. 16. Although the paramneters oftable 20-10 are niot cuirrently known for any 35. Wheni both partnerS are infected with one of these fourCSTtas, we assume African poipulatioin, the guesses here are rouIghly consistent in order ofmagni- that infectivity is miultiplied hv the square of the factort operating when only tuide with the estimates oif doubling time of the infectioin presented by one is infected. See aippendix 2ZOB for a formal statement. Anderson and MIcLean (1988). 36. These sums are represented graphically by the areas tinder the two curves 17. The case-fatality rates of thie other Csn,J~,, although important for the in the boutons panel of figure 20-14. static buirdeni calculations presented above, are too) smiall and too lonig after 37. For simnplicitv we model the condoim user as absolutely protected, infection to have an appreciable influence on this short-term model and thus ailthough in fact some condoms break or are incorrectly used and thus do not are assumed to be zero. For HIV we set D equal to ten years, the average time protect absolutely. Given the (unknown) frequency of condom failure, the uintil death, at which point We aSSumITe that a new susceptible is recruiited cost-effectiveness figures can be inflated by that amount. through migration on matuirationr to replace thie dead Individual. 38. The others are frequency of sexual partner change (a), selectivity (G), 18. New biomedical findings on the variation of inifectivity durinig the and theCSTD's durationofinfectivtty ID). The costs ofchaniging the frequency period of infection may affect these results, and duration are much harder to calculate. Theoiretically an iEcCprogram could 19. For other v-alues, of the selectivity coefficient G3, the two equilibrium also slow the epidemic in the noncore (and accelerate it in the coire) by prevalence levels are the solutions to tw'o sinmultaneous quadratic equations, encouraging increased selectivity oif sexual partners. Sensitivity analysis re- which are solved numierically by the simulations reported below'. For GS = 0 the ported elsewhere in this chapter, how'ever, reveals that even a substantial solutions are given in the last Two rows of table 20-Il1. increase in selectivity, has a relatively small effect on the speed of the 20. Furthermnore there is somne evidenice that other STI)s, suchi as chlamydial epidemic's spread. Furthermore, IEC programs toi encourage selectivity would infection and trichoinoniasis, also increase the efficiency of ti1v transmisscion inevitably also lead to increased stigmatization of peoiple who seem to, fit the (Laga and oithers 1 989, 1990; Pepin and others, 1989). programi's stereotype of "partners to avoid." In addition to creating deep and 2 1. A further question is whether HiV' and oihersTo)s discriminate between harmful social divisions and scapegoating, suich stigmati2atioin would drive the the poor and the rich in uirban areas. Limited evidence indicates that HIV epidemic further underground, where it would be even harder to combat. FOr prevalence r-ates may be positively coirrelated with) Income levels ansong urban) these reasons w'e do not coinsider this behavioral parameter to be the legitimate males and negatively coirrelated with income levels amiong urban femnales. This target oftan IEC programn. may, however, be only a temporary featUre, because the epidemic has not yet 39. Bertozzi ( 1991 ) presents two more complex models and sensitivity reached stable levels in most populations, analyses to show that they differ little from one another. 22. Indeed the authoirs later state that "C-urI parameter valuies ... niay well 40. This is the prevalence rate recently reported among blood donors in lead to coinclusions more representative of say Kinshasa and Nairobi than of Delhi, India, by Singh and others (1990). Zaire and Kenya as a whole' (p. 2 33). 41. In those countries, in which donor assistance has provided virtually 2 3. The last two columns of table 20-12 so'That the health) gain in unlimited financial assistance, the binding constraint is trained, coimpetent preventing a single case of STD in a woman is 1 9percent larger on average than manpower to manage those resources. preventing a single case in a man. 42. This is a consensus estimate by a WHt-is/oPA panel of experts reported in 24. Although antrhopologists have foicused on describing the range and C'hin, Sato, and Mann (1989). meaning of cultural practices, they have not typically attempted to estimnate 43 This is the same Elliot Ness who earlier coinducted such a vigorous and valid population-based frequencies of those praCtiCes. unsuccessful war against bootleg liquor. 25. For example, Forbes magazine recently reprinted and endorsed the 44. Brandt goes on to point out the illogic of placing the blame for STDs oxymorOnic request by the American activist group ACT UP for the public to, uniquely oin women. "The fact that controlling prostitution did not control "demand ... [elffective prevention educationi programs that target all people sexuality forced many to conifront the change in American sexual mores" in all cultural groups" (Atoms Coalition to Unleash Power 1989). (p. 168). 26. Thevyouth grouipentitled"'Les jeuines contre [es MST"`has been organized 45. By placing Htv-infected persons in quarantine, Cuba is the only countrv in Lubumbashi, Zaire. to combat STDS within the peer grotIP. in which a full-scale mandatory program foir the whole population is in efftect. 2?. To the extent that differenit mieasures of health gained result In different There the regulators are presumiably closely supervised (Bayer and Healton decisions, the decisionmaker will also have toi select a preferred measujre. The 1 989). Although the program's effectiveness is undoubtedly increased by the 522 Mead Over and Peter Piot absence of civil rights and by the surrounding ocean barrier at the borders, its Anderson, R. M. 1989. "Mathematical and Statistical Studies of the Epidemi- actual effectiveness is unknown. ology of AIDS." AIDS 3:333-46. 46. An average of 1.9 partners were niamed per cooperating index case. Anderson, R. M., and A. R. McLean. 1988. "Possible Demographic Conse- Interestinigly, when asked to choose whether personally to notify their own quences of AIDS in Developing Countries." Nature 332:228-34. partners (patienit referral) or to have the public health staff perform the Anderson, R. M., and R. M. May. 1988. "Epidemiological Parameters of HIV notification (provider referral), index cases chose the relatively anonymous Transmission." Nature 333:514-19. provider referral for 75 percent of all named contacts (Spencer, Raevsky and, Anderson, R. M., G. F. Medley, R. M. May, and A. M. Johnson. 1986. "A Wolf 1989). PrersonaRy G. F the R an d A. of thnson Immuno- 47. One hundred dollars is 0.5 percent of the U.S. 1987 per capita GNP. Preiminary Study of the Transmission Dynamics of the Human Immuno- 48. In a poor country it may be difficult to persuade the public that any deficiency Virus (HIV), the Causative Agent of AIDS." I,MA Journal of Math- allocation mechanism which distributes treatnent to only five or ten ofseveral emaics Appied i Medicie and Biology 3:229-63. thousand enrolled HIV-positive candidates is fair. One possible mechanism Antal, G. M. 1987. "The Epidemiology of Sexuallv Transmitted Diseases in would be a community review board similar to those used in the United States theTropics." In A. 0. Osoba ed., Sexually TransmittedDzseases in the Tropics. and Great Britain to allocate scarce kidneys to transplant canididates. Another 2 (I) of Bailli&re's Clinical and Tropical Medicine and Communicable Diseaes. possibility is simply to use a lottery. London: Baillire-Tindall. 49. Because no population-based studies yet exist which estimate the Amal, S. O., and K. K. Holmes. 1990. "Sexuallv Transmitted Diseases in the proportion ofadults with multiple sexual partners, these figures must be viewed AiDS Era." Scientific American 264(February):62-69. as hypotheses which require confirmation. The population-based figures, such Arya, 0. P., A. 0. Osoba, and F. J. Bennett. 1988. Tropical Venereology. 2d ed. as proportion of adolescents or of health care providers, can be adjusted to the Edinburgh: Churchill Livingstone. data of an individual country as they become available. Arya, 0. P., S. R. Taber, and H. Nsanze. 1980. "Gonorrhea and Female 50. As an analogy with the epidemiology of screening, think of the commu- Infertility in Rural Uganda." American Jotrnal of Ohstetrics and GNnecology nication as a screening test. Then the coverage of the target group by the 138:929-32. communication is analogous to the proportion ofall positives identified by the test, that is, the senisitivitv of the test. Austin, H., W. C. Louv, and W. J. Alexander. 1984. "A Case-Control Study 5 1. In accord with the preceding note, the "concentration" of a message on of Spermicides and Gonorrhea."JAMA 251:2822-24. a target group is analogous to the positive predictive value of a screening test Ballard, R. C., H. G. Fehler, and P. Piot. 1986. "Chlansydial Infections of for a specific disease. the Eye and Genital Tract in Developing Societies." In J. D. Oriel, 52. If the delivery of an [EC or training message to the wrong group has Geoffrey Ridgeway, Julius Schachter, David Taylor-Richardson, and political or social costs (for example, by inflaminig political antagonism to the Michael Ward, eds., Chlamydial Infections: Proceedings of the Sixth Inter- AID> prevention program or by stigmatizing prostitutes), then the "specificity" national S-mnposium on Human Chlamydial Infections, Cambridge: Cam- of the access group for the target group is also a desirable attribute. bridge University Press. 53. Behmian (1989) discusses the economics of social marketing programs Bamum, H. 1987. "Evaluating Healthy Days of Life Gained from Health in a family planning context. Projects." Social Science and Medicine 24:83 3-41. 54. For exaniple, Mposo and others ( 1989) found that HIV-infected infants Bayer, Ronald. 1990. Private Acts and Social Consequences: AIDS and the Politics in Zaire incurred annual health care costs that were 60 percent higher than of Publc Health. New Brunswick, N.J.: Rugers University Press. control infants. 55. The estimate of one DALY saved per case averred is derved by applying Bayer, R., and C. Healton. 1989. "Controlling AIDS in Ctuba: The Logic of 55~ ~ ~ ~~ ~~~~~~~~-. Th esiae.fn AYsvdprcseaerdi eie yapyn Quarantine. " Newj England]Journial of Medicine 320: 1022-24. the appendix A methodology to gonorrhea after setting to zero all incidence Qr rates for age greater than zero. Behrman, J. R. 1989. "The Simple Analytics of Contraceptive Social Market- 56. This policy is cost-effective in comparison with treating latercases, even ing." World Development 17(10):1499-1521. in populations that have prevalence rates of syphilis as loiw as 0.005 percent Bentsi, C., C. A. Klufmo, P. L. Perine, T. A. Bell, L. D. Cles, C. NI. Koester, in pregnant women (Stray-Pedersen 1983). and S.-P. Wang. 1985. "Genital Infections with Chlanndia trachomrans and 57. The presumptive treatment of syndromes has the additional disadvan- Neisseria gonorrhoeae in Ghanaian Women." Genitounaria Medicine 61: tage of exposing more people to antibiotics and thereby increasing resistance 48-50. to those antibiotics in the population of disease agents and potentially causing Berklevy S. 1991. "Parental Transmission of HIV in Africa." AIDS. 5:S87-S92. mild to severe side effects in individuals taking antimicrobials. Consideration Bertozzi Stefano. 1990. "The Economics of HIV Screening of Blood for Trans- of these drawbacks is beyond the scope of this chapter. fusion." Abstract. Fifth Intenmational Conference on AlDS in Africa, Kin- 58. According to World Bank sources, the cost per clinic-hour in one poor susa, Aire ct. Sub-Saharan African country is $2.20 (1990 U.S. dollars). shasa, Zaire. October. 59. See Hethcote and Yorke 1984, pp. 25-31, fordetails. . 1991. "Combating HIV in Africa: A Role for Economic Research." 60. For a more general specification of mixing in a one-sex mixing model, AIDS 5:S45-S54. see Blythe and Castillo-Chavez 1988 and the references therein. Bewes, P. C. 1973. "Urethral Stricture." TropicalDoctor 3:77-81. 61. Foramoredetaileddiscussion,thereaderisreferredotlrhe STDrtreatment Blvthe, S. P., and C. Castillo-Chavez. 1988. "Like with Like Preference and guidelines issued by WiHO (1989). Sexual Mixing Models." Comell University, Department of City and Re- gional Planining, Comell Universitv, Ithaca, N.Y. Bongaarts, J. 1988. "Modeling the Spread of HIV Infection and the Demo- References graphic Impact of AIDS in Africa." Working Paper 140. The Population Council, New York. Ades, A. E., M. L. Newell, C. S. Peckham. 1991. "Children Born to Wonmen . 1989. "A Model of the Spread of HIV Infection and the Demographic with HIV-I Infection: Natural History and Risk of Transmission." Lancet Impact of AIDS." Statistics in Medicine 8:103-20. 337:253460. Bongaarts, J., P. Reining, P Way, and F. Conant. 1989. "The Relationship AIDS Coalition to Unleash Power (ACT-ULP). 1989. "Statement." Forbes 144(1): between Male Circumcision and HIV Infection) in African Populations." AIDS 20-21. 3:373-77. Aladesanmi, A. F. K., G. Mumtaz, D. C. W. Mabey. 1989. "Prevalence of Bongaarts,J.,and P.O.Way. 1989. "GeographicVariationintheHivEpidemic Cervical Chlamydial Infection in Antenatal Clinic Attenders in Lagos, and the Mortality Impact of AIDS in Africa." Working Paper 1. Population Nigeria." GenirounnarN Medicine 65:130. Council, New York. HIV Infection and Sexually Transmitted Diseases 523 Braddick, M., and J. Kreiss. 1988. "Mother-to-Child Transmnission of HIV." In Quinn, and P. Piot. 1989. "HIV Infection in Patients with Tuberculosis in P. Piot and J. M. Mann, eds., AIDS and HIV infection in rhe Tropics. London: Kinshasa, Zaire." American Review of Respiratory Diseases 139:1082-85. Bailli&re-Tvndall. Cooper-Poole, B. 1986. "Prevalence of Syphilis in Mbeya, Tanzania: The Braddick, M.. J. C). Ndinva-Achola, N. B. Mirza, F. A. Plummer, G. Irungiu, Validitv of the VI)RL as a Screening Test." East African Medicine Journal S. K. A. Sinei, and P. Piot. 1990. "Towards Developing a Diagnostic 63:646-50. Algorithm for Chlam.dia trachsomatis and Neisseriagonorrhoeae Cervicitis in Coutinho, R. A., P. van Griensven, and A. Moss. 1989. "Effects of Preventive Pregnancy." Genitourinary Medicine 66:62-65. Efforts among Homosexual Men." AIDS 3(Supplement 1 ):S53-S56. Brandt, Allan M. 1987. No Magic Buller. New York: Oxford University Press. Curran, J. W. 1980. "Economic Consequences of Pelvic Intlammatory Disease Brown, S., F. Zacarias, and S. Aral. 1985. "sTn Control in Less Developed in the United States." Amencan Journal of Obstetn.s and GYnecolog- 138: Countries:TTheTime IsNow." InteranonalJournal ofEpidernioog-s 14:505-10. 848-51. Brunham, R. C. 1984. "Therapy for AcuIte Pelvic Inflammatory Disease: A Darta, P., M. LIaga, F. A. Plummer, J. O. Ndinya-Achola, P. Piot, G. Maitha, Critique of Recent Treatment Trials." Amenican Journaml of Obstetics and A. R. Ronald, and R. C. Brunham. 1988. "Infection and Disease after Gynecology 148:235-40. Perinatal Exposuire to Chilanydia trachomatis in Nairobi, Kenya." journal of Brunham, R. C., K. K. Holmes, and J. Embree. 1989. "Sexually Transnsitted Infectious Diseases 158:524-28. Diseases in Pregnancy." In K. K. Holmes, P. A. Mardh, P. J. Wiesner, and D'Costa, L. J., F. A. Plunmmer, ]. Bowtmer, L. Fransen, P. Puot, A. R. Ronald, P. F. Sparling, eds., Sexually Transniitted Diseases, 2nd ed.. New York: aind H. Nzanze. 1985. "Prostitutes Are a Major Reservoir of Sexuallv \McGraw-Hill. Transmitted Diseases in Nairobi, Kenya." Sexually Transmitted Diseases Bugingo, G., A. Ntilivamunda, D. Nzaramba, P. Van de Perre, A. Ndikuyeze, 12:64-67. S. Munyantore, A. Mutwewingaboi. and C. Bizinungu. 198S. "Etude suir la De Cock, K. M., A. Pozter, K. Odehoiuri. 1989. "Rapid Emergence of AIDS In s&opositivit6 liee a l'mfection au vinis de l'immunod6ficience hlmaine aln Abidjan, Ivory Coast." Lancet 2:408-10. Rwanda." Revue M8dicale Rweandaise 20:37-42. De Muylder, X. 1986. "Clinical Diagnosis of Pelvic Inflammatory Disease in Buisman, N. J. F., T. Abong Mwemba, G. Garrigue,]. P. Durand, J. S. Stilnsa. a Developing Country." Ainales de la Soci5ie beIge de Me'ecine Tropicale and T. M. Van Balen. 1988. "Chlamrvdia Ophthalmia Neonatorum in 66:339-42. Cameroon." DocitmeniaOphthalmologica 70:257-64. De Schampheleire, I.. L. Vani De Velden, E. Van Dvck, S. Guindo, W. Quint, Bulatao, R. A. 1985. "Expendituires on Population Programs in Developing and L. Franseni. 1990. "Maladies sexuellement transmissibles dans la popu- Regiois: Current Levels and Future Requirements." Working Paper 679, latiorn feminine a Pikine, Senegal." Annales de la Societe beige de Mddecine Population and Development 4. World Bank. Washington, D.C. Tropicale 70:227-35. Caldwell, I. C., P. Caldwell. and P. Quiggin. 1989. "The SOcial Context of AIDS Fakeya, R., B. Onile, anid T. Odugbemi. 1986. "AntitrepOnemal Antibodies in Sub-Saharan Africa." Populanon and Development Reiew 15(2): 185-235. among Antenatal PatienLs at the University of llonn Tedching Hospital." Cameron, D. W., 1. N. Simonsen, L. D'Costa. A. R. Ronald, G. M. Maitha, Atncan loitnal of Sexually Transmitted Diseases 2:9- 0. M. N. Gakinva, M. Cheang, J. 0. Ndinya-Achola, P. Piot, R. C. Brunhanm, Fleming, A. F. 1988. "Prevention ofTransmissiin of HIV by Blood Transfusion and F. A. Plummer. 1989. "Female to Male Transmission of Human Im- ing, A. F. Fleirino t and he eds., The Glol ,, 1 , , . ) \ 4 ,,~~~in en. in Developing Countries." In A. F. Flensinig and o.thers. eds., The Global inunodeficiency Virus Type 1: Risk Factors for Seroconversion inpact of AIDS. New York: Alan R. Liss. Lancet 2:403-7. Cates, W.. T. M. M. Farley, and P. 1. Rowe. 1985. "Worldwide Pattems of Frank, Odile. 1983. "Infertility in Sub-Saharan Africa: Estimates and Im- Infertility: Is Africa Different?" Lancet 2:596-98. plications." PopulatEio and Development Review 9(1!:137-44. Centers for Disease Control (c )L ). 1986. "Classification System for Human Fransen, L., H. Nsanze. L. D'Costa, R. C. Brunham. A. R. Ronald, and P. Piot. T-Lymphotropic Vinis Type III/Lymphadenopathy Associated Virus Infec- 1984. "Single-Dise Kanamycin Therapy of Gonococcal Ophthalmia Neo- tions." Morbidirt and Mortality Weekly Rep(ort 35:334-39. natorum." Lancet 2:1234-37. 1988. "Continuing Increase in Infectious Syphilis." Morbidity atnd Fransen, L., H. Nsanze, V. Klauss, P. van der Stuyft, L. D'Costa, R. C. Mortality WVeekly Report 37:35-38. Brunhain, and P. Piot. 1986. "Ophthalmia Neonatorum in Nairobi, Kenya: . 1989. "Tuiberculosis and Human Immunodeficiency Virus Infection: The Roles of Neisseria gonorrhoeae and Chhkmnndia trachorians." Journ(al of Recommendations of the Advisory Committee for the Ehnlination of Infectious Diseases 153:862-69. Tuberculosis." Morbidity andlMortality Weekly Report 38 236-50. Galega, F. P.,D1). L. Heyviian, and B. T. Nasah. 1984. "Gonococcal Ophthalmia Chin, J., P. Sato, and J. Mann. 1989. "Estimates aiid Projections of HIV/AIDS Neonatorum: The Case of Prophylaxis in Tropical Africa." Bulletin of the to the Year 2000." Paper presented at the Achieving Health for All World HealthOrganization 61:85-88. Symposium. September 3, Seattle, Wash. George. W. F. 1974. "An Approach to VD Control Based on a Study in Clavel, F., D. Guetard, F. Brun-Ve:inet, S. Chamaret. M. A. Rey. M. O. Kingston, Jamaica." Bntishjoumaal ofVenereal Diseases 50:22 2-27 Santos-Ferreira, A. G. Lauirent, C. Dauget, C. Katlama, C. Rouzioux, D. Ghana Health Assessment Project Team (cHAP). 1981. "Quantitative Method Klatzmainn, J. L. Chainpalimaud, and L. Montagnier. 1986. "Isolation of a of Assessing the Health Impact of Different Diseases in Less Developed New Huiman Retrovirus frotn West-African Patient with AIDS." Scienice Couintries." liternational JoUrnal of Epdenmiologp 10:73-80. 233:343-46. Goedert. J. J.. C. M. Kessler, L. M. Aledort, R. J. Biggar, W. R. Andes, G. C. Clumeck, N., H. Taelman, P. Hermans, P. Piot, iM. Schorumacher, and S. De White, j. E. Drummond, K. Vaidya, D. L. Mann, anid M. E. Eyster. 1989. Wit. 1989. "A Cluster of HIV Infection among Heterosexual People without "A Prospective Study of Human Immunodeficiency Virus Type I Infection Apparent Risk Factors." Neu EnglandJournal of Medicine 321:1460-63. and the Development of AItDS in Subjects with Hemoiphilia." Neuc England Cole, C. H., T. G. Lacher, J. C. Bailev, and D. L. Fairclough. 1980. "Vaginal Jounial of Medicine 321:1141-48. Chemoprophylaxis in the Reduction of Reinfectioin In Wonsen with Gon- Gravett, M. G., H. 1'. Nelson, T. De Rouen, C. Critchlow, D. Eschenbach, orrhoea." Brtishlournal of Venereal Diseases 56:314-18. and K. K. Holmes. 1986. "Independerit Assxociation .of Bacterial Vaginosis ColebLInders, R. L., B. Kapita, W. Nekwei, U. Y'. Bahwe, 1. Lebughe, M. and Chhlainydia trachotrnats Infection with Adverse Pregnancy Outcome." Oxtobv. and R. W. Ryder. 1988. "Breastfeediisg and Transmission of H-".' JAMA 256:1899-1903. Lancet 2:1487. Greenberg. A. E., P. Nguyen-Dinh, J. M. Nlann, N. Kabote, R. L. Colebunders, Colebunders, R. L., R. W. Ryder, N. Nzila, D. Kailunga, J. C. Willame, M. H. Francis, T. C. Quinn. P. Baudoux, B. Lyamba, F. Darachi, and others. Kaboto, B. Nkoko, J. Jeugmans, M. Kalala, H. L. Francis, J. NI. Nlann, T. C. 1988 "The Associatioln between Malaria, Blood Transfusions, and HIV 524 Mead Over and Peter Piet Seropositivitv in a Pediatric Population in Kinsiasa, Zaire." JAIMA 259: Kreiss, J. K., D. Koech, F. A. Plummer, K. K. Holmes, M. Lightfoote, P. Piot, 545-49. A. R. Ronald, J. 0. Ndinya-Achola, L. J. D'Costa, P. Roberts, E. N. Ngugi, Grimes, D. A. 1986. "Deaths Due to Sexually Transmitted Diseases: The T. C. Quinn. 1986. "AIDs Virus Infection in Nairobi Prostitutes: Spread of Forgotten Component of Reproductive Mortalitv." JAMA 255:1727-29. the Epidemic to East Africa." New EnglandJournal of Medicine 314:414-18. Gro6sbard-Schcchtman, Soshania, F. DuCharnic, and NI. Loomin, 1989. Laga, M., A. Z. Meheus, and P. Pior. 1989. "Epidemiology and Control of "Sex-Ratio Effects on the Incidenice of Sexually Transmitted Diseases," Gonococcal Ophthalmia Neonatorum." Bulletin of the World Health Organi- Population Association of America 1989 Arnuail Meetings: Program Abstra-cts. zntion 67:47 1--77. 52:94. Laga, M., W. Naimaara, R. C. Brunham, L. J. D'Costa, H. Nsanze, P. Piot, D. Guiness, L. F.. S. Sibandze, E. McGrath, and A. L. Cornelis. 1988. "Influence Kunimoto, J. 0. Ndinya-Achiola, L. Slaney, A. R. Ronald, and F. A. (of Antenatal Screening on Perinatal Mortality Caused by Syphilis in Plummer. 1986a. "Single Dose Therapy of Gonococcal Ophthalmia Neo- Swaziland." Gienitourinar's Medicine 64:294-97. natorum with Ceftriaxone." New England]ournal of Medicine 315:1382-85. Hammerschlag, M. R., C. Cummings, P. M. Roblin, T. H. Williams, and 1. Laga, M., H. Nsanze, F. A. Plummer, W. Namaara, G. Maltha, R. C. Delke. 1989. "Efficacy of Neoniatal Ocular Prophylaxis for the Prevention Brunhamii, J. 0. Ndinya-Achola, J. K. Mart, A. R. Ronald. and P. Piot. of Chlamvdial and Gonococcal Conjunctivitis." Neu England Journal of 1986b. "Epidemiology of Ophthalmia Neonarorum in Kenva." Lancet 2: Medicine 320:769-72. 1145 -48. Hethcote, H. H., and J. A. Yorke. 1984 Gonorrhea Trantistission Dynamicsatnd Laga, M., N. Nzila, A. T. Manoka, M. Kivuvu, F. Behers, B. Edidi, P. Piot, and Control. Lecture Notes in Biormathematics 56. New York: Springer-Verlag. R. Ryder. 1989. "High Prevalence and Incidence of HIV and Other Sexually 1990. "ModelingthelmpactofBreastfeeding byHiv-Infecred Wonien Transmitted Diseases (STD) aniong 801 Kinshasa Prostitutes." Paper pre- on Child SLurvival." American journal of Public Health 80:1305-1309. sented at the 5th International Conference on AIDS, June 4-9, Montreal. Hevman, S. J. 1990. "Modeling the Impact of Breastfeeding by Hlv-lnfected Laga, M., N. Nzila, A. T. Manoka, M. Malele, M. Tuliza, T. Bush. P. Piot, F. Women oii Child Survival." Amencan Journal of Public Health 80:1 305-9. Behets, W. L. Heyward, and R. Ryder. 1990. "Non Ulcerative Sexuallv Hira, S. K., and R. S. Hira. 1987. "Congenital Svphilis." In A. 0. Osoba, ed., Transmitted Diseases as Risk Factors for HIV Infection." Paper presented at Sexually Transmitted Diseases in the Tropics. Bailli&rc's Tindall, Londoil. rhe 6th Internarional Conference on AIDS, June 20-23, San Francisco. Hira, S. K., J. Kamanga, and G. J. Bhat. 1989. "Perinatal Transmissionoft HIV- I Laga, NI., and P. Piot. 1988. "HIV Infection after Plasma Donation in Valencia: in Zambia." Binish MedicalJournal 299:1250-52. Yet Another Case." Lancet 2:905. Holmes, K. K., P.-A. Mllrdh, P. F. Sparling, and P. J. Wiesner, eds., 1989. Laga, M., F. A. Plummer, P. Piot, P. Datta, W. Nansaara, J. 0. Ndinya-Achola, Sexually Transmitted Diseases. 2d ed. New York: McGraw-Hill. H. Nsanze, G. Maitha, A. R. Ronald, H. 0. Pamba, aisd R. C. Brunham. Hooper. R. R., G. H. Reynolds, (0. G. Jones, and K. K. Holmes. 1978. "Coh(ort 1988. "Prophylaxis of Ophthalmia Neonatorum: Silver Nitrate versus Tet- Studv of Venereal Diseases 1. The Risk of Gonorrhea Transmission frons racycline." New' Englaid Jourral of Medicine 318:653-57. Infected Wonsen to NIen." AmericanJournal ofEpidemniology 108:136-44. Lajmanovich, A. andJ. A. Yorke. 1971. "A Deterministic MdelforGonorrhea Homik, R. 1988. "The Knowledge-BehaviorGap in Public InformationCam- in a NonihomogeneouIs Population," Mathenatical Bioscience, 28:221-36. paigns: A Development Comniisnicarion View." Working Paper 110. Unt- Laleman, G., K. NMagazani. N. Badibanga, N. Kapila, M. Konde, V. Salemani, versity of Penissylvania, Annenberg School of Communications, State and P. Piot. 1992. "Prevention ofBlood Acquired HlvTransmission through College, Penn. a Decentralized Approach and Using a Rapid HIV Test in Shaba Province, - 1989a. "Channel Etfectiveness in Developmenit Communiication Zaire." AIDS 6: in press. Programs." Working Paper 111. University of Pennsvlvanta, Annenberg Lallemant, M., S. Lallemant-LeCoeur. D.Cheynier,J. Nzingoula,G. Jourdain, School of Communications, State (ollege, Penn. M. Sinet, M. C. Dazza, and B. Larouz6. 1989. "Mother-Child Transmission - 1989b. "General AID, Education: (Dross-National Evidence of Ef- of HIV-I and Infant Survival in Brazzaville, Congo." AtDS 3:643-46. fects." Working Paper 117. University of Pennsylvania, Annenberg School Lamptey, P., and G. A. W. Goodridge. 1991. "Condom Issues in AIDS Preven- of Communications, State College, Penn. tion in Atrica." AIDS 5:S183-S91. Jacobson, L. L., and L. Westrom. 1969. "Objectivized DLiagnosis ot Actite Lamptey, P., A. Neequay, S. Weir, and M. Potts. 1988. "A Model Prograns to Pelvic InflanimatoryDisease."AmencaTiJounialofObstemc.saidGynecolog's Reduce HIV Infection among Prostitutes in Africa." Paper presented at the 105:1088-98. 4th Intemational Conference on AIDS, June, Stockholm. Jama, H.. B. Heberstedt, S. Osman, K. Omar, A. Isse, and S. Bvgdeman. 1987. Lamptey, P., and M. Potts. 1990. "Targeting of Prevention Programs in "Syphilis in Women of Reproductise Age inl Mogadisiu, Somalia: Serolog- Africa." In P. Lamptey and P. Piot, eds., Handbook on AIDS Prevention in ical Suirvey." Genitouniary Medcine 63:326-28. Africa Durhatn, S.C.: Family Health Intemational. Janowitz, B. S., J. H. Brati, and D. B. Fried. 1990. "Investing in the Future: A Larson, A. 1989. "Social Context of Hiv Transmission in Africa: Historical Report on the Cost of Familv Planning in the Year 2000." Discussion paper. and Cultural Bases of East and Central AfTican Sexual Relations." Reviews Family Health International, Research Triangle Park, N.C. of Infectious Diseases 11:716-31. Johnson, A., and M. Laga. 1988. "Heterosextial Transnsission of HIV." AIDS Leclerc, A., E. Frost, MI. Collet, J. Goeman, and L. Bedjabaga. 1988. "Urogen- 2(Supplemenr 1 ):S49-S56. ital Chlamnydia trachomants in Gabon: An Unrecognized Epidemic." Genito- Kaptue, L., L. Zekeng, R. Salla, A. Trabucq, J. P. Lewis, Andele, A. Ndonmore, uinary Medicine 64: 308-11. Yanga, P. Lamptev, and S. Mitchell. 1990. "Setting Up a Sentinel Surveil- Louv, W. C., H. Austin, W. J. Alexander, S. Stagno. and J. Cheeks. 1988. "A lance Svstem for HIV Iisfectton in Cameroon." Paper presented at the 6th Clinical Trial ofNonoxynol-9 for Preventing Gonococcal and Chlamydial Intemational Conference on AIDS, June 20-23, San Francisco. lnfections."Journal ofInfecuous Diseases 158:518-23. Kennedy, K. I., J. A. Fortney, M. G. Bonhomme, M. Potts. P. Lamptey, and Lucas, R. E. B. 1985. "The Distribution of Wages and Employment in Rural W. Carswell. 1990. "Do the Benefits of Breastfeeding Outweigh the Risk of Botswana." In Dov Chemichovsky, R. E. B. Lucas, and Eva Mueller, eds., Postnatal Transmission of HIV via Breast Milk ' Tropical Doctor 20:25-29. The Household Economy of Rural Botswana. StaffWorking Paper 715. World Kesteleyn, P., J. Bogaerts, and A. Z. Meheus. 1987. "Gonorrheal Keratocon- Bank, Washington, D.C. junctivitis in African Adults." Sexually Transmitted Diseases 14:191-94. ILukeharm, S. A., E. W. Hook, 111, S. A. Baker-Zander, A. C. Collier, C. W. Klarman, H. E. 1965. "Syphilis Control Programs." In R. Dorfman, ed., Critchlow, and H. H. Handsfield. 1988. "Invasion of the Central Nervous Measunng Betiefits ofGovernment Invesnnent. Washington, D.C.: Brookings System by Treponema pallldum: Implications for Diagnosis and Treatment." Institution. Annals of Internal Medicine 109:855-62. HIV Infection and Sexuallv Transmitted Diseases 525 Luyeye, M., M. Gemiers, N. Lebughe, F. Behets, N. Nzila, B. Edidi, and M. Office of Technology Assessment. 1988. How effective is AIDS Education? Laga. 1990. "Prevalence et facteurs de risque pour les MST chez les femmes Congress of the United States, OTA Staff Paper 3, June. enceintes dans les soins de sant6 primaires a Kinshasa." Paper presented at Okpere, E. E., E. E. Obaseikl-Ebor, and G. M. Oyaide. 1987. 'Type oi the 5th Inrernational Conference on AIns in Africa. 10-12 October. Intra-Uterine Contraceptive Device (IUCD) Used and the Incidence of Kinshasa, Zaire. Asymptomatic Neisseriagonorrhoeae ." AfricanrJournal of Sexually Transmitred Mabey, D. C. W., N. E. Lloyd-Evans, S. Conteh, and T. Forsey. 1984. "Sexually Diseases 3:7-8. TransmittedDiseasesamongRandomlySelecredAttendersatanAntenatal Osegbe, D. N., and E. 0. Amaku. 1981. "Gonococcal Strictures in Young Clinic in the Gambia." British journal of Venereal Diseases 60:331-36. Patients." L'rology 18(1 :37-41. Mabey, D. C. W., and H. C. Whittle. 1982. "Genital andNeonatalChlamyd- Osoba, A. O., ed. 1987. "Sexually Transmitted Diseases in the Tropics." 2(1) ial Infection in a Trachoma Endemic Area." Lancet 2:301 -2. of Baillfi,re's Clinical and Tropical Medicine and Comrnmunicable Diseases. Lon- McCormack, W. M., R. J. Stumacher, K. Johnson, A. Donner, and R. don: Bailliere-Tyndall. Rychwalski. 1977. "Clinical Spectrum of Gonococcal Infection in Women." Over, Mead, S. Bertozzi, J. Chin, B. N'Galy, and K. Nyamuryekunge. 1988. Lancet 1:1182-85. "The Direct and Indirect Cost of HIV Infection in Developing Countries: Meheus, A. Z., R. Ballard, M. Dlamini. J. P. Ursi, E. Van Dyck, and P. Piot. The Cases of Zaire and Tanzania." In A. F. Fleming, et al, eds., The Global 1980. "Epidemiology and Etiology of Urethritis in Swaziland." International Impact of AIDS. New York: Alan R. Liss. Journat of Epidemiology 9:239-45. Over, Mead, S. Bertozzi, and J. Chin. 1989. "Guidelines for Rapid Estimation Meheus, A. Z., R. Delgadillo, R. Widy-Wirsky, and P. Piot. 1982. "Chlamydial of the Direct and Indirect Costs of HIV Infection in a Developing Country." Ophthalmia Neonatorum in Central Africa." Lancet 2:882. Health Policy 11:169-86. Meheus, A. Z., and A. Deschryver. 1989. "Syphilis and Safe Blood." WH-O/ Over, Mead, and J. Kutzin. 1990. "The Direct and Indirect Costs of HIV D\T/89.444. World lealth Organization, Geneva. Infection: Two African Case Studies." Postgraduate Doctor Middle East Meheus, A. Z., F. Friedman, E. Van Dyck, and T. Guyver. 1980. "Genital 13(11):632-38. Infections in Prenatal and Family Planning Attendants in Swaziland." East Pepin, J., F. A. Plummer, R. C. Brunham, P. Piot, D. W. Cameron, and A. R. African MedicalJournal 57:212-17. Ronald. 1989. "The Interaction between HIV Infection and Other Sexually Melbye, MI., E. K. Nselesani, and A. Bayley, 1986. "Evidence for Heterosexual Transmitted Diseases: An Opportunity for Intervention." AIDS 3:3-9. TransmissionandClinicalManifestationsofHumanImmunodeficiency Virus Perriens, J., Y. Mukadi, D. Nunn. 1991. "Tuberculosis and HIV' Infection: Infection and Related Conditions in Lusaka, Zambia." Lsncet 2:1113-15. Implications for Africa." AIDS 5. Monny-Lobe, M., D. Nichols, L. Zekeng, R. Solda, L. Kaptue. 1989. "Prosti- Phanuphak, P., Y. Poshychinda, T. Un-eklabh, and W. Rojanapithayakron. tutes as Health Educators for Their Peers in Yaounde: Changes in Knowl- 1989. "HIV Transmission among Intravenous Drug Users." Paper presented edge, Attitudes and Practices." Paper presented at the 5th Intemationial at the 5th Intemational Conference on AIDS, June 4-9, Montreal. Conference on AIDS, June 4-9, Montreal. Piot, P., and M. Carael. 1988. "Epidemiological and Sociological Aspects of Moses, S., F. A. Plummer, A. R. Ronald, J. 0. Ndinya-Achola. 1989. "Male HIV-Infection in Developing Countries." British Medical Bulletin 44:68-88. Circumcision in Eastem and Southem Africa: Association with HIV Sero- Plot, P., and K. K. Holmes. 1989. "Sexuall, Transmitted Diseases." In K. prevalence." Abstract (Th.G.0.27) of paper presented at the 5th Intema- Warren and A. Mahmoud, eds., Tropical and Geographical Medicine. New tional Conference on AIDS, June 4-9, Montreal. York: McGraw-Hill. Moss, A. R., and P. Baccherti. 1989. "Natural Historv of HIV Infection." AIDS Plot, P., J. K. Kreiss,J. 0). Ndinya-Achola, E. N. Ngugi, J. N. Simonsen, D. W. 3:55-62. Cameron, H. Taelman, and F. A. Plummer. 1987. "Heterosexual Transmis- Mposo, N., B. Engele, S. Bertozzi, S. Hassig, and R. Ryder. 1989. "Prospec- sion of HIV." AIDS 1:199-206. tive Quantification of the Economic and Morbid Impact of Perinatal HIV Piot, P., M. Laga, R. Ryder, J. Perriens, M. Temmerman, W. Heward, and J. Infection in a Cohort of 245 Zairian Infants Bom to HIV(+) Mothers." W. Curran. 1990. "The Global Epidemiology of HIV Infection: Continuity, Paper presented at the 5th Intemational Conference on AIDS, June, 4-9, Heterogeneity, and Change." Journal of AIDS 3:403-11. Montreal. Piot, P., and A. Z. Meheus. 1983. "Epid&miologie des maladies sexuellement Mvula, M., R. Ryder, and T. Manzila. 1989. "Response to Childhood Vacci- transmissibles dans (es pays en developpement." Annales de Ia Societe belge nations in African Children with HIV Infection." Paper presented at the 5th de Medecine Tropicale 63:87-110. International Conference on AIDS, June 4-9, Montreal. Piot, P., F. A. Plummer, F. S. Mhalu, J. L. Lamboray, J. Chin, and J. M. Mann. Nasah, B. T., R. Nguematcha, M. Eyong, and S. Godwin. 1980. "Gonorrhea, 1988. "AIDS: An International Perspective." Science 239:573-79. Trichomonas, and Candida among Gravid and Nongravid Women in Plummer, F. A., M. Laga, R. C. Brunham, P. Piot, A. R. Ronald, V. Bhullar, Cameroon." InterneationalJounal of Gynaecolog-y and Obstetrics 18:48-52. J. Y. Mati, J. 0. Achola, M. Cheang, and H. Nzanze. 1987. "Postpartum Ndilu, Mibandumba, D. Sequeira, S. Hassig, R. Kambale, R. Colebunders, M. Upper Genital Tract Infections in Nairobi, Kenya: Epidemiology, Etiology, Kashamuka,, and R. Ryder. 1988. "Medical, Social, and Economic Impact and Risk Factors."Journal of Infectious Diseases 156:92-97. of HIv Infection in a Large African Factory." Paper presented at the 4th Plummer, F. A., H. Nsanze, P. Karasina, L. !. D'Costa, J. Dylewski, and A. R. Intemational Conference on AIDS, Stockholm. Ronald. 1983. "Epidemiology of Chancroid and Haemophilus ducreyi in N'Galy, B., R. W. Ryder, B. Kapita, M. Kashamuka, R. L. Colebunders, H. Nairobi." Lancet 2: 1293-95. Francis, ]. M. Mann, and T. C. Quinn. 1988. "Human Immunodeficiency Plujmmer, F. A., J. N. Simonsen, D. W. Cameron, 1. 0. Ndinya-Achola, J. K. Virus Infection among Employees in an African Hospital." New England Kreiss, M. N. Galeinya, P. Waiyaki, M. Cheang, P. Piot, A. R. Ronald, and JournalofMedicine 319:1123-27. E. N. Ngugi. 1991. "Co-factors in Male-Female Sexual Transmission of Ngugj, E. N., . N. Simonsen, M. Bosire, A. R. Ronald, and F. A. Plummer. Human Immunodeficiency Virus Type I." Joural of Infectious Diseases 1988. "Prevention of HIV Transmission in Africa: The Effectiveness of 163:233-39. Condom Promotion and Health Education among High Risk Prostitutes." Pokrovski, V. V., 1. Eramova, V. Arzamastsev, V. Nikonova, and G. Mozha- Lancet 2:887-90. rova. 1990. "Epidemiological Surveillance for HIv-Infection in the USSR Nzila, N., K. M. De Cock, D. Forthal, M. Laga, P. Piot, and . B. McCormick. in 1987-1989." Paper presented at the 6th International Conference on 1988. "The Prevalence of Infection with Human Immunodeficiency Virus AIDS, June 20-23, San Francisco. over a 10-year Period in Rural Zaire." New England Joumal of Medicine Population Information Program. 1983. "Infertility and Sexually Transmitted 318:276-79. Diseases: A Public Health Challenge." Populatiot Reports 4. 526 Mead Over and Peter Piot Rajan. V. S. 1978. "SexuallyTransmitted Diseases (n a Tropical Island." Bitish Standaert, B., F. Niragira, P. Kadende, and P. Piot. 1989. "The Association of Joumal of Venereal Diseases 54:141-43. Tuberculosis and HIV Infection in Burundi." AIDS Research and Human Ratnam, A. V., S. N. Din, S. K. Hira, G. J. Bhat, D. S. Wacha, A. Rukmini, Retroviruses 5:247-51. and R. C. Mulenga. 1982. "Syphilis in Pregnant Women in Zambia." Bntish Stanley, E. A., S. T. Seity, P. 0. Way, P. D. Johnson, and T. F. Curry. 1989. Joumal of Venereal Diseases 58:355-58. "The lwg AIDS Model for the Heterosexual Spread of HIV and the Demo- Reeves, W. C., L. A. Brinton, and M. M. Brenes. 1985. "Case Control Study graphic Impact of the AIDS Epidemic." Paper prepared for the UNIvHO of Cervical Cancer in Herrera Province, Republic of Panama." International Workshop on Modeling the Demographic Impact of the AIDS Epidemic in Journal of Cancer 36:55460. Pattem 11 Countries, Los Alamos National Laboratories, December, Los Reeves, W. C., W. E. Rawls, and L. A. Brinton. 1989. "Epidemiology of Alamos, N.M. Genital Papilloma Viruses and Cervical Cancer." Reviews of Infectious Stray-Pedersen, B. 1983. "Economic Evaluation of Maternal Screening to Diseases 11:426-39. Prevent Congenital Syphilis." Sexually Transmitted Diseases 10:167-72. Reining, P. 1972. "lIaya Kinship Terminology: An Explanation and Some Tapia,R.,and A.Martin. 1990. "TheCostofAlDSin Mexico." Paperpresented Comparison." In P. Reining, ed., Kinship Studies in the Morgan Centenniial at the 6th Intemational Conference on AIDS, June 20-23, San Francisco. Year. Washington, D.C.: Anthropological Society of Washington. Temmerman, M., M. Laga, J. 0. Ndinya-Achola, M. Paraskevas, R. C. Rosenberg, M. J., K. F. Schultz, and N. Burton. 1986. "Sexually Transmitted Brunham, F. A. Plummer, and P. Piot. 1988. "Aetiology of Postpartum Diseases in Sub-Saharan Africa." Lancer 2:152-53. Endometritis in Nairobi, Kenya." Genitounnary Medicine 64:172-75. Rothenberg, R. B. 1983. "The Geography of Gonorrhea: Empirical Demon- Temmerman, M., N. Mirza, F. Plummer,J. 0. Ndinya-Achola, l. Wamola, and stration of Core Group Transmission," Amencan Journal of Epidemiology P. Plot. 1989."HIVlnfectionasaRiskFactorforPoorObstetricalOutcome." 117:688-94. Paper presented at the 5th Intemational Conference on AIDS, June 4-9, Rothenberg, R. B., and J. ). Potterat. 1988. "Temporal and Social Aspects of Montreal. Gonorrhea Transmission: The Force of Infectivity." Sexually Transmitted Temmetman, M., S. Moses, D. Kiragu, S. Fussallah, I. Wamola, and P. Piot. Diseases 15:88-92. 1990. "Impact of Single Session Postpartum Counseling of HIV Infected Rwandan Seroprevalence Study Group. 1989. "Nationwide Community-Based WomenonTheirSubsequent Reproductive Behavior." AIDSCare2:247-52. Survey of HIV-1 and Other Human Retrovirus Infections in a Central Temmerman, M., F. A. Plummer, N. B. Mirza, J. 0. Ndinya-Achola, 1. A. African Country." Lancer 1:947-49. Wamola, N. Nagelkerke, R. C. Brunham, and P. Piot. 1990. "Infection with Ryder, R. W., and F. S. Mhalu. 1988. "Blood Transfusion and AIDS in the Human Immunodeficiency Virus as a Risk Factor for Adverse Obstetrical Tropics." In P. Piot and J. M. Mann, eds., AIDS and HIV Infecuon itn the Tropics. Outcome." AIDS 4:1087-93. London: Bailliere-Tyndall. Toomey, K. E., and W. Cates, Jr. 1989. "Partner Notification for the Preven- Ryder, R. W., M. Ndilu, S. E. Hassig, M. Kamenga, D. Sequeira, M. Kasha- tion of HIV Infection." AIDS 3(Supplement 1):S57-S62. muka, H. Francis, F. Behets, R. L. Colebunders, A. Dopagne, R. Kambale, Tuliza, M., A. T. Manoka, N. Nzila, W. Way, M. St. Louis, P. Piot, and M. and W. L. Heyward. 1990. "Heterosexual Transmission of HIV-1 among Laga. 1991. "The Impact of STD Control and Condom Promotion on the Employees and Their Spouses at Two Large Businesses in Zaire." AIDS Incidence of HIV in Kinshasa Prostitutes." Paper presented at the 7th 4:725-32. Intemational Conference on AIDS, June 16-21, Florence. Ryder, R. W., W. Nsa, S. Hassig, M. Rayfield, B. Ekungola, A. M. Nelson, U. United Kingdom, National Health Service. 1988. "AIDS Prevention through Mulenda, H. Francis, and K. Mwangagalirur. 1989. "Perinatal Transmission Health Promotion: The U.K. Experience at National Level." Health Edu- of HIV- to Infant of Seropositive Women in Zaire." New EnglandJournal of cation Authority, London. Medicine 320:1637-42. Ryder, R. W., and P. Pior. 1988. "Epidemiology of,iv Infection in Africa." In U.S. Bureau of the Census. 1991. "Recent HIV Seroprevalence Levels by P. Piot and ]. M. Mann, eds., AIDS and HIV Infection in the Tropics. London: Country: February 1991." Washington, D.C.: Research Notes 3, Health Bailli&e-Tyndall. Studies Branch, Center for Intemational Research. Scirovsky, A., and Mead Over. 1988. "AIDS: Costs of Care in the Developed Urquhart,J. 1979. "Effect of the Venereal Disease Epidemicon the Incidence and the Developing World." AIDS 2(supplement ):S71I-S81. of Ectopic Pregnancy-Implications forthe Evaluation ofContraceptives." Senyonyi, F. 1987. "Prevalence des infections gynecologiques A Chlamydia Conrraception 19:455-80. trachomatis et a Neisseriagonorrhoeae chez 199 femmes examin6es en consul- Ursi, J. P., E. Van Dyck, C. Van Houtte, P. Piot, J. Colaert, M. Dlamini, and tations pr6natales ou en consultations pour planning familial A Kigali et 3 A. Z. Meheus. 1981. "Syphilis in Swaziland: A Serological Survey." British Kabgayi." Revue Medicale Rwandaise 19:85-89. Journal of Venereal Diseases 57:95-99. Simonsen, J. N., D. W. Cameron, M. N. Gakinya, J. O. Ndinya-Achola, L. J. Valleroy, L. A., J. R. Harris, and P. 0. Way. 1990. "The Impact of HIV-I D'Costa, P. Karasira, M. Cheang, A. R. Ronald, P. Piot, and F. A. Plummer. Infection on Child Survival in the Developing World." AIDS 4:667-72. 1988. "Human Immunodeficiency Virus Infection among Men with Sexu- Van de Perre, P., M. Carael, D. Nzaramba, G. Zissis, J. Kayihigi, and J.-P. ally Transmitted Diseases." New England Journal of Medicine 319:274-78. Butzler. 1987. "Risk Factors for HIV Seropositivity in Selected Urban-Based Singh, Y. N., A. N. Malaviya, S. P. Tripathy, K. Chaudhuri, S. D. Khare, A. Rwandese Adults." AIDS 1:207-11. Nanu, and R. Bhasin. 1990. "Human Immunodeficiency Virus Infection in van Griensven, G. J. P., E. M. M. de Vnoome, J. Goudsmit, and R. A. the Blood Donors of Delhi, India." Journal of Acquired Immune Deficiency Coutinho. 1989. "Changes in Sexual Behavior and the Fall in Incidence Syndrome 3(2):152-54. of HIV Infection among Homosexual Men." British Medical Journal 298: Sittitrai, W. 1990. "Outreach to Bar Workers in Bangkok." Hygie9(4):25-28. 218-21. Sparling, P. F. 1990. "Natural History of Syphilis." In K. K. Holmes, P.-A. Warner, K. E. 1983. "Bags, Buckles, and Belts: The Debate over Mandatory MArdh, P. F. Sparling, and P.]. Wiesner, eds., Sexually tranrsmitted diseases. Passive Restraints in Automobiles."Journal ofHealth Politics, Policy.and Law McGraw-Hill, New York. 8(1):44-75. Spencer, Nancy, C. Raevsky, and F. Wolf. 1989. "Results and Benefit-Cost Washington, A. E., P. S. Amo, and M. A. Brooks. 1986. "The Economic Cost Analysis of Provider-Assisted HIV Partner Notification and Referral." Ab- of Pelvic Inflammatory Disease." JAMA 255:1735-38. stract WAo21 in Intemational Development Research Centre, 5 Inrerna- Washington, A. E., W. S. Browner, and C. C. Korenbrot. 1987. "Cost-Effec- tional Conference on AIDS: The Scientific and Social Challenge. Montreal, tiveness ofCombined Treatment for Endocervical Gonorrhea Considering Canada. June:6-7. Co-infection with Chlamydia trachomanis."JAMA 257:2056-60. HIV Infection and SexuaUy Transmitted Diseases 527 Washington, A. E., R. E. Iohnson, and L. H. Sanders. 1987. "Chlamydia WHO) (World Health Organization). 1978. "Neisseriagonorrhoeae and Gono- trachomatis Infections in the United States. What Are They Costing Us?" cOccal Infections." Technical Report 616. Geneva. JAMA 257:2070-72. . 1985a. "Control of Sexually Transmitted Diseases." Geneva. Wasserheit, J. N. 1989. "The Significance and Scope of Reproductive Tract _ 1985b. "Simplified Approaches for Sexually Transmitted Diseases Infections among Third World Women." lnternaionaliJouyrrwlof GYnecology (sTD) Control at the Primary Health Care (PHC) Level." WHO/VDT/85.437. and Obstetrics 3(supplement): 145-68. Geneva. Way, Peter W., anid K. Stanecki. 1991. "The Demographic Impact of an AIDS . 1989a. "Consensus Statement on STDs as Risk Factors for HIV Trans- Epidemic in an African Country: Application of the lwg AIDS Model." CIR mission." W'HOGPA/1989-l. Geneva Staff Paper 58. U.S. Bureaui of Census, Center for International Research, . 1989b. "STD Treatment Strategies." WHoI,VDT/89.447. Geneva. Washingtn,n D.C. WsignD.. 1991 a. "Global Programme on AIDS Update, AIDS Cases Reported Westrom, L. 1975. "Effect of Acute Pelvic Inflammator) Disease on Intertil- toSurveillanceForecastingand ImpactAssessmentUnit." Mimeo.Genev'. itv. " Amnerican Journal of Obstetrics andr3 Gnecolobgy 121.707- 13.t useiac oeasn n pc sesen nt. m eea -ry "^Amn ounalof OstericsandGNeoloy12:70-I . 1991b. "Management of Patients with Sexually Transmitted Dis- -.1980. "Incidence, Prevalence, and Trends of Acute Pelvic Inflamma- eases." WHo Technical Report 810, World Health Organization, Geneva. tory Disease and Its Consequences in Industrialized Countries." Amencan JournalofObstetncsaandtGynecolog; 138:880-92. World Bank. 1989. Worlds Developent Report. Baltimore: Johns Hopkins Westrons, L., I. Serafim, L. Svensson, and P-.A. Mardh. 1979. Infertilitv after University'HPress. Acute Salpingitis: Results of Treatment with Different Antibiotics."; Cur- Yorke,J. A., H. W. Hethcote, and A. Nold. 1978. "Dynamics and Control of rerituTherapeutic Research, 26:752-59. the Transnsission ofGonorrhea." Sexually Transmitted Diseases 5:51-56. Willianms. Eka, N. Hearst, and 0. Ndofia. 1989. 'Sexual Practices and HIV Yvert, F., J. Y. Riou, E. Frost, and B. Ivanoff. 1984. "Les infections gono- Infection if Feniale Prostitutes in Nigeria," Abstract WAo 24 in Intema- cocciques au Gabon (Haut Ogooue)." Pathologie Biologie 32:80-84. tional Development Researach Centre, 5 Internanonal Conference Otl AlLiS Zachariah, K. C., and M. T. Vu. 1988. W'orld Population Projectons. World The Scientfic aiid Social Challenge. Montreal, Canada. June 1985. 985. Bank, Washington, D.C. Wilson, David and Amir Mehrvar. 1981. "The Role of AIDS Knowledge, Ziegler, J. B., D. A. Cooper, R. D. Johnson, and J. Gold. 1985. "Postnatal Attitudes, Beliefs aind Practices Research in Sub-Saharan Africa." AIDS Transmission of AIDs-Associated Retrovirus from Mother to Infant." Lancet 5:S 17 7-S81. 1:896. I 21 Cancers Howard Barnum and E. Robert Greenberg Human cancer consists of more than 100 distinct diseases, each importance of cancer has increased as the age structure of the defined by its anatomic site of origin and microscopic features population has changed (see, for example, Bumgamer 1992). (cell type). The two characteristics shared by all cancers are Third, once an individual is past the hurdle of childhood the uncontrolled proliferation of cells and their invasion into diseases, cancer looms as one of the three largest causes of other tissues. Cancers differ, however, in their clinical features. death (the other two being accidents and cardiovascular dis- Some, such as esophageal cancer, progress rapidly, are rela- eases) even in the lowest-income countries of Africa and Asia tively intractable to treatment, and are almost uniformly fatal. (Stjemsward and others 1985). Last, changing demographics Others, chronic lymphocytic leukemia, for example, usually and increasing tobacco consumption virtually ensure that an follow an indolent course and may persist for decades with epidemic of lung cancer will occur in many developing coun- little morbidity. tries during the next century (Stanley 1986). It is important to Cancers differ also in their etiology. As a consequence, act now to prevent tobacco use rather than to wait until the individual cancer types tend to have distinct epidemiological epidemic is manifest. features, preferentially affecting particular populations as de- Despite its importance in adult mortality, cancer has not fined by geography, culture, and personal habits. For example, been considered in shaping health policy in most developing cervical cancer occurs more often in the developing world and countries. The discussion that follows provides a brief survey among women with low income and a history of many preg- of the epidemiology of cancer and gives suggestions for incor- nancies. In contrast, breast cancer tends to strike more affluent porating cancer planning in health policy. The appendix pro- women who live in industrial countries and have a history of vides a review of the salient characteristics of ten of the most fewer or no pregnancies (Kelsey and Hildreth 1983). important cancers and the environmental, behavioral, physi- An analysis of the health policy implications of cancer can ological, and occupational circumstances-collectively re- therefore not proceed very far using an aggregate concept of ferred to as risk factors-that have been associated with their cancer because aggregation obscures important details of cause occurrence. and potential interventions. Two countries can have the same aggregate rate of occurrence of cancer but need to employ very Public Health Importance different types of intervention. The discussion that follows uses a level of disaggregation that distinguishes the ten most im- Accurate information on cancer occurrence (mortality and portant cancers in the developing world and considers their incidence rates) and on the prevalence of risk factors is essen- individual etiological and clinical characteristics. tial in assessing the public health importance of cancer and in The focus of this chapter is the implications of cancer planning control strategies. epidemiology for health policy in developing countries. A common notion is that cancer is primarily a disease of indus- Data Sources trial countries, occurring late in life as a consequence of affluent lifestyle. It thus is felt to be not important in develop- Determination of cancer mortality rates primarily requires a ing countries, where the focus should be on infectious diseases reliable system of death registration and accurate demographic and childhood. . here are good reasons why cancer deserves data and thus can be accomplished within the context of attention, however. First, several important cancers, including mortality measurement for other diseases. Determination of stomach and liver cancer, occur most often in poorer countries cancer incidence is best achieved through specialized tumor (Stjemsward and others 1985; Parkin, Laara, and Muir 1988). registries (Waterhouse and others 1976). These require an Second, some low-income and middle-income countries, such effort directed toward cancer, and in their absence cancer as China, Sri Lanka, Malaysia, and Brazil, have reduced fertil- incidence may be approximated by applying estimated case- ity and causes of infant and childhood death, and the relative fatality rates to cancer mortality data (Parkin, Laara, and Muir 529 530 Howard Barnum and E. Robert Greenberg 1988). The prevalence of risk factors is measured by surveys munization should lead to better control of infectious diseases, employing questionnaires or by direct measurement of physical decreasing deaths from these causes. Declining fertility and and biochemical characteristics, reduced infant and childhood mortality will eventually shift Countries vary greatly in the availability and reliability of population age structures from a pyramidal to a more columnar their cancer data (Muir and Nectoux 1982). This fact largely pattern (with a greater proportion of older adults, whose risk reflects the limited resources devoted to the measurement of of cancer is higher). In addition, tobacco use is increasing vital statistics by many governments in the developing world. in many developing countries, and increases in cancer will Also, the reliability of cancer data depends on coding practices undoubtedly follow (Crofton 1984; Stanley 1986). (Percy and Muir 1989), on the level of medical care, and on The pattem of cancer occurrence now differs greatly be- the availability and quality of diagnostic procedures. Cancer tween developing and industrial countries. In industrial coun- incidence and mortality rates are likely to he underesti- tries, most cancer deaths are due to tumors of the lung, mated in countries in which people have little access to high- colorectum, breast, prostate, and pancreas. In developing technology medical care. Although some excellent rumor countries the m-nain causes of cancer death are tumnors of the registries have been established in several low-income coun- stomach, esophagus, lung, liver, and cervix (Parkin, Laara, and tries (Muir and others 1988), they cover cancer incidence in Muir 1988). In table 21-1 we provide estimates of numbers of only a minor fraction of the developing world. There also are deaths and new cases for the ten most frequent cancers in large gaps in the internatiotnal data on cancer mortality rates developing countries.' Note that the relative importance of (Kurihara, Aoki, and Tominaga 1984). Thus, the incidence these cancers differs depending on whether one focuses on and mortality rates quoted here must be interpreted with the deaths or new cases. This is because cancers of the lung, realization that there are severe deficiencies in cancer data, stomach, and esophagus are almost invariably fatal, whereas prolonged survival and cure are common for breast, oral, and Global Epidemiology cervical cancer cases. The degree of industrial development is an unreliable guide Cancer accounts for approximately 8.5 percent of the 51 to the pattern of cancer occurrence in a given country. For million deaths occurring in the world each year (Hakulinen example, although esophageal cancer is generally more fre- and others 1986). Of the estimated 4.3 million cancer deaths, quent in developing countries, its incidence is also high in more than half (2.5 million) occur in developing countries. In northwestern France and parts of eastem Europe (Ghadirian, relation to other causes of death, however, cancer seems less Thouez, and Simard 1988). Similarly, stomach cancer is com- important in these countries, in which it accounts for approx- mon not only in developing countries but also in Japan, where imately 5.5 percent of all deaths, well below those due to it is the leading cause of cancer death (Kurihara, Aoki, and infections (40 percent), circulatory diseases (19 percent), and Tominaga 1984). The occurrence of most cancers actually perinatal events (8 percent). In most industrial countries can- appears to he deternined by factors (such as tobacco consump- cer accounts for approximately 20 percent of deaths, second tion, diet, and reproductive practices) that are related only only to circulatory diseases (about 50 to 55 percent). indirectly to industrial development (Doll and Peto 1981). There are four principal reasons for the relatively lower Accordingly, statements about the cause and prevention of importance of cancer as a cause of death in developing coun- cancer in the developing world (considered as a whole) may tries. One is the continued high death toll from infectious and not be applicable to all developing countries. Policies to ad- parasitic conditions that have been largely eliminated as causes dress the cancer problem must be formulated on a country-by- of death in the industrial countries. A second reason is that the age structure of the populations of most developinig countries is heavily weighted toward young children. Cancer occurs Table 21-l. Estimated Annual Deaths and New most frequently among older adults, and this group accounts Cases of Ten Most Important Cancers in the for a small percentage of the populations of these countries. Developing World, 1980 The third reason is that aggregate cancer risks are truly lower (thousands) in many (though not all) developing countries. Among coon- Site or type of caticer Deaths Neu cases tries such as Cuba and Costa Rica, age-adjusted cancer mor- tality rates are approximately one-third to two-thirds less than Stomach 280 336 in typical heavily industrialized countries (Kurihara, Aoki, and Esophagus 231 254 Tominaga 1984). Many of these differences, however, are Lunegr 174 206 accounted for by lower rartes of lung cancer and other tobacco- Cervix 154 370 associated cancers, and the plcture is complicated considerably Colon/rectLm 108 183 when we consider other cancers, such as those of the stomach Mouth/pharynx 101 272 and liver. Last, cancers are more likely to he unrecognized in Breast 97 224 populations with less access to advanced diagnostic facilities, Lymphoma 81 122 so the actual effect of cancer is underestimated. Leukemnia 81 106 In the near future, cancer will become increasingly import- Note: Excluides skin cancers. ant in developing countries. Improved sanitation and im- Sozsrce Parkin, Laara, and Muir 1988. Cancers 531 country basis, taking into account particular features of cancer increases in deaths from tobacco-associated cancers (Bailar occurrence and the prevalence of particular risk factors in each and Smith 1986). These latter cancer deaths are occurring as population. Clearly, reliable country-specific data will be cru- a result of tobacco smoking initiated decades in the past. cial to this process. Will these changes be repeated in the developing countries? Tobacco-associated cancers are certain to increase in countries Time Trends in Cancer Occurrence where tobacco use has risen. Cancers of the breast and col- orectum are also likely to increase in populations that adopt There is relatively little information on cancer trends in de- the reproductive patterns and diet of inore industrialized coun- veloping countries themselves, and one must try to draw tries. A drop in the occurrence of stomach cancer seems likely parallels from industrial countries that have a longer history of in countries in which diets have come to include more fresh collecting cancer incidence and mortality data. Over the past fruits and vegetables and fewer spoiled foods. Changes in fifty years, the most striking changes in cancer occurrence in sexual practices may lead to declines in cervical cancer. many of these countries have been increasing rates of lung Changes in health care and hygienic practices could lead to a cancer and falling rates of cancers of the stomach and uterine reduction in hepatitis and, hence, liver cancer. In the absence cervix (Kurihara, Aoki, and Tominaga 1984; Stanley, of direct intervention, however, these favorable changes are Stjernsward, and Koroltchouk 1988). The profound increase apt to occur gradually, and increases in lung, colorectal, and in lung cancer is almost entirely attributable to tobacco use, breast cancer may offset declines in stomach, liver, and cervi- which became prevalent (in the United States and western cal cancer. Europe) early in this century (Doll and Peto 1981). There has also been a smaller increase in other tumors, such as bladder The Current Burden of Cancer cancer, that are related to tobacco use but not as strongly as is lung cancer. The fall in uterine cervical cancer deaths is not The total economic cost of cancer to society can be conve- fully understood (Knox 1982). Possible contributing factors, niently partitioned into indirect and direct costs. The indirect besides Pap testing, include changes in sexual practices, im- cost is the cost to society from the loss of productive life, and proved genital hygiene, and increased frequency of hysterec- the direct cost is the value of the resources (including those tomy (removing the uterus removes the risk of cervical used for health sector services) required by cancer. The pro- cancer). Reasons for declines in stomach cancer mortality in portional split of total cost between these two categories varies developing countries are uncertain (Stanley, Stjernsward, ancl considerably by disease. In a study by Rice, Hodgson, and Koroltchouk 1988). A plausible explanation is that diets have Kopstein (1985) of the cost of disease in the United States for improved and consumption of spoiled or mold-contaminated 1980, direct costs for cancer were only 26 percent of total cost, food (due to better food storage and refrigeration) is thus lower and the remainder was composed of the indirect costs of (Bjelke 1982). Also, increased consumption of antioxidants morbidity (11 percent) and mortality (63 percent). Only inju- (as food additives and vitamin C) may have contributed to the ries, at 23 percent, had direct costs that were a lower percent- decline. Lastly, reduced rates of infection with Helicobacter age of the total. For comparison, the direct costs of diseases of pvlori, a pathogenic bacterium implicated in gastric carcino- the genitourinary system were 80 percent. The importance of genesis, may have resulted from better sanitation in industrial indirect costs of cancer is attributable to the high case-fatality countries (Correa 1992). rates for cancers and the fact that in the developing countries Within the industrial countries there has been little change cancer primarily affects people during their productive years during the last several decades in deaths due to cancers of the (albeit often relatively late in life). breast (Stanley, Stjernsward, and Koroltchouk 1988) and Although morbidity and mortality rates, and therefore the colon (Boyle, Zaridze, and Smans 1985). Reported increases in total costs of cancer, vary considerably on a global basis, with some countries of breast cancer and pancreatic cancer are afew exceptions, the parametersdetermining the proportional difficult to interpret because there have been improvements in importance of indirect cost per case do not vary markedly diagnostic capabilities and greater efforts to find cancer, par- between countries. In figures 21-1, 21-2, and 21-3, respec- ticularly among the elderly. A modest decline in large bowel tively, we show the age-specific mortality rates for lung cancer, cancer mortality in some industrial countries also cannot he liver cancer, and stomach cancer in China and the United easily explained, although dietary changes (Boyle, Zaridze, and States. As for most cancer, the force of mortality is highest late Smans 1985) or earlier diagnosis and improved treatment may in life. Leukemia (shown in figure 21-4) is a notable exception, be partly responsible. with appreciable mortality occurring earlier in life. In North America and westem Europe, mortality from A general measure of the burden of specific cancers is given childhood leukemia and lymphoma has fallen dramilatically. by calculation of years of life lost (YLL), measured as the This observation is attributable to improved treatment meth- difference between the age of death for victims of the disease ods developed during the past thirty years through a massive in comparison with life expectancy. In table 21-2 we give the investment in cancer research (Miller and Mckay 1984). In average YLL per case and the total number of years lost out of a countries such as the United States, however, decreases in population of 1 million for eight main cancers in a prototypical cancer deaths achieved by improved treatment of leukemia, developing country. In order of total burden, cervical, breast, Iymphoma, and other, uncommon, cancers have been offset hy and stomach cancers are the most important for women, and 532 Howard Barntm and E. Robert Greenberg Figure21-1. Age-SpecificMortalityforCancer Figure 21-2. Age-Specific Mortality for Liver in China and the United States Cancer in China and the United States Mortality (per 100,000) Mortality (per 100,000) 300 100 O China O China 250 - G United States 80 - United States 200- 60- 150- 100 l rUh1VI 40V 50 -20V 0 10 20 30 40 50 60 70 0 10 20 30 40 50 60 70 Age (years) Age (years) Source: China: unpublished Disease Surveillance Point data 1986; U.S.: Source: China: unpublished Disease Surveillance Point data 1986; U.S.: National Institutes o0 Health 1988. National Institutes of Health 1988. Figure 21-3. Age-Specific Mortality for Stomach Figure 21-4. Age-Specific Mortality for Leukemia Cancer in China and the United States in China and the United States Mortality (per 100,000) Mortality (per 100,000) 250 50 O China El China 200 - U United States 40 - E United States 150 30- 100 20- 50 -2 10 I 0 10 20 30 40 50 60 70 0 10 20 30 40 50 60 70 Age (years) Age (years) Source: China: unpublished Disease Surveillance Point data 1986; U.S.: Source: China: unpublished Disease Surveillance Point data 1986; U.S.: National Institutes of Health 1988. National Institutes ol Health 1988. Cancers 533 Table 21-2. Average Years of Life Lost from Premature Mortality from Cancer in Developing Countries Years of life lost per million Site or type Average age at death Average years of life lost per death populationa of cancer Male Female Male Female Male Female Total Stomach 64 62 16 18 853 561 1,414 Liver 56 57 22 22 814 312 1,126 Esophagus 65 65 16 16 714 402 1,116 Leukemia 32 34 42 41 599 418 1,016 Lung 63 66 17 15 721 210 931 Cervix n.a. 61 n.a. 18 n.a. 837 837 Colon/rectum 61 60 18 20 313 304 617 Breast n.a. 57 n.a. 22 n.a. 645 645 Total 59 58 20 21 6,611 5,906 12,517 n.a. Not applicable. a. Based (in population distribution, incidence, and case-fatality rates. Source: Parkin, Laara, and Muir 1988. stomach, liver, lung, and esophageal cancers are of primary is an important determinant of their competitiveness. Thus, concern for men. the relative potential of preventive interventions is dependent on the ranking of the targeted cancer as a cause of lost life and Prevention economic productivity, the technical effectiveness of the in- tervention, the feasibility of implementing the primary pre- A large number of primary and secondary prevention activities vention activity, and the cost per person covered by the have been proposed for control of cancer mortality. These activity. These factors are considered in the specific interven- activities vary greatly, however, in their cost and potential tions discussed below. effectiveness as part of a national cancer strategy. A consider- able literature has emerged that discusses causes of cancer and PROGRAMS TO REDUCE TOBACCO CONSUMPTION. A clear the effectiveness, hazards, and (to a regrettably lesser extent) causal relationship is established between tobacco consump- the costs of preventive activities. Most of this literatLre has tion and many cancers, including tumors of the lung, mouth, been oriented toward industrial countries, but some general upper respiratory and digestive tracts, kidney and urinary tract, conclusions apply in the developing world. In the next two and other sites, with the most important link being to lung subsections we discuss the implications of current knowledge cancer (USDHHS 1982; IARC 1986). The many cancer deaths of cancer risks for primary and secondary prevention programs attributable to tobacco consumption mark antitobacco pro- in developing countries. grams for special consideration among cancer prevention ac- tivities. In Europe and North America the dramatic increase Primary Prevention in lung cancer in the last thirty years has made it the leading cause of cancer mortality and one of the main causes of death. Primary preventive activities are designed to lessen exposure The increase has been firmly linked to earlier changes in to risk factors. The main potential risk factors identified for smoking behavior. Because of formerly low levels of cigarette intervention include tobacco consumption, food, alcohol, in- consumption, lung cancer has been a relatively less important fections, environmental and occupational chemicals, and ra- cause of death in developing countries, but as life expectancy diation (Doll and Peto 1981). Evaluation of the potential and smoking prevalence increase, lung cancer and other smok- usefulness of alternative interventions has several compo- ing-related causes of death are becoming more important. nents: (a) the importance of targeted cancers and the preva- Additionally, chewing tobacco, especially betel nut chewing lence of targeted risk factors in individual countries must be in India, has long been an important cause of oral cancer (WHO established by surveillance; (b) the technical effectiveness of 1984). specific interventions is being explored in experiments in Studies of the hazards of tobacco consumption have in- industria! countries, but little is known about the ease of cluded examinations of the difference in risk for alternative transferring the results of these experiments to other countries; forms of consumption (cigarettes, pipes, cigar), dose (daily (c) the feasibility of preventive activities, even though they number of cigarettes, tar and nicotine content, and method of may be technically effective, is impeded by the practical diffi- ingestion), and duration (age started or years of habit [Peto culty of alrering underlying behavior and customs, interfering 1986]). The risk of lung cancer for those who smoke approxi- in production or consumption processes, and managing large- mately one to ten cigarettes per day is three to five times greater scale preventive campaigns; and (d) prevention activities must than for nonsmokers; for those who smoke one pack (twenty compete with other resource uses, and the cost of the activities cigarettes) the relative risk is seven to nine times greater; and 534 Howard Barnum and E. Rohert Greeribcrg for two packs, nine to twenty-five times greater. Smokers who despite the relatively loose social structure and limited govem- report that they do not inhale nevertheless have a risk four to ment ability to produce behavioral change. The United States eight times that of nonsmokers. effort has benefited from effective national coordination pro- A striking aspect of the link between smoking and cancer is vided by the Surgeon General's office and the Office of Smok- that duration of smoking appears to be more important than ing and Health in the Department of Health and Human daily dose in determining lung cancer risk. For example, in one Services. Progress has been retarded, however, by agricultural summary of the literature the author concludes that "a three- and commercial interests within and outside government. fold increase in the daily dose-rate may produce only about a Thus, a full national program to contain tobacco consumption three-fold increase in effect, while a three-fold increase in in a given country would also likely involve several ministries duration might produce about a 100-fold increase in effect" directly, notably the education and health ministries, and (Peto 1986 p. 23). The effect of duration of the smoking habit would need to be coordinated through a special agency or on lung cancer is crucial in assessing past trends of cancer in committee to ensure interagency cooperation. industrial countries and in projecting the future effect of can- The cost of an antismoking program is difficult to estimate cer in developing countries. Current high rates of lung cancer but is probably not great; implementation is more a matter of in industrial countries result from a high percentage of the political and social will than specific costly activities. The population that started smoking at an early age and continued primary cost of a basic antismoking education program is that to smoke for several decades. Cultural and social changes in of organizing and supplying an informational effort. An infor- developing countries are lowering the starting age for smoking, mation campaign could include warnings of the adverse health increasing the duration of smoking, and increasing the per- consequences of smoking on cigarette packets; antismoking centage of the population that smokes. messages in posters, billboards, newspapers, and on the radio; Tobacco consumption is increasing in developing countries information dissemination through the health system; and at the same time that it is leveling off or decreasing in the school programs. Cost items would include advertising ex- industrial world (Crofton 1984). Between 1980 and 1986, pense; short-term training courses and seminars for health consumption increased at an average of 5.4 percent per year in professionals, teachers, and local leaders; and supervision and developing countries. The percentage of the population that management. The cost would be dependent on the available smokes is higher in many developing countries, but per capita infrastructure for information and education dissemination consumption tends to be greater in higher-income countries and on the scale of the program. Some estimate of the cost can (IARC 1986). Rising per capita income, however, is expected to be made from examination of other programs that use infor- increase per capita tobacco consumption, and this effect is mational campaigns to bring about behavioral change. On the greater for lower-income countries. A 10 percent increase in basis of the cost of training programs and information and per capita income can be expected to create a 7 percent education activities in several World Bank projects, the cost increase in tobacco consumption in middle-income countries per capita could vary between 0.005 and 0.025 percent of the and more than a 13 percent increase in lower-income coun- annual gross national product per capita (GNPN) in a typical tries.i Thus, it is clear that, unless preventive measures are low-income country. taken, tobacco consumption will continue to increase as de- A particularly important problem for antitobacco programs velopment proceeds in developing countries. is that tobacco, because of its nicotine content, produces a Although the current pool of smokers will inevitably gener- strong consumption dependence that must be recognized in ate more lung cancer in the future, even greater increases can prevention strategies. For this reason it seems more feasible, be avoided by preventing starts. Possible components for pre- and less costly, to prevent new starts than to convert those who vention programs include prohibiting cigarette advertising, have already developed a dependence.4 The beneficial effects requiring warning labels on cigarette packages, using anti- of an antismoking campaign would be to prevent new smokers smoking advertising, instituting educational programs in from starting, to convert previous smokers to nonsmoking schools and in work groups, banning tobacco smoke from status, to convert some smokers to the use of low-tar cigarettes, workplaces and public areas, increasing the price of tobacco and to reduce passive smoking. It is possible to use a few products through special taxes, and finally (with limited effec- plausible and simple assumptions from available information tiveness) decreasing the carcinogenic content of cigarette to demonstrate the low cost per unit of effect in preventing smoke through tobacco processing and filters. new starters and encouraging smoking cessation in a hypothet- Most of these program components involve behavioral ical low-mortality population of I million. change, so their feasibility is highly dependent on the cultural Out of I million people with a smoking prevalence of 0.5 context in which they are carried out. In addition, the numer- among the adult male population and 0.2 among adult females, ous elements in a smoking prevention program require na- the annual number of new starters is estimated to be 8,307 tional coordination, monitoring, and motivation (Mackay (6,390 men and 1,917 women). Of these, approximately 25 1989). The feasibility of an effective program is thus apt to vary percent will die from causes attributable to smoking, and the considerably across developing countries. Some encourage- average age of death will be fifty-five years, resulting in a pre- ment can be taken from the moderate success in reducing mature loss of thirteen years of life for each smoking-induced smoking in the United States during the last twenty years, male death and sixteen years for each female death. The total Cancers 535 Table 21-3. Cost per Year of Life Gained from an the program expenditure, some adjustment is needed to take Antismoking Campaign, as a Function of Smoking account of time. Discounting provides a method of comparing Starts Averted events that occur at different points in time. If we use a (percent of per capita GNP) discount rate of 3 percent and use the difference between the Program cost per capita Proporion of new smoking starts averted average age of starting smoking (twenty-one years) and the (percent GNPN) 0.10 0.25 0.50 average age of a smoking death (fifty-five years) as the discount period, the cost per discounted unit of effect is still only 1.9 0.005 1.8 0.7 0.4 percent GNPN (table 21-4).5 0.010 3.5 1.4 0.7 The cost-effectiveness calculations above were made on the 0.020 75. 2.8 1.4 assumption that the total effect of the program would be to prevent new starts. There will also be effects in bringing about Note: For a hypothetical program. Information for the Unired States indi- cessation in response to the basic informational and educa- cates that 25 percent of smokers will die from smoking-arrrihuted causes (50 tional campaign, and these effects, added to prevention of new percent from coronary, 20 percent from cancer, and 15 percent each from cerebrovascular and pulmonary diseases. When weighted across 'LL per starts, will increase the cost-effectiveness of the program. Ces- death by caLise, there is an average of thirteen YLL for men and sixteen YLL sation is difficult to achieve but can be cost-effective. Over a for women. The total years of life lost from all smoking starts is 28,435 per ten-year period of sustained informational effort, without any million persons in the absence of any intervention program. secial prog tareed seficall at s lted usmoke Source: Authors. special programs targeted specifically at selected smoker groups, a 10 percent quit rate would be possible in the industrial countries (Altman and others 1987; Warner 1987), and a years of life lost will be 28,000. If a national smoking campaign, well-designed program conceivably would be more successful costing 0.01 percent GNPN per person could have prevented 50 in some developing countries. Compared with preventing new percent of new starts, the cost per year of life gained (YLG) starts, the effects of quitting would be less, however. The would have been 0.7 percent CNPN. Obviously, this is a low cost relative risk (RR) of mortality from all causes for smokers as for such a significant benefit. opposed to nonsmokers is about 2 on average, and a smoker's The cost of an antismoking campaign per YLG is sensitive to chance of dying from a cause attributable to smoking is roughly the program cost per capita and the efficiency of the program 0.25. About ten years after cessation the RR falls to about 1.25 measured as the proportion of new starts averted. In table 21-3 and a former smoker's chance of dying from a smoking-related we give the cost-effectiveness of an antismoking campaign cause falls toabout0.0625. Withaprogram cost of0.0l percent under several alternative assumptions. Even under adverse GNPN and a quit rate of 10 percent over ten program years, the assumptions-a cost per capita of 0.02 percent GNPN and an cost perYLG is 0.5 percent GNPN undiscounted or 1.6 discounted efficiency of only 0.10 of new starts averted-the program at 3 percent. remains relatively inexpensive at only 7 percent GNPN per year of life gained. PRIMARY PREVENTION OF VIRUS-ASSOCIATED CANCERS. Several It might be argued that, because most effects of the smoking cancers are associated with viral infections. These include liver campaign expenditure would not occur until many years after cancer, which is an outcome of infection with the hepatitis B Table 21-4. Cost-Effectiveness of Antismoking Programs Percent of GNPN per death pretented a Percent of GNPN per year of life gained Study ra - 0 ra-0.03 ra- 0 ra -0.03 Hypothetical example in text New starters only 9.6 26.3 0.7 1.9 Quitrers included 7.5 17.6 0.5 1.6 United States cessation programs Smokingcessationclasses 11.8 18.4 0.8 1.2 Incentive-based contest 6.5 10.1 0.4 0.6 Self-help antismoking kit 2.1 3.2 0.2 0.3 Eddyc cessation program 12.0 18.0 0.8 1.0 Notes: Effects include prevention of excess mortality from lung cancer, other cancers, cardiovascular disease, chronic obstructive lung disease, and other causes; it is not practical to identify cancer as a sole objective. a. Discount rate. b. Altman and others 1987 present cost per quitter, which has been converted to deaths prevented using an excess mortality rate of 0.1875 and to years of life gained assuming fourteen years of life lost per smoking-associated death. Discounting assumes that the mean age of quitters is forty-two and the mean age at death is fifty-seven. The figures given are based on the additional cost of the special programs and do not include the cost of the national information pro- gram. c. Based on Eddy 1981. CLonverted to percent GNPN using U.S. GNPN of $12,800. Source: Eddy 1981; Altnian and others 1987. 536 Howard Barnum and E. Robert Greenberg virus (HBv); cervical cancer, which is related to the human tropical Africa, and the Amazon region of South America), 70 papilloma virus; and nasopharyngeal cancer and Burkitt's lym- to 95 percent of the population show HBv antibodies and 7 phoma, which are both related to the Epstein-Barr virus. Of percent to 20 percent are carriers; neonatal infection is fre- these, the most immediate interest is in the link between FIBV quent. Three-fourths of the world's population live in areas of and liver cancer because of the strength of the evidence indi- intermediate or high prevalence. In areas of low prevalence cating causality and because of development, during the last the ITFH recommends that only high-risk groups (such as med- fifteen years, of an effective vaccine (WHO 1983; Beasley 1988). ical and dental workers) be immunized. In areas of intermedi- The etiology of HBv infection is complex. The virus is ate or high prevalence the most effective preventive strategy transmitted through close contact with serum-derived fluids, is immunization of neonates as part of immediate postnatal including blood, dental exudates, skin exudates, and semen. care because of the large proportion of infections that are Of those who develop infection, the proportion dying within acquired at birth or perinatally. a few months is approximately 0.00125, although a much The feasibility of immunization programs closely parallels higher proportion have symptoms of hepatitis. Longer-term that of WHO's Expanded Programme on Immunization (EPn) in consequences of the infection are much more frequent and are general. In countries with a low quality of maternal child associated with development of a chronic carrier state of the health services the obstacles to a correctly timed delivery of virus. From twenty to forty years following HBV infection those the first of the three doses required must be overcome to ensure with the carrier state are at much greater risk of mortality from coverage of the neonatal population. In countries such as both primary liver cancer (PLC) and cirrhosis. The percentage China and Indonesia, where development of preventive of persons who become chronic virus carriers is closely related health services is more advanced, the feasibility of an effective to the age of infection; most infected infants develop the HBV program is high. carrier state, but fewer than 10 percent of those infected in Hepatitis immunization can be added to existing EPI pro- adulthood become carriers. A study of HBv carriers in Taiwan grams (The Gambia Hepatitis Study Group 1989), with the found that the incidence of PLC among carriers was about 200 additional requirement that, to interrupt transmission, new- times greater than among noncarriers (Beasley and others horn children of infected mothers must be inoculated as soon 1981). Other studies indicate that aflatoxin consumption acts as possible after birth (usually within seven days). Subse- with the HBV carrier state to increase risks of PLC and cirrhosis quently, a second dose is to be delivered with EPI vaccines as (Bulatao-jayme and others 1982;Van Rensburg and others convenient from four to twelve weeks later, and then a third 1985; Yeh and others 1989). The long-term excess risk of dose two to twelve months following the second. Two strate- mortality from HBV-linked causes is about 15 to 30 per 1,000 gies are possible. The first is to immunize all newborns. The cases (ITFIH 1988). second is to screen mothers for hepatitis B surface antigen There is no standard treatment for HBV infection. Preven- positivity and immunize only the infants of mothers who were tion strategies include immunization, better sanitation, and positive (that is, carriers). Because of the cost and current low improved sterilization of medical instruments, needles, and production of the vaccine, the second strategy is tempting. In syringes. During the last ten years, development of both low-income countries with high endemicity, however, the first plasma-derived and recombinant DNA types of vaccine against strategy appears preferable. Epidemiological modeling by the HBV virus has made it technically possible to prevent infection ITFH indicates that the screening strategy would not reduce the (Szmuness and others 1981; Francis and others 1982; Wain- HBV carrierstate by more than 30 to 50 percent compared with wright and others 1989). Both types of vaccine are safe and 75 a long-run reduction of at least 90 percent for a continued to 95 percent effective, depending on delivery conditions. strategy of total newborn vaccination. The capacity for screen- Until recently the price of vaccine was high (more than $ 100 ing and effective follow-up may also be difficult to develop. for a full course of three shots using the recombinant DNA type Immunization against the hepatitis B virus is a relatively of vaccine), but a fall in the price of HBV vaccine to below $1 new technology, and the cost has only recently fallen to levels per injection has increased the possibility of large-scale pro- that make it a possible strategy. There are no existing large- grams for the control of hepatitis. scale programs that can be examined for cost-effectiveness. The International Task Force on Hepatitis B Immunization Tentative projections, however, have been made of the cost- (ITFH) distinguishes between areas of low, intermediate, and effectiveness of immunization strategies for China and Indo- high HBV prevalence. In areas of low prevalence (western nesia, where programs for HBV vaccination are under active Europe, Australia, North America, and southern South Amer- consideration or in the early stages of implementation. In ica), 4 to 6 percent of the population show HBV antibodies and addition, the ITFH has made estimates for a prototypical pro- only 0.2 to 0.5 percent are carriers; neonatal infection is gram in a high-prevalence country. infrequent. In areas of intermediate prevalence (eastern Eu- Recast in annual percentage of per capita GNP, the costs for rope, countries of the former U.S.S.R., the Mediterranean and the three nonscreening programs lie within a range of 19 to 88 the Middle East, Central and South America, and North per undiscounted death prevented and 1 to 6 per undiscounted Africa), 20 to 55 percent of the population show HBV antibod- year of life gained (table 21-5). These costs, although greater ies and 2 to 7 percent are carriers; neonatal infection is fre- than those for tobacco prevention programs, make HBV im- quent. In areas of high prevalence (China, southern Asia, munization an attractive candidate for inclusion in a cancer Cancers 537 Table 21-5. Cost-Effectiveness of Hepatitis B Vaccine in Various Settings Percent GNPN per death prevented Percent GNPN per year of life gained a Setting ra- O ra- 0.03 Tar- O ra- 0.03 b Indnesia Immunize all newborns 87 404 6 25 Screen mothers and vaccinate newborns at risk 151 700 9 44 China Immunize all newboms 74 340 5 22 Screen mothers and immunize newborns at risk 77 356 5 23 I Hypothetical exampled Low-cost estimate 19 87 1 6 II igh-cost estimate 88 408 6 26 Notes: Effects include prevention of excess mortality from hepatitis B, liver cancer, and cirrhosis; it is not practical to identify liver cancer as the sole objec- tive. The discounted values assume that intervention occurs at age 0 and the average HBv-caused death occurs at age fifty-two. The GNPN is estimated at 810 yuan. a. Discount rate. b. Widjaya (1988) calculated the cost per carrier prevented. This has been converted to deaths prevented and years of life gained using an excess mortality rate of0.25 and sixteen years of life lost per HBV-associated death. Widjaya assumed vaccine cost of us$15 (does not include delivery cost). The program costs have been recalculated using vaccine cost of us$1.50 per dose and estimated delivery cost of us$1.20 per dose. Calculated for 1985, the GNPN is estimated to be 588,000 rupiah. c. Circa 1986. d. Asstimes no screening. The ITFH 1988 estimates, which are given in U.S. dollars, have been converted to percent GNPN. Assumes per capita GNP of L15$400. Soutrce: Wid]aya 1988; Bamum 1988; ITFH 1988. prevention strategy. Discounting affects the estimated cost- logical evidence is not conclusive. Consumption of foods effectiveness of HBV programs more than many other interven- contaminated with aflatoxin (produced by Aspergillus flavus tions because of the long lag between delivery of the fungal growth on stored peanuts and grains) is likely to be a intervention (which occurs at birth) and prevented death cause of primary liver cancer. With the exception of these (which, for liver cancer or cirrhosis occurs at an average age of substances, however, no single component of diet has been about fifty-five years). Using a 3 percent discount rate, we find conclusively shown either to cause or to prevent cancer (Doll that the cost per death prevented ranges from 87 to 408 and and Peto 1981; Willett and MacMahon 1984; Byers 1988). the cost per YLG ranges from 6 to 26. Even after discounting, Neverthelessanumberofintriguingrelationshipsbetweendiet HBV vaccine remains an attractive program possibility. and cancer have emerged from epidemiological studies. These The cost-effectiveness of a maternal screening program is includepositiveassociationsbetweenpoorlypreservedorpick- equal to the newbom program in the China example but is led foods and stomach cancer, traditionally prepared salted fish much less cost-effective than the newbom program in the and nasopharyngeal cancer (among southem Chinese), and Indonesia exarriple (table 21-5). In both examples, however, total fats (or perhaps saturated fats) and large bowel cancer. cost-effectiveness calculations include only the deaths directly Interesting negative (protective) associations have been ob- prevented in the newbom cohort. Additional deaths will be served between vegetable or fruit consumption (or their con- prevented from reduced transmission of hepatitis to other stituents such as beta-carotene and fiber) and a variety of cohorts in the future. This indirect effect will be substantial for epithelial cancers, most notably lung and colorectal cancer. the total immunization strategy but only modest for the screen- Estimates of the proportion of cancers attributable to diet in ing strategy. Thus the total immunization strategy is superior, the industrial world range widely from about 10 percent to because it holds out the possibility of a dramatic containment more than 70 percent. Given this uncertainty, and the lack of of hepatitis B in the future, instead of continued endemicity firm conclusions about most postulated dietary causes of can- that would result from the screening strategy. cer, it is not possible to predict the benefits (in reduced cancer) that may accrue from dietary interventions. An assessment of FOOD AND ALCOHOL. Abundant epidemiological evidence the costs and value of dietary interventions must await the links dietary habits to occurrence of cancer. There is still results offurtherstudies, which may likely include randomized considerable uncertainty, however, both about the actual di- controlled trials. Several relationships between diet and can- etary constituents that influence risk and about the proportion cer, however, are particularly relevant in the developing world, of cancer occurrence that can be explained by diet. Alcohol and these may well figure in a natural cancer control program. consuinpt ion clearly increases risk of cancers of the oral cavity Especially noteworthy are the relation between aflatoxin- and esophagus. Alcohol has been implicated as a cause of contaminated food and liver cancer and between traditional breast cancer and large bowel cancer, although the epidemio- Chinese salted fish and nasopharyngeal cancer. Other aspects 538 Howard Barnum and E. Robert Greenberg of diet that merit attention in the developing world include consumption of fresh fruits and vegetables should be carefully consumption of alcohol (related to esophageal and oral can- considered in discussions of national food policy. At the min- cers), poorly preserved foods (related to stomach cancer), imum, it would be prudent to avoid strategies that discourage and obesity. production and consumption of these foods. In countries with high rates of liver cancer, a program to Obesity itself, apart from constituenits of diet, is also associ- reduce aflatoxin intake (by better storage ofgrains and peanuts ared with cancer mortality (Lew and Garfinkel 1979; and by monitoring the aflatoxin content of commercial foods) Simopoulos 1987). The association is particularly notable for may be justifiable as an adjunct to hepatitis B virus immuniza- female breast and endometrial cancers. Of course, obesity is a tion. At present, one cannot precisely estimate the amount of complex phenomenon, and its occurtence is determined by liver cancer that could be avoided through reduction of atla- genetic factors and physical activity as well as by caloric intake. toxin. Although hepatitis B immunization programs may Still, obesity often appears to accompany other features of eventually prove to be the most cost-effective method for economic development, and it is associated with diabetes, preventing liver cancer, they will not benefit people who are cardiovascular diseases, and other conditions in addition to already infected with HBV and thus at risk for developing liver cancer. It therefore merits consideration for public health cancer. Aflatoxin appears to be an important cofactor leading preventive programs. to liver cancer development in i-BV carriers (Bulatao-Jayme Diet affects many diseases besides cancer, and it would be and others 1982; Yeh and others 1989); thus reduction of unwise to alter diet without considering all the potential aflatoxin consumption offers promise for reducing liver cancer health consequences. Also, dietary change has economic and mortality in the near future, whereas immunization programs social implications. Planning and application of dietary strat- are unlikely to show a substantial benefit for forty or more egies to prevent cancer should therefore occur only within the years. Consumption of traditionally prepared salted fish in context of an overall national policy on diet. The relative southern China (and in areas settled by mnigrants from southern importance of nutritional deficiency diseases, obesity, cardio- China, such as Singapore) is strongly related to nasopharyn- vascular disease, diabetes, cancer, and other diet-related con- geal cancer (Yu and others 1986). Educational programs to ditions must be considered country by country. Likewise, the discourage the practice of feeding salted fish to infants, wean- expected economic consequences, in changing agriculture and lings, and children may be jusrifiable in these populations. It food marketing, must be considered if a successful dietary also seems reasonable to discourage consumption of other intervention is to be accomplished. traditionally prepared foods that are high in volatile nitrosa- mines in countries where nasopharyngeal cancer is common ENVIRONMENTAL AND OCCUPATIONAL PROGRAMs. Exposure to (Poirier and others 1987). environmental and occupational factors accounts for an unde- In certain countries the possibility of reducing cancer occur- ternmined but probably small number of cancers in the devel- rence may provide additional justification for implementing oping world. In the industrial world the actual contribution of programs to decrease alcohol consumption, although the pri- such factors to cancer occurrence has been the subject of mary value of these programs is to reduce injuries and other considerable debate (Doll andPeto 1981). Muchconfusionhas ills. The cancer prevention benefits of reduced alcohol con- arisen about the definition of "environmental" causes of can- sumption (historically, difficult to achieve through govern- cer. If "environmental factors" are taken to mean only agents ment policy) cannot be reliably assessed, however, because the in the ambient environment (excluding diet and habits such proportion of oral and esophageal cancers caused by alcohol is as smoking, chewing, and so on), then only a small proportion uncertain. The contribution of tobacco use to these cancers (probably less than 10 percent) of cancers can be attributed to appears to be greater than that of alcohol. these causes. Nevertheless, in certain highly exposed groups, Many other foods and nutrients are strongly suspected of usually defined by occupation, environmental agents are more influencing cancer occurrence, and these have figured in na- important. tional dietary policies in the industrial world. For example, Several environmental agents have been identified as pos- Americans are advised to eat a diet containing lessfat and more sible causes of lung cancer. These include both ambient and fiber and to consume larger amounts of fresh fruits and vege- occupational exposure to industrial chemicals and radon. tables. There is no conclusive information from scientific These causes are important in selected subpopulations in studies that these recommendations will lower cancer risks. which exposure is intense. For example, workers involved in Nevertheless, several expert review groups have concluded the manufacture of asbestos, chromates, and ion exchange that one or more of these recommendations should be pre- resins (involving chloromethylethers) are all at elevated risk sented to the public as prudent advice for dietary change. of lung cancer. Likewise, underground miners exposed to high The justification for advising people to eat more fruits and levels of radon have an increased risk of lung cancer. vegetables is that epiderniological stLdies have shown de- Cigarette smoking acts with environmental factors to in- creased risks for many types of cancer in people whose diet is crease greatly the risk of lung cancer. Before implementation high in these foods (National Research Council 1982). Also of protective measures for asbestos workers in the United fruits and vegetables may displace other, less desirable, foods States, asbestos workers who smoked cigarettes had approxi- fromn the diet. The evidence linking lowered risk of cancer with mately five times the risk of lung cancer as smokers who did Carners 539 not work with asbestos and more than fifty times the risk of breast cancer hold greatest promise as candidates for national- people who neither smoked nor worked with asbestos. Simi- level screening programs. These are discussed in detail later. tarly, the risk for uranium miners who smoke cigarettes is four Large bowel cancer and oral cancer are also of interest, times the risk of smokers who are not miners. Nonetheless, although evidence supporting the value of screening for these within the United States population as a whole, these other two tumors is inconclusive. Several procedures for colorectal environmentalfactorsareoflittlesignificancewhencompared cancer screening have been proposed, but none has yet been with the effect of tobacco smoking. After taking account of the shown clearly to reduce the risk of death from this condition effects of other known causes of lung cancer, tobacco smoking (Chamberlain and others 1986; Clayman 1989; Fleischer and still appears to account for at least 80 to 90 percent of lung others 1989; Knight, Fielding, and Battista 1989). Institution cancer deaths in North America (Doll and Peto 1981). of screeningprogramsscannotnow berecommendedasapublic Present information does not appear to warrant a general policy, but further tests of screening efficacy and controlled recommendation regarding environmental and occupational clinical trials are clearly worth undertaking. Examination by interventions to control cancer. For limited exposed popula- flexible sigmoidoscopy appears particularly promising as a pro- tions these types of intervention may be cost-effective or cedure. although not in countries with limited health re- necessary on moral grounds. Asbestos merits particular con- sources. Screening for oral cancer in conjunction with other cern in this regard for two principal reasons. First, it is widely medical procedures, such as dental care or general care, theo- used in many developing countries in applications for which retically should prevent deathsfrom thisdisease. Theexpected there are no cheap substitutes. Second, the rising prevalence low cost of screening (by simple inspection), and the possibility of cigarette smoking greatly magnifies the potential for asbes- of identifying premalignant changes (leukoplakia), make oral tos-related lung cancer. Programs to reduce workers' exposure cancer a good prospect for intervention (McMichael 1984). and to use safer forms of asbestos will likely be worthwhile for Also, high-risk individuals (those who chew or smoke tobacco) many countries. For much of the developing world, however, are readily identifiable. There have not, however, been anv control of environmental and occupational carcinogens ap- published studies showing the effectiveness of oral cancer pears to deserve a lower priority than other, proven cancer screening. In countries, such as India, with high mortality rates prevention activities. for oral cancer, screening programs may now be worth under- taking but only within a research context. CANCER PREVENTION LENEFITS AND OTHER HEALTH MEASURES. Most othier cancers such as lung, liver, esophageal, and Programs directed toward other important diseases may, as an stomach cancer are not suitable targets for a screening program ancillary benefit, also help to control cancer. For example, in a developing country.' Although they can be detected in a schistosomiasis is clearly associated with urinary tract cancer presymptomatic phase, no controlled studies have shown that (in the case of Schistosomiasis hematobium) and possibly related screening reduces the risk of death from these tumors. Also, to bowel cancer (in the case of S. japonicctm and S. mansoni). these tumors tend to progress rapidly to a clinically symptom- An evaluation of the cost-effectiveness of a mass treatment atic phase, so there is only a brief period during which the program to control S. hemnatobium, for example, should take cancer is screen detectable but not symptomatic. account of the expected reduction in urinary cancers that would result. Likewise, control of Cloriorchis and Opisthorchis CERVICALCANCER SCREENING tPAPSMEARI. Because of its high infection, besides reducing acute morbidity, will lessen the risk incidence in many lower-income countries, screening for cer- of biliary tract cancer. Last, programs to decrease spread of vical cancer is among the most promising of the secondary sexually transmitted diseases seem likely to diminish cervical prevention possibilities. The objective of cervical cancer cancer occurrence. In all these examples the reduction of screening is to detect neoplastic cells at an early stage when a cancer occurrence would be a secondary benefit that would be relatively low-cost and low-risk surgical procedure can be used realized only vears (or decades) after an effective program was to remove the cells and prevent occurrence or spread of inva- implemented. sive cancer. The cost-effectiveness of cervical cancer screening has been debated in the literature, but evidence in support of Secondary Prevention screening has become compelling (Cramer 1982; Lynge, Madsen, and Engholm 1989), and cervical cancer screening The cost-effectiveness of secondary prevention (that is, is an important component in the World Health Organiza- screening and early detection) programs is dependent on the tion's strategy for combating cancer (WHCO 1988b). The cost- incidence of the disease in question, the technical feasibility effectiveness of screening, however, varies greatly with the of screening and treatment at early stages in the cancer's setting in which a program is carried out. Important factors development, the possibility of targeting to reduce costs by determiniing the advisability of the test include incidence and covering groups at highest risk, and the availability and cost of mortality rates (and therefore the prevalence of occult dis- appropriate health infrastructure so that the screening can be ease), type of program (mass screening or integrated medical carried out with accuracy and the findings followed with an examination), availability of adequate facilities for following effective intervention. Among the ten primary cancers, on the the test findings, accuracy of laboratory facilities, and cost of basis of the experience in industrial countries, cervix and the test and follow-up procedures. 540 Howard Barnum and E. Robert Greenberg The World Health Organization (1 986b) has recommended physical examination combined with mammography can re- that in low-income countries every woman should be screened duce breast cancer mortality by about 20 to 30 percent (Sha- once in her lifetime between the ages of thirty-five and forty piro and others 1982; Tabar and others 1985; Day and others years. If additional resources are available, screening should 1986). Breast self-examination has also been promoted as a take place at intervals of ten years for women between the ages secondary prevention policy, but evidence for its effectiveness of thirty-five and fifty-five. In middle-income countries the in reducing mortality is weak (Day and others 1986). interval should be increased to five years. The cost-effectiveness The cost-effectiveness of breast examination using mam- of tests falls off rapidly as the frequency of testing goes much mography is much less than that of cervical cancer screening above three to five years and as the age of testing falls below in the examples we considered. The estimated cost per YLG thirty-five. from mammography ranges from about 100 to 200 percent Estimates of the cost-effectiveness of cervical cancer screen- GNPN using a discount rate of 3 percent. The estimated cost per ing, in percentage of per capita GNP per year of life gained, YLG from the physical examination alone is, however, compet- range from about 70 to 300 percent per undiscounted death itive with other prevention activities. Using only a physical prevented, or 4 to 18 percent per year of life gained (table examination and discounting costs and effects at 3 percent, the 21-6). Because of the brief time between the start of screening cost per YLG was 12 percent GNPN as recalculated from the and the average age of death, the effect of discounting is not results reported in a study by D. Eddy (1981; see table 21-7). great. Using a discount rate of 3 percent, we find that the cost per year of life gained ranges from 5 to 26 percent GNPN. At the Treatment upper extreme, cervical cancer screening is not competitive with other health interventions, whereas at the lower extreme The three primary modalities for treating cancer are surgery, the program would be an attractive component of a chronic radiation therapy, and chemotherapy (including hormonal disease strategy. Clarification of the epidemiological basis and manipulation). Surgery alone offers a chance of cure for can- cost-effectiveness of program design should be made in indi- cers of the breast, uterine cervix, colorectum, and oral cavity. vidual countries before embarking on a program of secondary Although some cancers of the stomach, lung, liver, and esoph- cervical cancer prevention. Cost-effectiveness will be greater agus are also curable by surgery, their number is very small. The in countries with higher rates of cervical cancer and in those surgical treatment required for early stage cancers of the breast, with clearly identifiable high-risk groups. It will be lower where cervix, colorectum, and oral cavity does not require highly cultural norms prevent women from having pelvic examina- technological facilities, nor does it require training beyond tions and where the quality of cytology laboratory work is low. that ordinarily received by surgical specialists. The procedures generally take approximately two to four hours to complete BREAST CANCER SCREENING. Breast cancer screening can be and are associated with a hospital stay of seven to fourteen carried out through a physical examination by a health care days. Accurate assessment of the degree of cancer spread worker (Baines, Miller, and Bassett 1989) or a combination of (staging) is important so that performing surgery is avoided on a physical examination and mammography. Controlled clini- patients in whom the procedure cannot be curative, and it cal studies with long-term follow-up have demonstrated that a necessitates reliable diagnostic imaging facilities. Table 21-6. Cost-Effectiveness of Screening for Cervical Cancer Percent GNPN per death prevented Percent GNPN per year of life gained Study ra-0 ra- 0.03 ra- 0 ra-0.03 Parkin and Moss (United Kingdom) Five-year interval, age 35-64 66 94 4 5 Five-year interval, age 25-64 154 220 7 11 Barnum (China) Five-year interval, age 35-59 310 390 18 26 Luce (United States)d Five-year interval, age 30-39 138 250 13 25 Three-year interval, age 30-39 145 290 12 23 a. Discount rate. b. Based on model identified as `H3C" in Parkin and Moss 1986. Their results are scaled in terms of unit costs representing the cost for a routine screening. To convert their results to percent GNPN, I unit of input was evaluated as 4.6Y (1982 prices), using a price index for medical services in the United Kingdom, and then rescaled using a 1982 GNPN of 4,907L. Discounting assumes an average of twelve years from the age of screening to the averted cancer mortality. c. Estimates are not based on as complete a model as other two studies. For example, false positives and false negatives are nor included. d. Luce results reported in terms of 1979 U.S. dollars per death prevented and year of life gained from using a 10 percent discount rate. These have been converted by interpolating from Luce's discount rate sensitivity table and using a 1979 GNPN of us$10,810. Luce's "Low cost" public provider model has been used. Source. LUce 1980; Parkin and Moss 1986; Bamum 1988. Cancers 541 Table 21 7. Cost-Effectiveness of Screening for Breast Cancer Percent GNPN per death prevented Percent GNPN per year of life ganed Study ra-0 ra-O.03 ra - Q ra - 0.03 Eddyb Physical exam only - 12 Physical exam and mammoqraphy, one-year interval, age 50 and older d - 97-210 Schwartz, three-year interval, age 40-70 1 5 135 - Data nor available. a. Discount rate. b. Assumes U.S. CNFPN of $12,800. c. The costs and effects are incremental with the addition of mammography to a routine annual physical examination. Based on U.S. GNPN of $10,600 (1979). Based on present value of total incremental cost ranging from $482 to $1,042. Based on the increase in life expectancy from age forty. Assumes a cost of us$ 100 per screening and eleven screenings between age forty and seventy. Based on U.S. GNPN of $10,600 (1979). Source: Schwartz 1978; Eddy 1981. Radiation therapy involves administration of high-energy cytic leukemia, treatment may continue for a prolonged period radiation in an effort to kill tumor cells. The technique requires and require repeated hospitalization, intensive chemotherapy, sophisticated equipment and skilled therapists, who are likely and radiation therapy. During the course of treatment the child to be found only in technologically advanced tertiary care may require sixty to ninety days in the hospital. centers. Radiation therapy is capable of curing lymphoma A policy decision to provide access to this type of techno- (particularly Hodgkin's Iymphoma) and cancer of the cervix. logically advanced treatment has implications beyond the Radiation therapy is useful for control of breast cancer and lung immediate cost of the procedure. First, a facility must be cancer, although it only rarely cures lung cancer. Most courses developed that has the technological capacity to support ad- of radiation therapy must be administered over a prolonged vanced therapies. This involves diagnostic imaging capabili- period (usually several weeks), but the patient may not have ties, advanced laboratory facilities, and radiation therapy to be kept in the hospital. devices. Second, highly trained and experienced personnel Chemotherapy is administration of drugs that kill cells or are needed from both the medical and nursing professions to inhibit their growth. Chemotherapy is potentially curative in carry out the therapy. Third, the patients who are likely to leukemia and certain lymphomas. It also contributes to length benefit from advanced multimodal therapy are relatively few of survival and possibility of cure in breast cancer. Chemother- in number, so referral to a few specialized treatment facilities apy provides palliation in certain forms of lung cancer. Inten- will be necessary. To provide ready access to advanced can- sive chemotherapy such as that given for most lymphomas and cer treatment will require the development of a system of central leukemias requires highly trained physicians who have sub- treatment facilities with an outlying referral network. This stantial experience with administration of these toxic agents. implies a heavy investment in facilities located in urban Less toxic chemotherapy for relief of symptoms often can be areas and will further concentrate resources away from rural carried out in primary medical settings, providing that consul- populations. tant guidance is readily available. Many of the drugs used for chemotherapy are expensive, and they require close monitor- Cost-Effectiveness of Treatment ing of laboratory tests and intense skilled nursing support because of side effects. An exception is the antiestrogen drug Given the resource intensity of cancer treatment, evaluation Tamoxifen. It clearly reduces risk of recurrence and death in of its cost-effectiveness is particularly important to make in- breast cancer yet is relatively free of toxic side effects, is simple formed planning decisions. Cancers differ greatly in severity to administer in oral form, and should be available at low cost and potential for treatment or prevention, and, for the most in most countries (WHO 1985). part, it is necessary to consider the effects and costs for separate More recent advances in cancer treatment have shown the types of cancer. In the remainder of this section we provide a importance of multidisciplinary management, involving more cost-effective analysis of cancer treatment using United States than one treatment modality. This strategy requires the com- data to model effectiveness and general information to model bined efforts of highly trained professionals and is possible only costs in a prototypical country. The results are acknowledged in the context of a technologically sophisticated tertiary care to be highly approximate, and our intention is to allow rough center. Many of the gains noted in the probability of survival comparisons of the cost-effectiveness of treatment and preven- of patients with certain cancers (particularly leukemias and tion. For certain cancers, notably cervix and breast, and per- lymphomas) are attributable to use of multimodal therapy. haps colorectal and oral cancer, the close link between Application of the most advanced therapy can be vastly secondary prevention and treatment prevents a clear assign- expensive. For some cancers, for example, childhood lympho- ment of gains in survival to treatment per se. For these cancers 542 Howard Barnum and E. Robert Greenkerg screening and treatment need to be considered as joint inter- entirely attributable to improved therapy. Part of the recorded ventions if the fLll cost-effectiveness of treatment is to be increase may also be due both to earlier diagnosis and to achieved. In fact, the cost-effectiveness oftreatment ofcancers detection of less aggressive tumors, factors which give an for which screening is feasible exceeds the estimates in this illusion of improved survival without any true change in the section (which are based on an average for all stages), because underlyinlg risk of death. The size of these effects is debated in the effect of treatment on early stage cancer is greater and, in the literatlre and unknown, but in any case, the apparent some instances, the cost of treatment is less. increase in survival almost certainly overestimates the effect of new therapy techniques. EFFECTS. We approximate the effectiveness of treatment The data show very little improvement in survival for using data on the change in survival rates for cancer in the esophageal, stomach, and liver cancer. Lung cancer im- United States from the period 1945-50 through the years provements that are shown are small and difficult to attribute 1955-60 to 1975-80. Conceptually, the objective is to esti- totechnical improvements. Although therehave been modest mate the benefits in developing countries of treatment in a gains in treating small cell carcinoma, this cell type represents higher-level hospital with relatively modern technology in less than 20 percent of all lung cancer. Much of the increase comparison with treatment at an entry-level hospital or with that is shown is likely an art-ifact resulting from earlier diagno- no treatment or only palliative care. As an approximation, the sis. Improved survival for leukemia (and perhaps for breast higher-level hospital is equated with the level of care in the cancer) is more reflective of the effects of new therapies. In the United States in the period 1975-80, and the low-level care analysis below we lack the quantitative information to make alternative is equated with the level of treatment in the United an adjustment for the effects of earlier diagnosis and different States during 1945-5O. Baseline, minimal treatment survival spectra of tumor aggressiveness, but these sources of error rates are specified arbitrarily after a literature search of techni- should be borne in mind when one interprets the results. cal improvements in cancer prior to 1940. The effectiveness of treatment is estimated by calculating Equating institutional care at various levels with chronolog- the implied gain in years of life in going from the baseline ical change in cancer care in the United States is obviously fatality rates that would exist with only minimal treatment to only an oversimplification because,con the one hand, there are the fatality rates after treatment equivalent to the United a few modern techniques that can be delivered with effect and States level in the period 1975-80. The fatality rates implied at low cost at smaller, low-level institutions and, on the other by the published survival data and the baseline fatality rate hand, there were some effective treatments used in large, used in the analysis are summarized in table 21-9. technically advanced hospitals in the United States in the years 1945-50. During the last forty years, however, there have COSTS. Surprisingly few studies have been done on the total been substantial advances in radiation, chemotherapy, and direct costs incurred over the course of specific cancers. Cost- surgery. Radiation therapy has advanced through development ing cancer is particularly difficult because for given cancers of new sources of radiation and techniques that allow better there may be a number of treatment options at each stage of control of the intensity and the focus of radiation. Chemother- the disease, and the duration of treatment may last several apy has advanced with the development of new drugs and their years. Also, treatment procedures and therefore the costs of use in combined therapies. Surgery has improved in safety and treatment vary considerably for the different types of cancer. precision. Another change during the thirty-year period has been in the accuracy of diagnostics, which has made it possible to identify cancers at an earlier stage. Table 21-9. Estimated Five-Year Fatality for Canc-er, All these factors are likely contributors to the changes that by Year can be seen in relative survival in table 21-8.7 It would be a - Baseline mistake to view the apparent improvements in survival as mimimal treat- Site or type of cancer ment' 1950 1960 1970 1980 Table 21-8. Five-Year Relative Survival for Cancer --- -- - - -- - - in the United States, by Year Mouth/pharynix 0.82 0.58 0.58 0.60 0.50 -- ---- ----- Esophagus 0.97 0.97 0.97 0.97 0.95 Site or type of cancer 1950 1960 1970 1980 Stomach 0.90 0.90 0.91 0.89 0.86 Mouth/pharynx 0.45 0.45 0.43 0.53 Coloni/rectum 0.75 0.64 0.61 0.59 0.52 Esophagus 0.04 0.04 0.04 0.06 Liver 0.99 0.99 0.98 0.97 0.97 Stomach 0.12 Q.11 0.13 0.16 Lung 0.95 0.95 0.93 0.91 0.89 Colon/rectum 0.41 0.44 0.46 0.53 Breast 0.75 0.47 0.43 0.38 0.31 Liver 0.01 0.02 0.03 0.03 Cervix 0.80 0.50 0.50 0.45 0.41 Lung 0.06 0.08 0.10 0.13 Leukemia 0.90 0.90 0.86 0.78 0.67 Breast 0.60 0.63 0.68 0.75 Note: Five-vear fatality computed as F = I - R * S. where R is the relative Cervix 0.59 0.58 0.64 0.67 survival given in table 21-8 and S is overall survival (fir all stages) from Leukemia 0.10 0.14 0.22 0.33 United States life tables fir 1950-SO. a. Either 1950 fatality or higher rate implied froim literature search. Source: Axtell, Asire, and Myers 1976; American Cancer Society 1987. Source: Axtell, Asire, and Myers 1976; American Cancer Society 1987. Cancers 543 For the purposes of this study, we approximated the costs of suggest that it is much more cost-effective to treat cancers at treatment using information on the relative costs of treatment younger ages; in particular, the cost per year of life gained for for different cancers in the United States (Cromwell and treatments at age fifty and over are very high. Gertman 1979; Rice, Hodgson, and Kopstein 1985). These The analysis above underestimates the true costs and over- costs were then scaled to the cost of tertiary treatment and estimates the effects and therefore must be interpreted cau- expressed as a percentage of per capita GNP for a lower-middle- tiously. First, only the direct costs of health services are income developing country. Although this procedure is inex- included. Costs to patients and their families in travel to the act, it is probably sufficient for the comparisons between hospital and provision of supplementary care or food are not cancers and between prevention and treatment that are sought included. Second, for several cancers, especially of a certain here. In table 2 1-10 we present the estimates of total costs per type and in advanced stages, the number of years of additional case, varying fromalowof about 175 percent GNPN for cervical life is very small and may be largely an artifact of earlier cancer to 780 percent G;NPN for lung cancer. detection, diagnosis of less malignant tumors, and staging changes (Feinstein, Sosin, and Wells 1985). Third, there is no COST-EFFECTIVENESS. The costs and effects derived in the two adjustment for quality of life. This omission can be especially preceding sections can be compared to give the costs per year important for the marginal gains (in extended life) for treat- of life gained from treatment. Looking first at the cost per YLG ment of later stage cancer and for treatment of esophageal, for the average of cancer in all stages (table 21-1 I ), we find stomach, lung, or other cancers for which the gain in fractional that the results suggest that it is relatively cost-ineffective to years of life follows debilitating treatments that usually result treat esophageal, stomach, liver, and lung cancer and more in only short periods of remission. cost-effective to treat colorectal, cervical, and breast cancer. The analysis is also based on improvements in survival rates To compare extremes, the cost per year of life gained from in the United States, where there has been an increased treatment for esophageal cancer is more than 100 times that emphasis on early detection during the last forty years and for cancers of the breast, mouth, and cervix. The results also where there is substantial training and treatment capacity. This link between detection and treatment is especially im- portant as an underlying factor in the cost-effectiveness of Table 21-10. Cost per Case Treated treatment for cancers of the mouth, cervix, breast, and colon. (percent GNI'N) It is important to emphasize the link between detection and Costs per case in Cost per case in treatment because it is not useful to develop treatment capac- Site or type Costs relative to high-income lower-middle- ity in the absence of improved detection; nor is it useful to of cancer average for _ c_ntrNb income countryC develop detection programs in the absence of treatment capac- Mouth/pharvnx 0.76 79 243 ity. This link has been underlined by the World Health Orga- Esophagus 1.11 115 709 nization, which stresses the importance of linking the Stomach 1.07 112 687 development of therapy with early referral (WHO uses the term Colon/rectum 1.05 110 336 "down staging") and improved training for cancer detection Liver 1.13 118 727 among primary health care workers (Stjernsward 1990). Lung 1.22 127 782 Breast 0.65 67 206 PAIN RELIEF. An important component of palliative care for Cervix 0.5049 154 700 cancer patients is adequate treatment ofpain (Portenoy 1988). More than 80 percent of cancer is not detected until an Average 1.00 104 641 advanced stage at which treatment other than palliation is not a. Based on 1969-71 average direct costs for cancer treatments as re- effective (Stjemnsward 1988). A VJHO document setting out ported in Cromwell and Gertman 1979. guidelines on cancer pain relief estimates that 50 to 70 percent b. The number of hed-day equivalents for the cost of the average cancer of cancer patients experience pain (WHO 1 986a). Of these, pain in the United States was multiplied by the cost of a bed-day (expressed as percent GNPN) to obtain the cOst of an average case of cancer. This average is moderate to severe in about 50 percent and very severe in was, in tum, multiplied by the relative costs in the first column to obtain 30 percent. The World Health Organization recommends a the costs for individtial cancers. In 1980 the average direct cost of cancer in three-stage analgesic program going from nonopioids (such as the United States was fifty-three times the cost of the average bed-day based on a reanalvsis of data in Rice, Hodgson, and Kopstein 1985 andL aspirin or paracetamol) for mild pain, through weak opioids thirty-seven times based on 1969-71 data in Cromwell and Gertman 1979; (such as codeine plus paracetamol) for moderate pain, to strong 40 is used for these calculations. The percentage GNFN per bed-day in a opioids (morphine) for intense pain. They note that these large urban hospital remains relatively stable over a sample of countries, 2.6 was used for these calculations. agents, possibly supplemented by adjuvant drugs (see the de- c. Because the technical hospital procedures involved in cancer therapy tails in WHO 1986a and Stjernsward 1988), can provide relief can have a high foreign exchange cointent when cost is determined by inter- in 90 percent of cancer patients with pain. national prices, the estimated cost i, a weighted sum of a foreign exchange component (including specialized training, equipment, and pharmaceiiti- Past inadequacy of cancer pain control is attributable to a cals) and a local component. The calcuilations are made for a lower-middle- lack of recognition by health care professionals that effective incoime country with a per capita GNP of us$I 50X. As an approximation, it methods existed for cancer pain management, a lack of avail- is assumed that the foreign exchange content is 0.2 tor early stages of oral, cervical, breast, and rectal caricer treatment and 0.5 for al) other cancers. ability of the required drugs, unreasonable fears concerning Source: Cromwell and Gertman 1979; Rice, Hodgson, and Kopstein 1985. addiction, and poor education of health professionals on can- 544 Howard Barnum and E. Robert Greenberg Table 21-I1. Cost per YLGG ained from Tertiary-Level Treatment in a Lower-Middle-Income Country, by Age at Diagnosis (percent GNPN) Sire Average age 65 years 60 years 55 years 50 years 45 Nears 40 vears Undiscounted Mouth/pharynx 44 179 80 51 38 30 25 Esophagus 3,574 8,755 3,891 2,502 1,843 1,459 1,208 Stomach 2,462 6,525 2,900 1,864 1,374 1,087 900 Colon/rectum 126 472 210 135 99 79 65 Liver 3,315 13,011 5,783 3,717 2,739 2,168 1,795 Lung 1,183 3,075 1,367 879 647 512 424 Breast 26 93 46 31 23 19 16 Cervix 27 69 35 23 17 14 12 Discounted at 3 percent Mouth/pharynx 55 187 89 62 49 41 36 Esophagus 4,056 9,147 4,367 3,010 2,374 2,008 1,772 Stomach 2,826 6,817 3,254 2,243 1,769 1,496 1,320 Colon/rectum 154 493 235 162 128 108 96 Liver 4,083 13,593 6,489 4.473 3,528 2,984 2,633 Lung 1,354 3,213 1,534 1,057 834 705 622 Breast 33 122 58 40 32 27 24 Cervix 32 90 43 30 23 20 18 Note: Costs are averaged over a[l stages. Costs per YLG gained from tertiary level treatment of leukemia diagnosed in a ten-year-old are 48 percent of GNP undiscounted and 100 percent discounted at 3 percent. Source: Authors. cer pain management. Legislative reform, better pharmaceuti- * Provide for decentralized supportive care and symptom cal management, and improved training can remove these relief for the vast majority of other cancer patients using blocks to better pain therapy. community-based alternatives for palliative care. For pa- Adequate treatment of pain could alleviate much suffering tients with breast cancer, Tamoxifen therapy could also be while placing fewer demands on medical resources than inef- accomplished at this level. fective attempts at higher-level curative treatment. For exam- Such alternatives include community-staffed nursing pIe, for an average duration of pain therapy of ninety days, the homes, health center outpatient care and hospices for the cost of drugs and outpatient delivery in a lower-middle-income terminally ill, and home beds managed from lower-level hos- developing country would be about $18 to $65 per case, or 1 top4npe cntry for, respect, mil to severe pain man- pitals. Almost all patients with stomach, liver, and lung cancer toa4gement This would respresent les th 1eperepn oftecs would receive care at this level. Older patients and those with agement.' This would represent less than I percetit of the cost advanced disease would not be referred for curative therapy, Of inpatient tertiary treatment of, sav Iling or esophageal ofainpatient t but reasonable efforts would be made to alleviate pain and cancer. suffering through use of palliative medications. COST-EFFECTIVE TREATMENT POLICY. The results of the cost- effectiveness analysis can be helpful in fashioning a policy to Developing a Strategy for Cancer increase dramatically the effective use of secondary and terti- ary hospital resources in place of available space to treat all The World Health Organization has recommended the devel- cancer on the basis of random referral or first come, first serve. opment of national cancer strategies based on surveillance and Applying a criteria of no more than 50 percent GNPN per year prevention programs tailored to local needs (Stjernsward and of life saved and bearing in mind the cautionary comments others 1985; WHO 1988b). The recommendations of WHO for noted in the preceding paragraph, a possible policy would be the development of a national cancer strategy are especially the following: important in countries in which infectious diseases have been reduced and life expectancies are lengthening. It is not effec- * Use existing referral capacity to treat younger patients tive to allow the resources devoted to combating cancer to be with early stage cancers amenable to curative treatment, allocated without plan on an ad hoc basis depending on local especially breast, mouth, cervical, and possibly colorectal interests. Instead, a national cancer strategy needs to be devel- cancer. oped formally. Such a strategy would recognize local variations * Use existing tertiary-level referral capacity to treat chil- in cancer occurrence and risk factor prevalence. It would also dren with leukemia and lymphoma. allow setting an overall national strategy to combat cancer and Cancers 545 Table 21-12. Subjective Potential for Prinmary Preventive Activities for Cancer Potential reductioni Potential technical Persons covered Activitv __ _ ____ of YLL Feasibilitya effectb per unit cost Priority Antismoking measures High MediumLI High High Highest Hepatitis immunization Medium High High Medium High Control of carcinogens in foodc Low Medium Medium High Medium Reduced fat and increased fiber in diet Unknown Low Unknown High Low Occupational hazards Agriculture: pesticides, fertilizers, equipment Low Medium High Low Low Mining (dust exposure) Low High Medium Medium Medium Industrial safety Low High High Medium Medium Air pollution control Low Low Low Low Low Home environment: heating Medium Low Medium Low Medium Genetic screening and counseling Low Low Medium Low Low a. As demonstrated, for example, from implementation in other countries. h. Related to prevalence of diseases affected, age and social characteristics of people affected, and effectiveness of program. A possible measure of effective- ness is healthy days of life saved. c. Including aflatoxin. Source Authors. mobilization of national resources to reduce cancer morbidity would likely be at least several times less costly per year of life and mortality. This can be accomplished by setting priorities saved than any other anticancer program. Smoking prevention based on epidemiological information, demonstrating the and cessation programs could reduce lung and other cancers as value of specific cancer control activities, and anticipating well as provide benefits in reduced mortality from cardiovas- changes in cancer incidence and resource needs. Several coun- cular and nonmalignant respiratory diseases. Programs to re- tries have developed national cancer control strategies on the duce tobacco chewing are needed in countries in which oral basis of WHO recommendations. (Examples are Chile [WHO cancer is prevalent. Hepatitis immunization is also likely to be 1988al; Indonesia [wHO/Indonesia, Ministry of Health 19891; cost-effective and should be a part of national strategies in India [India, Ministry of Health and Family Welfare 1984; Nair countries in which HBV incidence is high, perhaps in conjunc- 1988; and Bhargava, n.d.]; and Sri Lanka [Warnakulasuriya tion with efforts to reduce aflatoxin exposure. These findings and others 19841.) are consistent with a subjective evaluation of alternative pri- The cost-effectiveness analysis in the preceding sections mary prevention possibilities in the context of the limited is suggestive of appropriate priorities and does not take the resources of developing countries. A list of potential primary place of the analysis of country-specific situations. The analy- prevention activities is given in table 21-12. The activities are sis does, however, provide an indication of the main compo- rated as to the importance of the associated disease as a cause nents of a general cancer strategy. This section recommends a of years of life lost, feasibility (including ease of behavioral four-component general strategy for cancer consisting of pri- modification), potential technical effectiveness, and cost per mary prevention, secondary prevention, case management, person covered. and surveillance. The strategy is in direct contrast to the current picture in Europe and the United States in that it Secondary Prevention places less emphasis on centralized higher-level technology and acute care and greater emphasis on diffused basic institu- Analysis of the cost-effectiveness of individual secondary pre- tional care and prevention. The strategy is based on the supe- vention programs is needed in specific country situations be- rior cost-effectiveness of prevention and lower-level case fore they can be adopted as components of a national cancer management. strategy. Of all secondary prevention possibilities, cervical cancer screening through periodic pelvic examinations and Primnary Prevention Pap tests shows the most potential as a universal strategy component. In table 21-13 we list secondary prevention activ- Primary prevention programs are the key element in a strategy ities and give an evaluation of their potential as components for cancer. Cancers targeted by the strategy and the program of a national strategy. Again, as with the primary prevention design need to be consistent with the pattern of cancers and activities, the list is subjective and only meant to be used as a risk factors in individual countries. The analysis above suggests basis for discussion. A salient aspect of the table, confirmed in that antismoking programs are of utmost priority and ought to several studies of specific secondary prevention activities, is be implemented as soon as possible. An antismoking campaign the high cost of vertical secondary prevention programs in would be the single most cost-effective preventive activity and relation to benefits. 546 Howard Bcarnum and E. Robert Greenberg The high cost is a result of the fact that often thousands of patients must be made on a disease-by-disease basis and with people must be screened to identify one case of the disease and differentiation by stage of disease and other prognostic factors. that most people who test positive in a screening program are Limited centralized facilities for treatment of selected re- found (on definitive testing) not to have cancer. Thus, the cost ferred cancer patients may be appropriate as discussed in the of screening per new case can be very high. Generally, second- section on treatment, above, especially if improved treatment ary prevention is more attractive if the screening and early capacity is coordinated with earlier referral and diagnosis. treatment costs are low, earlv therapy is effective, and the cost Criteria for referral to treatment in higher levels of the health of late treatment is high (Kristein, Arnold, and Wynider 1977). care system would make it possible to achieve longer and Secondary prevention generally appears less cost-effective higher-quality lives for more people with existing and pro- thani smoking prevention and HBv immunization. Still, well- jected levels of resources. For most cancers, however, treat- planned and executed programs of cervical and (perhaps) ment with higher-level tertiary care is not cost-effective, and breast cancer screening (using physical examination without the investment by low-income countries in the special capac- mammography) directed at high-risk women may be cost- ity for cancer treatment is not warranted. The small gains in effective in areas of high incidence. Other screening programs lite expectancy that may be achieved are often associated with are either of unproven benefit or do not seem highly cost- a poor quality of existence, great discomfort, and stress. Use of effective in most situations. But as the health sector develops scarce technical medical resources to treat cancer reduces their and evidence of effectiveness accumulates, additional screen- availability for other diseases for which there is greater proba- ing strategies may become cost-effective. Especially strong bility of a favorable outcome from treatment and a higher candidate programs for future evaluation are screening for oral quality of the years of life saved. Instead, development of less and large bowel cancer. centralized, lower-level, treatment or case management alter- Studies in the United Statesand Canada indicate that many natives, including adequate pain relief, would not only save secondary prevention programs are not cost-effective individ- resources but would also allow more humane care for the ually but may be important components of an integrated terminally ill. program. For example, pelvic examinations carried out as a Alternative lower-level and community-based care that result of mass vertical campaigns have been shown to be could be developed includes substitution of outpatient for expensive per year of additional life gained among U.S. inpatient care; use of nonphysician professionals; use of home women, but when carried out as an integrated part of other beds, community nursing care, and hospices; and an emphasis activities, such as a periodic check-up or for contraceptive on palliative care rather than cure. Such care is not only of visits, the cost-effectiveness increases. Targeting also increases lower cost but may also be more humane. the cost-effectiveness of secondary prevention activities. Surveillance and Research Case Management Rational planning of health resource allocation requires a Given the inadequacy of existing and projected resources to continuous flow of information on disease occurrence, preva- cope with chronic disease through conventional tertiary-level lence of etiological factors, and the cost and effectiveness of care, alternative modes of case management must be devel- health interventions. This information can be used to set oped. Expansion of treatment care to cope with all cancer at health investment priorities, to develop guidelines or triage the secondary or tertiary level would far exceed projected rules for the use of health services, and to examine the cost- resources during the next twenty years in developing countries, effectiveness of health activities from broad aspects of program and in any case it is not advisable, given the practicality of strategy down to the assessment of specific technologies. lower-level case management. Criteria need to be defined for Ongoing data collection and research programs in most developing and using upper-level and specialized facilities. To developing countries do nor provide a good basis for building facilitate development of treatment policies, an evaluation of the capacity to analyze health program effectiveness in the the effectiveness of treatment in extending the life of cancer future. These programs need to be strengthened and use made Table 21-13. Subjective Potential for Secondary Preventive Activities for Cancer Potential reduction Potental Persons covered Activity of YLL Feasibility' technical effect per unit cost Priority Pelvic exam Medium High Medium Low Medium Rectum, colon, prostate exam Medium High Unknown Medium Low Breast exam (mammography) Low Medium Medium Low Low Lung, stomach, esophagus cytology High Medium Low Low Low Oral exam Medium High Unknown Medium Medium a. As demonstrated, for example, from implenmentation in other countries. Source: Authors. Cancers 547 of the analytical findings in cancer planning. For example, in the stomach, a superficially spreading mass on the stomach China the Disease Surveillance Points System is providing mucosa, or an infiltrating process in the stomach wall. In its crucial mortality information for setting health priorities on a early stages stomach cancer usually causes no symptoms. With disease-specific basis. This information, supplemented with more advanced disease patients experience lack of appetite, cost data, will make it possible to analyze the cost-effectiveness weight loss, abdominal pain, and other nonspecific digestive of specific preventive and curative care interventions. It will symptoms. Diagnosis is usually made by barium x-ray studies also help to improve program efficiency by formulating cri- or gastroscopy. Within the United States more than 85 percent teria for prevention programs and triage rules for the use of of patients have advanced disease at the time of diagnosis, and curative care. only 10 to 15 percent survive five years. In-country capacity for research on questions of operational prevention programs needs to be developed. Particularly im- oCCCURRENCE. About 670,000 cases of stomach cancer occur portant are questions on the cost-effectiveness of health care in the world each year, the cases being about equally divided technology. Cancertreatments and diagnostic procedures, per- between the industrial and developing countries. Mortality haps more than those for other diseases, have led to adoption rates are particularly high (above 30 per 100,000 for males) in of high-cost technology with little established effectiveness. Japan, Chile, Hungary, and Poland. They are much lower in The contrast between the benefits of prevention and cure is United States whites (about 6 per 100,000), reflecting the acute for cancer. This contrast needs to be effectively and dramatic decline during this century in the occurrence of continually questioned using local analytical capacity. stomach cancer. Stomach cancer incidence among Japanese Prototypical models for longitudinal operational research immigrants to the United States falls in succeeding genera- programs to reduce mortality and morbidiry related to cancer tions. Incidence tends to be 1.5 to 2 times higher in men than and other chronic diseases need to be developed. Such pro- in women. There is a progressive increase in incidence with grams would encompass a surveillance system for chronic increasing age. diseases, an experimental framework for testing community- based interventions, and collaboration with the existing dis- ETIOLOGY. International variations and results of migrant ease prevention and health care system in testing new studiesstronglysupportaroleofnongeneticfactorsinstomach programs. Activities that might come under examination in- cancer etiology. The actual causative factors, however, remain clude alternative antismoking measures, early detection of uncertain. The most promising explanation is that the diet breast and cervical cancer, and a program of case management may contain both carcinogenic and anticarcinogenic sub- that employs home care and attempts to coordinate the capac- stances that influence stomach cells. Likely candidates for ity of different levels of care to provide cost-effective pallia- dietary carcinogens are (a) nitrosamines and (b) toxins pro- tion. Establishment of such models in several regions with duced by molds and bacteria in the course of food spoilage or diverse epidemiological and socioeconomic environments pickling. Highly salted foods also appear to increase risk. would make it possible to test alternative primary and second- Candidates for protective substances in the diet include vita- ary prevention programs in specific local conditions. min C and beta-carotene, a vitamin A precursor. Nitrosamines are present in preserved or smoked foods and can be formed in the stomach through metabolism of naturally occurring ni- Appendix 21A: The Ten Most Important Cancers trates and nitrites in food and water. Antioxidants, such as in the Developing World vitamin C, appear to inhibit nitrosamine formation. Although nitrosamines are potent carcinogens, their hypothesized role We consider each of the cancers individually, including a brief in the etiology of human stomach cancer remains unproven. description of its biological characteristics, pattern of occur- Infection with Helicobacterpylori is a cause of gastritis and may rence, important risk factors, effectiveness of treatment, possi- predispose individuals to the carcinogenic effects of dietary bility for screening, and potential for primary prevention. The nitrosamines. descriptions, of necessity, are brief. Our purpose is to provide a background to the discussion in the text. More complete TREATMENT. When detected early, stomach cancer may be information can be obtained from any of several references, treated by removal of all or part of the stomach. The great although Schottenfeld and Fraumeni 1982 is particularly rec- majority of patients are not candidates for this surgery either ommended. Tables 21A-1 and 21A-2 provide rates and per- because the tumor has spread beyond the stomach or because centages of cancer incidence, respectively. they are too ill to withstand a major operative procedure. Other forms of cancer therapy such as radiation or drugs are Stomach Cancer not curative and have little palliative effect. Stomachi cancer arises within the glandular cells that line the SCREENfNG. There is evidence, principally from Japan, that inside of the stomach. Almost all these tumors are adenocar- screening by barium x-ray studies, gastroscopy, or gastric cytol- cinomas. They may show varying degrees ofdifferentiation and ogy may detect stomach cancer at a stage when surgical cure may take several forms; a polyp projecting into the lumen of is more likely. None of these techniques has had a rigorous 548 Howard Barmum and E. Robert Greenberg Table 21IA-1. Estimated Crude Rates of Cancer by Site, Sex, and Region, 1980 (per 1 00,000 people) All developing All industrial Africa Latin America China Other Asia countries countries Site or type of cancer Male Femnale Male Femrale Male Femnale Male Femrale Male Female Male Femnale Mouth/pharvnx 6.2 3.7 8.7 3.3 6.1 4.4 15.2 8.3 10.5 5.9 14.7 4.4 Esophagus 3.0 1.1 5.8 2.2 21.0 12.3 4.9 3.5 9.6 5.6 7.3 2.9 Stomach 3.2 2.0 17.7 10.2 24.6 15.6 6.2 3.5 12.6 7.6 35.8 23.2 Colon/rectum 2.9 2.5 9.1 10.1 8.3 7.7 4.2 3.3 5.8 5.2 34.4 34.1 Liver 7.5 2.9 2.9 2.0 15.2 7.1 4.4 1.6 8.0 3.5 6.7 3.9 Lung 3.1 0.8 17.7 5.0 8.5 4.7 9.5 2.4 9.2 3.1 65.3 16.3 Breast n. a. 12.3 n.a. 30.8 n.a. 6.4 n.a. 15.0 n.a. 13.8 n.a. 59.2 Cervix n.a. 18.1 n.a. 27.0 nia. 27.4 n.a. 20.0 n.a. 22.7 n.a. 16.4 Lymphoma 8.8 4.9 6.3 4.6 2.2 1.7 4.4 2.3 4.6 2.7 11.4 9.1 Leukemia 2.3 2.0 4.7 4.0 4.7 3.4 3.1 2.1 3.6 2.7 8.3 6.3 Other 31.7 26.3 57.4 48.1 35.4 18.3 27.0 22.1 33.5 24.5 108.8 83.0 Total 68.6 76.7 130.4 147.3 126.0 109.0 78.9 84.2 97.3 97.4 292.6 258.9 n.a. Not applicable. Source; Parkin, Laara, and Muir 1988. scientific evaluation (through a controlled clinical trial), so ofspoiled foods or those preser-ved by pickling. Epidemiological their actual value remains uncertain. In countries with high data linking lowered risk to consumption of fresh fruits and incidence and mortality from stomach cancer, the effective- vegetables (particularly those containing large amounts of ness of screening programs may be worth investigating. Stom- vitamin C and beta-carotene) should be considered in discus- ach cancer does not occur with sufficient frequency in most sions of national food policy. For example, it seems prudent to populations, however, to justify mass screening activities. In avoid strategies that discourage production and consumption these populations more selective screening of high-risk indi- of these foods. viduals may be practical if any screening measure is eventually proven to be effective in preventing stomach cancer death. Esophageai Cancer PREVENTION. Knowledge of risk factors is insufficient to Esophageal cancer arises from the cells lining the inside of the recommend specific primary preventive strategies. In the in- esophagus. The tumors may be either squamous cell or adeno- dustrial countries, particularly in North America, time trends carcinoma in histological appearance. Usual presenting symp- in stomach cancer suggest a benefit from increased consump- toms are difficulty swallowing and steady chest pain. Diagnosis tion of fresh fruits and vegetables and decreased consumption is made by barium x-ray studies or endoscopy. At diagnosis Table 21A-2. New Cancer Incidence, by Site, Sex, and Region (percent) All developing All industTial Africa Latin Amrenica China Other Asia countries countries Site or type of- cancer Male Female Male Female- Male Female- Male Female- Male Female- Male Female- Mouth/pharvnx 9.0 4.8 6.7 2.2 4.8 4.0 19.2 9.8 10.7 6.0 5.0 1.7 Esophagus 4.4 1.4 4.4 1.5 16.7 11.3 6.2 4.2 9.9 5.8 2.5 1.1 Stomach 4.6 2.6 13.6 6.9 19.5 14.3 7.8 4.2 12.9 7.8 12.2 9.0 Colon/rectum 4.2 3.2 7.0 6.9 6.6 7.1 5.3 3.9 5.9 5.4 11.7 13.2 Liver 11.0 3.7 2.2 1.3 12.1 6.5 5.6 1.9 8.2 3.6 2.3 1.5 Lung 4.5 1.0 13.6 3.4 6.7 4.3 12.1 2.9 9.4 3.2 22.3 6.3 Breast n.a. 16.0 n.a. 20.9 n.a. 5.9 n.a. 17.8 n.a. 14.1 n.a. 22.9 Cervix n.a. 2 3.7 n.a. 18.3 n.a. 25.1 n.a. 23.8 n.a. 23.3 n.a. 6.3 Lymphoma 12.8 6.4 4.9 3.1 1.7 1.6 5.6 2.7 4.7 2.8 3.9 3.5 Leukemia 3.3 2.6 3.6 2. 7 3.7 3.1 3.9 2.5 3.7 2.8 2.8 2 .4 Other 46.1 34.3 44.0 32.6 28.1 16.8 34.2 26.2 34.4 25.1 37.2 32.1 n.a. Not applicable. Note. Figures reflect percentage of each cancer in the region. Source: Parkin, Laara, and Mluir 1988. Cancers 549 most esophageal cancers have spread into adjacent tissues. The Lung Cancer prognosis is dismal, and fewer than 5 percent of patients survive for five years. Cancers of the lung may arise at any site in the respiratory tree beyond the trachea, but most of these tumors occur in small OCCURRENCE. Esophageal cancer accounts for approximately airways within the lung itself. There are four main histological 310,000 new cases of cancer per year, four-fifths of which occur subtypes: adenocarcinoma, squamous cell, small cell undiffer- in the developing world. Esophageal cancer incidence and entiated, and large cell undifferentiated. The relative fre- mortality rates vary greatly in different parts of the world. In quency of these different histological types varies according to the areas of Iran and the former Soviet Union which surround sex and exposure to cigarette smoking. Among nonsmokers, the Caspian Sea, mortality rates of more than 100 per 100,000 adenocarcinoma is most common. In smokers, squamous cell per year pertain for both men and women. Within North tumors are most common, although women smokers also have America the rate is less than 5 per 100,000 per year. The a high frequency of small cell undifferentiated cancers. The occurrence of esophageal cancer increases progressively with principal importance of histological type is that small cell age. Rates tend to be about twice as high among men as women undifferentiated tumors behave differently from the others. in most parts of the world, although in some areas of high They are virtually always metastatic when diagnosed, and they incidence the male-female differences are less, whereas in are more responsive to combined chemotherapy and radiation. others they are much greater. In the United States, esophageal Initial symptoms of lung cancer are cough, bloody sputum, cancer rates are higher among blacks than whites, and there chest pain, or systemic symptoms of weight loss and fatigue. At appears to be a socioeconomic gradient, the disease being more least two-thirds of lung cancers have spread beyond the local frequent among the poor. Mortality rates for esophageal can- site by the time of diagnosis. These tumors progress both by cer, unlike those for stomach cancer, have not declined much regional extension and distant metastases to the brain and in recent years in industrial countries. other organs (this is particularly true for small cell undifferen- tiated tumors). About 90 percent of patients die within two ETIOLOGY. Tobacco and alcohol are the principal risk factors years of diagnosis. for esophageal cancer in industrial countries. The joint effect of these two substances in increasing risk appears to be greater OCCURRENCE. Lung cancer accounts for approximately than the sum of their individual effects. There is no clear 660,000 new cases of cancer per year; just over 200,000 of these explanation for the extremely high rates of esophageal cancer occur in developing countries. Countries with the highest in areas such as the Caspian littoral and parts of northern mortality rates (for males) include the United Kingdom (about China. Dietary characteristics have been implicated, but no 115 deaths per 100,000 per year) and other countries of north- particular component or habit has yet been proven to account ern Europe and North America (about 75 deaths per 100,000 for the highly elevated risk. Factors of interest include toxins per year). Rates are lower in countries in which smoking is less in pickled foods and fermented drinks, carcinogens from burn- common, such as Costa Rica (6 per 100,000) and Israel (25 per ing tobacco or opium, and deficiencies in nutrients such as 100,000). In most countries mortality rates for females are vitamin A, zinc, or selenium. approximately one-fifth to one-third those of males. Lung cancer incidence rates increase progressively with age except TREATMENT. Surgical resection offers the only chance of cure where cigarette smoking has been taken up more recently; in in localized disease, and surgery is also useful for relief of such countries relatively few older adults have smoked, so lung symptoms in more advanced cases. Radiation therapy may also cancer rates are lower among the elderly. Incidence and mor- relieve symptoms, particularly the inability to swallow. tality rates from lung cancer have increased dramatically in the past fifty years in many countries, and lung cancer has become SCREENING. The feasibility of screening for esophageal can- the largest cause of cancer death in the world as a whole. The cer has been considered in high-risk populations of northern increases parallel (with a delay of at least twenty years) the China. Thus far, screening by endoscopy or cytology has not uptake of tobacco smoking. been shown to reduce mortality. This topic merits further study in areas of extremely high risk. ETIOLOGY. The most important risk factor for lung cancer is cigarette smoking. Numerous studies have shown that smokers PREVENTION. In countries in which there is relatively low risk have ten to fifteen times the risk of lung cancer as nonsmokers. of esophageal cancer, the most reasonable primary preventive Among heavy smokers the risk is approximately twenty-five strategies are avoidance of tobacco and reduced consumption times that of nonsmokers. Risk is most closely related to the of alcoholic beverages. In countries with extraordinarily high number of years a person has smoked. Filtered and low-tar rates of esophageal cancer, the actual risk factors have yet to cigarettes appear to be less risky than other types with regard be identified, so there is little information on which to base a to causing lung cancer (although they arc not clearly less primary prevention program. Studies of dietary supplementa- hazardous with respect to cardiovascular disease). In certain tion with vitamins now being conducted in northern China limited populations other risk factors are also important in lung may provide a better scientific basis for intervention. cancer etiology. These include asbestos (related to mining and 550 Houward Barnum and E. Robert Greenhbrg industrial exposure), radon (principally related to under- histological category. These tumors develop within the liver ground mining but perhaps related to household exposures), andcauselocalizeddestructionandenlargementofthatorgan. and industrial exposure to substances such as bischloro- Presenting symptoms are abdominal mass, pain, and weight methylether (BCME), nickel, and chromates. Dietary and sermm loss. The tumor tends to progress rapidly by local growth and studies show a decreased risk in persons eating large amounts extension. Distant metastases may occur but usually are not an of foods containing beta-carotene and in those with higher important feature. Fewer than one person in twenty survives blood levels of this substance. The importance of these findings five years following diagnosis. is uncertain. OCCURRENCE. Primary liver cancer accounts for 250,000 new TREATMENT. Surgical resection of lung cancer offers the only cases of cancer per year; more than 190,000 of these occur in real chance of cure. Fewer than one-third of patients are developing countries. Areas of highest incidence and mortality candidates for this surgery, either because the tumor is no are in Sub-Saharan Africa and southern Asia. Rates tend to be longer localized or because the patient is too unwell to with- lower in North America and western Europe. In Hong Kong, stand a major operation. Radiation therapy can palliate symp- mortality rates for males are roughly 39 per 100,000; in the toms of lung cancer and, in some patients, may extend survival. United States, they are less than 2 per 100,000. Mortality rates A small fraction of patients with small cell undifferentiated for females are generally one-half to one-quarter of the male tumors appear to have prolonged survival following intensive rates. Substantial differences in rates occur for different ethnic radiation or chemotherapy. groups within countries; for example, in the United States, ethnic Chinese have liver cancer rates approximately five to SCREENING. Various methods of early detection of lung can- eight times those for whites. In countries of high incidence, cer have been tested, but none has been shown to be effective. liver cancer rates appear to plateau in middle adulthood, Methods tried include periodic chest radiographs and exami- whereas in areas of low incidence they continue to rise pro- nation of sputum cytology. The principal problem with screen- gressively with age. ing is that lung cancer progresses rapidly from the time when it is first detectable by screening methods to when it is no ETIOLOGY. There is compelling evidence from epidemiolog- longer curable by surgery. Screening programs designed to ical and laboratory studies that hepatitis B virus can cause increase this "window of opportunity" are under consideration hepatocellular carcinoma. Liver cancer is most common in and may eventually prove beneficial in selected groups of very countries with high rates of HBV infection; the vast majority of high risk individuals. patients with HCC have serological evidence of HBV infection, particularly the carrier state; and laboratory studies have PREVENTION. The principal preventive efforts are discourag- shown viral DNA within hepatocellular carcinoma cells. In ing uptake ofcigarette smoking and encouraging its discontin- areas with high liver cancer incidence and mortality, HBV uance among those who already smoke. There is evidence that infection occurs early in life, probably during the perinatal adult and teenage education can decrease use of cigarettes. period, and is spread from carrier mothers to their children. More effective programs, however, will likely be aimed at Other likely etiological factors include alcohol consumption restricting cigarette marketing (through prohibition of adver- and aflatoxin. In areas of low incidence, alcohol use and tising) or sales (by heavy taxation). Efforts to portray cigarette cirrhosis are usually (though not always) identified as risk smoking as antisocial behavior also may be effective in groups factors in epidemiological studies; but in high incidence areas, which are responsive to these social pressures. Reduction of alcohol is probably of minimal importance. In tropical Africa occupational exposures to asbestos and other lung carcinogens and Asia, food preservation is difficult, and growth of a mold is also warranted. (Aspergillus flavus) can contaminate foods with aflaroxins, which are strongly carcinogenic in laboratory animals. Epide- Liver Cancer miological data support a higher risk of liver cancer in persons who consume relatively larger amounts of foods (particularly Most primary liver cancers are hepatocellular carcinomas peanuts and grains) likely to be contaminated with aflatoxins. (HCCs) and arise from hepatic parenchymal cells. Another This factor appears to act jointly with hepatitis B virus in important cell type is cholangiocarcinoma, which arises from explaining much of the worldwide distribution of primary liver the bile ducts. It accounts for fewer than 25 percent of all cancer. Other exposures (such as smoking) and hereditary primary liver cancers in low-incidence areas and probably disorders (such as hemochromatosis) are also associated with fewer than 10 percent in areas with high primary liver cancer the risk of liver cancer, but these factors could only account incidence. Risk for cholangiocarcinoma appears to be in- for a small proportion of primary liver cancers in the develop- creased in persons infected with Clonorchis and Opisthorchis, ing countries. which are liver fluke infections prevalent in Southeast Asia and are acquired by eating uncooked fish. Within the devel- TREATMENT. If detected very early, liver cancers can be oping world the great majority of liver cancers are HCCs, and surgically resected, although the procedure is a lengthy one and the remainder of this discussion primarily focuses on that requires great surgical skill. Some patients have been treated Cancers 551 successfully with total resection of the liver and transplanta- include Colombia (50 per 100,000) and Brazil (40 per tion. Other treatment modalities, such as radiation and che- 100,000); very low incidence is found among Israeli Jews (5 motherapy, are of little or no benefit for this condition. per 100,000). There is a strong socioeconomic gradient, the poorer countries and poorer groups of women within countries SCREENING. Hepatocellular carcinomas produce alpha- having the highest risk. Peak incidence of invasive cervical fetoprotein, which can be detected by assay of peripheral cancer occurs in late adulthood, and lower rates are observed blood. There are reports from uncontrolled studies in very high in the elderly. In North America, cervical cancer mortality risk groups that alpha-fetoprotein screening detects asymp- rates have declined dramatically during the past fifty years. tomiatic personls who may be cured by surgical resection. The This decline began before institution of Pap testing. Neverthe- screening procedure has not, however, been the focus of any less, there is strong evidence from studies in the United States, rigorous scientific study. Canada, and northern Europe that widespread use of the Pap test has contributed to the decline in mortality from this PREVENTION. In high-risk areas, preventive efforts are di- condition. In some countries of westem Europe there has been rected toward neonatal immunization with HBV vaccine. The a recent reversal of the trend toward falling rates, and cervical effectiveness of this vaccine in preventing liver cancer has not cancer appears to be increasing among younger women. been established, although current studies in China and the Gambia should eventually provide a better basis for assessing ETIOLOGY. Cervical cancer has many features of a sexually its value. Because of the strong evidence linking YIBV infection transmitted disease. Risks are highest anmong women who have to primary liver cancer, and because of the apparent safety of had multiple sexual partners (or whose husbands have had the vaccine, widespread programs of immunLization are a pru- multiple partners), who began sexual intercourse early in life, dent course of action until vaccine field trials have been and who have sexual partners of lower socioeconomic status. completed. Other preventive efforts should focus on maintain- Barrier methods of contraception appear protective; oral con- ing food stores, such as peanuts, under conditions that mini- traceptives increase risk moderately. Intensive efforts to iden- mize growth of molds. Testing commercial food stores for tify an infectious cause of cervical cancer initially focused on aflatoxin is alsoareasonable strategy. In areas endemic for liver herpes simplex virus 2 (Hsv2). Indeed, cervical cancer risk is fluke infection, control of these parasites may reduce stronglyassociated witheevidenceofHsv2infection,andlabo- cholangiocarcinoma occurrence. ratory studies have found nisv2 genetic material in cervical cancer cells. Nevertheless, more recent studies strongly suggest Cervical Cancer an etiological role for human papilloma virus (HPV), particu- larly types 16 and 18. This virus is more strongly associated Cervical cancer arises in cells covering the lower, vaginal with cervical cancer risk than is Hsv2. Human papilloma virus portion of the uterus. Progression from normal cells to cancer- causes benign growths (genital warts), and HPV DNA has been ous ones appears to occur through phases of dysplasia, carci- found intercalated in the human DNA of cervical cancer cells. noma in situ, and, finally, invasive cancer. Some terminology Another risk factor is cigarette smoking. In most studies smok- breaks down the preinvasive phases into three categories: ers have cervical cancer risks 1.5 to 2 times higher than cervical intraepithelial neoplasia 1,11, and 111. In most women, nonsmokers. Although this association might be noncausal the transformation from normal cells to invasive carcinoma and due to the relationship between smoking and early sexual occurs over a long period (probably fifteen to thirty years), so activity, the weight of evidence appears to favor a causal role there is long time when the condition can be detected at an for smoking in cervical cancer. early, noninvasive stage. In some women, however, the process appears to progress much more rapidly. Once cervical cancer TREATMENT. Treatment of cervical neoplasia varies accord- becomes invasive it tends to spread by direct extension to the ing to stage. For noninvasive lesions, initial destruction of the adjacent tissues of the pelvis and by metastases to distant sites affected tissue by incision or laser and periodic follow-up may such as the lung and liver. In the United States approximately be all that is necessary to achieve a cure. As an alternative, 50 percent of invasive cervical cancers are still localized when hysterectomy removes the affected tissue and obviates any detected. Approximately 80 percent of patientssurvive forfive need for further surveillance. In more advanced disease the years if tumors are still localized and approximately 60 percent treatment options depend on the extent of tumor spread and survive for five years through all stages. Most cervical cancers the preferences of the patient. Extensive pelvic surgery and are found by screening examinations; however, symptoms of radiation are capable of curing disease which has spread be- vaginal bleeding and pelvic pain occur if the disease progresses yond the cervix, although success in more advanced stages is beyond the initial stage. less likely. OCCURRENCE.Cancerofthecervix accountsfor470,000new SCREENING. Pap testing involves cytological examination of cases of cancer per year in the world. Approximately 80 per- cells scraped from the cervix and is the established screening cent of these occur in developing countries, where they cause method for cervical cancer. Properly performed, the Pap test is about 155,000 deaths annually. Areas of highest incidence capable of detecting the vast majority of cervical neoplasms 552 Howard Barnum and E. Robert Greenberg before they become invasive. Positive tests require follow-up the Orient. Traditionally prepared Chinese salted fish con- by repeat testing, biopsy, and, perhaps, culdoscopic examina- tains carcinogens (perhaps volatile nitrosamines) which can tion to establish the diagnosis. Although annual Pap tests are produce nasopharyngeal tumors in rats. Volatile nitrosamines recommended for high-risk women, less frequent testing (even exist in other traditionally prepared foods in endemic areas, as rarely as every five years) should be highly effective in and further study of this issue may lead to an effective preven- reducing mortality. The principal obstacle in Pap test programs tive strategy. is that the women at highest risk are hard to reach because they are poor and avoid doctors. Failure to follow up positive or TREATMENT. Treatment of nasopharyngeal cancer involves suspicious tests and unreliable cytology readings have also been high-dose radiation therapy. The tumor is generally respon- problems in some programs. In most settings, highest priority sive. Most patients achieve palliation and approximately one- should be assigned to achieving widespread acceptance of the quarter are cured. Pap test program and to providing adequate definitive care. More frequent testing of already screened women is usually less SCREENING. There are no proven effective screening mech- important. anisms for nasopharyngeal cancer. PREVENTION. No primary prevention programs have proven PREVENTION. Primary prevention methods are speculative. effective. The fact that cervical cancer is amenable to early Development of an Epstein-Barr virus vaccine has been pro- detection and the lack of certainty about its etiology have posed as a possible preventive method, but none has yet been directed public health efforts to screening. Some authors pro- developed, and the value of this method is uncertain. Changes pose development and implementation of HPV vaccines, but in infant feeding practices (should the initial findings regard- these suggestions (like earlier suggestions for an HSV2 vaccine) ing salted fish and other foods be confirmed) appear to offer appear to be premature. more promise for prevention. Nasopharyngeal Cancer Oral and Other Pharyngeal Cancers Cancers of the oral cavity and pharynx fall into two relatively Other cancers of the oral cavity and pharynx principally arise separate epidemiological categories. The first is nasopharyn- in the lining of the mouth and throat. Most of these tumors geal cancer, which occurs principally among the populations have a squamous cell histology. They typically present either of southern China and Southeast Asia. The second category, as an oral ulcer or as enlarged lymph nodes in the neck, and discussed separately below, includes other tumors of the mouth they progress through local and regional extension rather than and pharynx, which occur with greatest frequency in people distant metastases. The usual outcome varies according to who habitually chew tobacco, betel, or similar substances. where the tumor arises within the mouth or pharynx. Cancers Nasopharyngeal cancer arises in epithelial cells lining the of the lip are generally found early and are cured by surgery. surface of the nasopharynx. Usual presenting symptoms are Those of the tongue or pharynx are approximately 70 percent lymph nodes in the neck; nasal bleeding and obstruction; fatal within five years. symptoms of nerve compression, including headache; and ear- ache or hearing loss. The five-year survival rate after treatment OCCURRENCE. Worldwide, cancers of the mouth and phar- (radiotherapy) is approximately 22 to 35 percent. ynx account for more than 300,000 new cases of cancer per year, about two-thirds of which occur in developing econo- OCKCURRENCE. Although rare in most of the world, nasopha- mies. Incidence and mortality rates are highest in India, Hong ryngeal cancer is the most common cause of cancer death in Kong, Puerto Rico, Brazil, France, and Singapore. In most parts some southem Chinese populations. Rates are high in areas of the world, incidence rates are many times higher for males with large groups of Chinese origin, as in Singapore and than for females. Rates increase progressively with age, and Malaysia. Nasopharyngeal cancer occurs with peak frequency there is a strong socioeconomic gradient, the poorer people among adults between the ages of thirty-five and sixty-five, but having higher rates. it also occurs relatively often in children and adolescents. Risk is somewhat higher in males than in females. The condition is ETIOLOGY. The principal risk factor for oral and pharyngeal associated with lower socioeconomic class in China and per- cancers is tobacco smoking and chewing. In countries such as haps in other countries. India, tobacco is often mixed with betel leaves, lime, and other substances. Chewing betel, or areca, nut may be a risk factor ETIOLOGY. The principal identified risk factors for nasopha- even in the absence of tobacco use. Cigarette smoking (espe- ryngeal cancer are Epstein-Barr virus infection and the con- cially reverse smoking) is also associated with risk. Other risk sumption during infancy of traditionally prepared Chinese factors include alcohol consumption and lower intake of fruits salted fish. The role of Epstein-Barr virus in this condition is and vegetables (which contain beta-carotene as well as other unclear, because Epstein-Barr virus infection occurs through- possible anticarcinogens). Cancers of the lip are strongly re- out the world but nasopharyngeal cancer is seen principally in lated to sunlight exposure. Cancers 553 TREATMENT. Treatment of these tumors principally involves ETIOIOGY. The causes of large bowel cancer are unknown. surgery and radiation therapy. Surgery can be curative if tumors International comparisons and migrant studies implicate are found early, but it often results in substantial loss of func- dietary habits in the etiology of this disease. Dietary fat tion and requires intense rehabilitative efforts. Radiation ther- consumption is a strong candidate for a causal role, but the epi- apy is used in conjunction with surgery and may contribute to demiological data are inconclusive. Relative deficiency of survival. Radiation also may palliate patients with nonresect- vegetable fiber is also implicated in large bowel cancer etiol- able tumors. ogy, particularly in persons with a high dietary fat intake. The risk of large bowel cancer is lower in people who consume large SCREENING. Although no screening programs have been amounts of vegetables and fruit. Possible protective substances proven to be effective, they appear likely to be beneficial. from these sources include vitamins C and E, beta-carotene, Examination of the mouth and pharynx of high-risk persons indoles, and the like. Adenomatous polyps and bowel cancers (perhaps linked to dental examinations) can detect many also appear to be at least partly determined genetically. tumors before they have advanced to an untreatable stage. Oral examination also detects leukoplakia (white patches), a pre- TREATMENT. Surgical resection of tumor and adjacent large malignant condition which requires close observation to mon- bowel is curative if the cancer is detected early. For rectal itor development of invasive cancer. cancer, radiation therapy may also contribute to the chance of cure and palliation. Chemotherapy of advanced lesions does PREVENTION. Primary prevention should focus on reduced not prolong survival materially, although some patients are use of tobacco, either for chewing or smoking. Currently improved by this treatment. Patients with rectal cancer often researchers are alsco investigating whether dietary supplemen- require a colostomy and need postoperative rehabilitation. tation, particularly with beta-carotene, can prevent oral and pharyngeal cancers. This notion remains to be proven. SCREENING. Although several screening practices have been broadly recommended for early detection of large bowel can- Large Bowel Cancer cer, none has been conclusively shown to reduce mortality from this condition. Screening methods fall into three catego- Cancers of the large bowel usually arise from the glandular ries: stool occult blood testing, digital rectal examination, and epithelium and are classified histologically as adenocarcino- sigmoidoscopic examination. Occult blood screening involves mas. Most of these tumors appear to result from malignant testing a small specimen of stool for the presence of hemoglo- transformation of benign adenomatous polyps. Tumors may bin. The test is easy to perform, inexpensive, and generally arise anywhere in the large bowel; in the United States about acceptable to patients. The principal problems with the test three-quarters occur in the colon and one-quarter in the rec- are that it results in a high number of false-positive findings, tum. Although rectal and colonic cancers differ somewhat in which require subsequent and (often costly) definitive diag- their epidemiology and clinical behavior, they are considered nostic studies, and it has a relatively low sensitivity in the together in this discussion. Large bowel cancers typically pres- detection of small polyps and very early cancers (because these entasrectalbleeding,anemia,orsignsofpartialbowelobsttuc- generally do not bleed). Large clinical trials of occult blood tion. They tend to metastasize to regional lymph nodes and the screening, despite involving tens of thousands of patients, have liver, and about two-thirds have spread beyond the local stage thus far failed to show a statistically significant reduction in by the time of diagnosis. About 50 percent of patients survive large bowel cancer mortality associated with use of the test. It five years. appears, therefore, that even if occult blood testing is effective, the magnitude of benefit is not large. A second method of OCCURRENCE. Large bowel cancers cause more than 570,000 screening is finger examination of the distal rectum; this can new cases of cancer in the world per year; about 30 percent of be performed at the time of a complete physical examination. these occur in developing countries. Areas of greatest occur- Advantages of this screening method are simplicity and low rence are the more westernized and industrialized countries. cost. The primary disadvantage is that only a small proportion Mortality rates for males in New Zealand, Denmark, England, of large bowel cancers are potentially detectable in this way. and Hungary range between 20 and 25 per 100,000. Typical The third screening method is visual examination of the areas of intermediate mortality are Greece (6 per O00,000) and rectum and sigmoid colon; this is best accomplished with a Spain (I I per 100,000). The lowest rates are reported from less tlexible sigmoidoscope. The procedure reliably detects early industrialized countries, such as Costa Rica (5 per 100,000) cancers and adenomatous polyps in the distal large bowel and Nigeria (probably below 2 per 100,000). Within areas of (usually the last 60 centimeters). Advantages of this procedure high incidence there are often populations with lower inci- are that it appears to identify as many as 80 percent of persons dence. For example, in New Zealand, incidence rates in the who harbor polyps (although identification of all polyps in Maori are approximately one-third those of the non-Maori. these people will require examination of the remainder of the Incidence and mortality rates in women are generally about 20 large bowel by colonoscopy or contrast x-rays). Because polyp percent lower than those in men. Risk increases progressively formers are the group at greatest risk of later developing bowel with age. cancer, they should be kept under surveillance after being 554 Howard Bamum and E. Robert Greenberg identified by sigmoidoscopic examination. Disadvantages of delayed pregnancy into their fourth decade. In most studies, the procedure are that it requires a skilled examiner (usually a earlier menarche and later menopause are associated with physician. although trained nurses and paramedical personnel higher risk. A positive family history of breast cancer is also a have been used successfully), the equipment is expensive, and risk factor. Increased consumption of dietary fat and calories is the procedure entails more time and discomfort than other implicated in breast cancer etiology, but the epidemiological screening methods. Although flexible sigmoidoscopic screen- data are inconclusive. In general, women who are more obese ing has not yet been tested in a controlled clinical trial, on have higher death rates from breast cancer, although it is not present evidence it appears to offer the best prospect for clear whether this is due to a greater risk of developing the reducing colon cancer mortality. disease or a greater risk of dying once it has occurred. Alcohol consumption also is implicated as a risk factor, particularly for PREVENTION. Uncertainty about the cause of large bowel high levels of consumption. cancer has largely prevented implementation of primary pre- vention programs. Dietary changes that include reduced fat TREATMENT. Surgical resection of the tumor is the standard and increased fiber intake have been recommended on the initial therapy for breast cancer. Surgerv may remove only the basis of epidem iological and laboratory data. The value of these tumor or, more commonly, the entire breast and associated interventions is largely a matter of conjecture, but it would be axillary tymph nodes. Radiation therapy is often given to the prudent for developing countries to resist dramatic changes breast and adjacent lymph nodes of women treated with simple away from their current high-fiber and low-fat diets. excision. For an increasing proportion of women, primary treatment of breast cancer has come to include chemotherapy Breast Cancer and hormonal therapy with Tamoxifen. There is strong evi- dence that these additional therapies lessen recurrence follow- The great majority (80 to 90 percent) of breast cancers arise ing surgery. Metastatic disease is treated with radiation or from epithelial ductal cells; perhaps 10 percent are of lobular chemotherapy and hormonal therapy to reduce local symp- origin. The disease usuallv presents as a nodule or mass in the toms and to induce remission. Many women with metastatic breast. Metastatic spread is common and may affect sites breast cancer respond well to treatment and have prolonged througholut the body. Involvement of the axillary lymph nodes periods of normal function. In both industrial and developing is a usual feature of larger tumors. Other common sites of countries Tamoxifen offers safe, inexpensive, and effective metastatic spread are the bones, liver, lung, and brain. Most treatment. breast cancers progress relatively slowly, and prolonged sur- vival (several years) with active disease is common. After SCREENING. There is conclusive evidence from randomized initial treatment the patients often follow a path of remission clinical trials that screening programs, consisting of mammog- and relapse. In the United States approximately 70 percent of raphy and physician examination of the breast, reduce breast breast cancer patients survive five years. An important feature cancer mortality by about 30 percent. The value of mammog- of breast cancer, unlike most other tumors, is that recurrence raphy alone is less certain. Hazards of screening are a small risk is common after a prolonged survival free of disease. of cancer from radiation exposure and a risk of needless surgery generated by false-positive results. Of all cancers, however, OC.CURRENCE. There are approximately 575,000 new cases of evidence is strongest for a beneficial effect of screening in breast cancer in the world each year; about 40 percent occur breast cancer. in developing countries. Areas of highest incidence include North America and western Europe (between) 50 and 80 per PREVENTION. Epidemiological information on breast cancer 100,000), whereas incidence is lowest in Japan (about 20 per is insufficient to support institution of any primary prevention 100,000). Breast cancer incidence rates have been increasing program. The clearest risk factors relate to reproductive prac- innearlyallcountriesthathavereliabledata.Theproportional tices which are not readily amenable to preventive interven- increase has beencgreatest in areas of previously low incidence, tion. In fact, policies in developing countries are likely to such as Japan. Some of the increase in incidence in the United emphasize reduced parity and delayed childbearing as ways to States, and perhaps elsewhere, is due to better detection of reduce population growth. These policies, if implemented tumors. Breast cancer mortality rates in the United States have successfully, are likely to increase breast cancer occurrence. been relatively constant in the past fifty years. Mortality and Although several groups have recommended reduced dietary incidenice rates increase progressively with age until the time fat intake to control breast cancer, the epidemiological data of menopause, after which point the rate of increase is less. do not consistently support this course of action. Most efforts at breast cancer control are now correctly directed toward ETIOLOGY. Ionizing radiation is the only exogenous exposure earlier detection and improved treatment. that has been clearly shown to cause breast cancer. Neverthe- less, several dietary and reproductive practices are related to Lymphomna breast cancer occurrence. Risk is increased in nulliparotis women, and multiparity is associated with lower risk. Among These tumors arise from cells of the immune system, including women who have been pregnant, risk is highest in those who lymphocytes, histiocytes, and their precursor cells. Lympho- Cancers 555 mas are a heterogeneous group of neoplasms which differ also responds dramatically to chemotherapy, and more than 90 greatly in their epidemiology and clinical course. The reasons percent of patients achieve complete remission of their tumors. for including them in this chapter are (a) the Burkitt's lym- Perhaps 50 percent or more of them can be cured with cur- phoma type occurs with great frequency in young children in rently available therapy. parts of East Africa, (b) the Hodgkin's type tends to occur in childrenand inyoungadultsduringtheirproductiveyears, and SCREENING. There are no practical screening programs for (c) both varieties are often cured by intensive therapy. Lym- lymphoma. phoma usually presents as enlarged lymph nodes. Concomitant features include fever, weakness, loss of appetite. itching, and PREVENTION. Programs to control or eradicate malaria con- other symnptoms of general illness. As it progresses it involves ceivably will decrease Burkitt's lymphoma. There is no good multiple sites in the body. Accordingly, lymphoma can pro- evidence, to date, that this occurs, however. Development of duce a variety of complicating problems and usually causes an Epstein-Barr virus vaccine has also been suggested as a death unless quickly treated. preventive strategy for Burkitt's lymphoma. If an effective Epstein-Barr virus vaccine is developed, it logically should be OCCURRENCE. Lymphomas, considered as a group, account tested in controlled trials for effectiveness in preventing for approximately 135.000 new cases of cancer per year in the Burkitt's lymphoma. world; just over half of these occur in developing countries. Occurrence of different varieties of lymphoma varies from Leukemia country to country. Burkitt's lymphoma occurs with highest frequency in parts of East Africa and New Guinea, where Leukemias are malignancies of the blood-forming cells. They malaria is endemic. Intemational pattems of Hodgkin's dis- are generally categorized according to the cell of origin and to ease vary according to age group. Childhood Hodgkin's disease whether the disease is chronic or acute in onset. The clinical occurs most often in developing countries, whereas the dis- behavior and treatment requiretments differ for the various ease in young adults is more common in niorthern Europe and types of leukemia. We consider all leukemias together here North America. Both Hodgkin's disease and Burkitt's lym- because they often respond to treatment, and because the phoma affect males more often than females. technological requirements for their treatment are similar. Leukemias are characterized by an overproduction of either ETIOLOGY. Burkitt's lymphoma is strongiv associated with mature (chronic leukemias) or immature (acute leukemias) Epstein-Barr virus infection in epidemiological and labora- bone marrow cells. Consequences of leukemia relate to defi- tory studies. The exact role of Epstein-Barr virus in causing ciencies of normally functioning red cells, white cells, and Burkitt's Iymphoma is uncertain, however. Areas with highest platelets (necessary for control of bleeding). Accordingly, pre- rates of Burkitt's lymphoma also have endemic malaria, and senting symptoms of leukemia are anemia, loss of resistance to this geographic relationship has not beeri filly explained. Why infection, or bruising and bleeding. Untreated, acute leuke- should Epstein-Barr virus infection he associated with Burkitt's mias are rapidly fatal. Chronic leukemias may persist for years lymphoma in Africa and with nasophiaryngeal cancer in without causing dehilitating symptoms. Alternatively, chronic China? Also, Epstein-Barr virus is a ubiquitous virus (it is the myelogenous leukemia may enter an accelerated phase which cause of infectious monlonucleosis). What circumstances of mimnics an acute leukemia and is rapidly fatal. infection or other factors are necessary for Burkitt's lymphoma, and how can they explain the dramatic geographic variations? OCCURRENCE. Leukemias account for approximately These questions remain to be answered. 190,000 new cases of cancer per year worldwide. About The cause of Hodgkin's disease is unknown, although some 105,000 of these cases occur in developing countries. Leuke- epidemniological data support an infectious etiology. The pat- mia incidence rates peak first in early childhood, decline, and tem of occurrence is in many ways co-nsistent with the late then progressively rise again with age. Much of the childhood effects of an early childhood infection. Thus far, however, no peak in leukemia is due to acute lymphocytic cell type, which infectious agent has been identified. The causes of lymphornas is also highly responsive to intensive therapy. The adult leu- other than Burkitt's lymphoma and Hodgkin's disease also are kemias are more often of the acute or chronic myelogenous unknown. Patients on imnmunosuppressive therapy (for exam- varieties. International variations in leukemia incidence and ple, following organ transplantation) are at higher risk of mortality are not as pronounced as those for most other can- lymphomas in the brain, but this is of relatively little import- cers, except that chronic lymphocytic leukemia is rarely seen ance to developing countries. in the Far East. Leukemia occurs more often in males than in females and in whites than in blacks or Asians. TREATMENT. Both Hodgkin's disease and Burkitt's lym- phoma are highly responsive to therapy with radiation and ETIOLOGY. There is clear evidence from atomic bomnb survi- chemotherapy. Many patients can be cured of Hodgkin's dis- vors in Japan and from other exposed groups that radiation ease even if they have advanced disease. In early disease, cure causes acute leukemia. Increased leukemia rates are detectable mav be achieved by radiation alone. More advanced disease within three years following acute radiation exposure, and the requires iitensive chemotherapy. African Burkitt's lymphoma excess persists for decades. The increased risk is principally for 556 Howard Barnum and E. Robert Greenberg the myelocytic types. Other environmental exposures, includ- in most oftihe developing world, however. that they nearly o-mpensate fb r sex ing exposure to chemicals (particularly benzene), have been vdifferences in other cancers and the overall rate of canicer in women is nearly ingexplicted as causesiof leukeia.t Alylatng drug sc a equal ito that of men. In tables 21A-1 and 21A-2 we give rates tor men and iinplicated as causes,of leukeinia. Alkylating drugs sucb as women by main developing country region. Also refer to Parkin, Laara, and melphalan and bUsulfan, which are used in cancer chernother- Mluir ()1988) for an excellent disctission of detailed rates. apy, also can cause leukemiiia. For rhe vast majority of cases, 2. Without anv adjustment for price or cultural differences, an estimated however, there is no history of exposure to a known leukemo- relationship between 1982 manufactutred cigarette consumption and income gen other than nortnal background levels of radiation. hased on 3 cross-section sample of eightv-four countries is cag/pop = 361 + . 25 NNP/pop -0.000009 (NNP/pop) R = .56 TREATMENT. Treatment of acute leukemias, particularly (4 4) (5.9) (3,4) acute lymphocytic leukemia of childhood, has advanced dra- matically in the past three decades. Intensive chemotherapy At the mean pet capita income ($3,500) for the sample the estimated income of acute lymphocytic leukemia appears capable of curing 50 elasticity is 0.7. At incomes below $1,000 the income elasticity of cigarette percent of affected children. Treatment requires intensive consumption is above 1.3. The t-statistics are in parentheses (World Bank 1988; 1ARs. 1986). medical support with close monitoring of chemotherapy side 3. The puLrpose ot using the percentage of per capita w:,r or %CtNPN, rather effects. Infectious complications are common. After an initial than monetary units is to reduce measures of program costs across countries to phase of induction therapy, children require prolonged periods approximately comparable units. The measure is deficient in that it primarily of maintenance with intermittent chemotherapy. Adults with adjusts for lahor cost differences between countries but does not account well acute mye,ogenous leukemia require even iiiore intensive ther ortot differences in toreign supply costs or in productivity. The deficiencies are offset, however, bv the convenience of the measure. apy, but there is now the possibility of curing somne of these 4 The effect of increasing tobacco prices, for example. is greater in reducing patients. The chemotherapy regimens are intensely toxic to co,nsumptionbvtheyoutingthaninreducingconsumptionamongthectirrently thenormalbone marrowandotherorgans,andpatientsrequire addicted. Similarly, the costs if programs to convert present smokers to close monitoring and support to treat complications of infec- nionsmioking status can he high (Altman and others 1987). tion, hieding, nd anemi. Newertechniqus ofbon marrow 5. After correcting nomiinal rates of interest for inflation, the real rate of tion, bleeding, and anemia. Newer techniques of bone marrow interest has been in the vicinity of 3 percent in much of the wotld during the transplantation, using either a closely matched donor or last twenty years. treated rnarrow cells from the patient, offer prospects for curing 6. Under exceptional circumstances and at very high cost, screerning proce- more leukemia patients. This therapy requires great techno- diires appear to detect esophageal, stomach, or liver cancer at a stage at which logical sophistication and intense supportive care by medical early sirgery i ay be suiccessful. Research in Japan is under wav oin the and nursing personnel. practicality of endoscopy programs in areas of high stomach and esophageal cancer incidence. 7. Relative cancer survival is the survival probabilits for cancer patients in SCREENIN(G. Acute leukemias progress very rapidly from the relation to that expected for persons of similar age in the general population. time at which they might be detectable by screening to the 8. Assuming an outpatient visit every foUrteen days at $1.50 per visit, the time of svmptoms. There is no known value in treating chronic cost of aspirin or paracetamol is $0.10 per day, the cost of codeine is $1.00 per day and the cost ofmorphmne is $0.20 per day. For severe symptoms it is assumed leukemias before they are symptomatic. For these two reasons, that there wouldbeforty-fivedaysduring whichcodeine and paracetamol were screening has no role in the management of these conditions. required and forty-five days during which morphine was needed. PREVENTION. Avoiding unnecessary exposure to medical ra- diation is probably the mnost feasible approach to letukemia References prevention. Other strategies include reduced occupational exposure to known leukemogens (such as benzene), and avoid- Altman, D. G.,J. A. Flora, S. P. ForTtman, andJ. W. Farquhar. 1987. "The Cost ing environmental contamination with these substances. Effectiveness ofThreeSmokingCessation Programs." Americanijournal of These preventive strategies, even if effectively implemented, Public Health 77(2) 162-65. are unlikely to produce a measurable decrease in leukemia American Cancer Society. 1987. "Cancer: Basic Data." In Cancer Facts anid | . ,. , , , , ~~~~~~~~~Figures- 1987. American Cancer Societv. Attanta, Ga. occurrence, because few cases are directly related to these Fm rs-97AeicnCcrSoet.tltaGa Axtell, L. M., A. J. Asire, and M. H. Myers. 1976. Cancer Patient Survival. factors. Repon 5. U.S. Department of Health, Education, and Welfare; National Cancer Institute, Washington, D.C. Bailar, John, ill, and Elaine M. Smith. 1986. "Progress against Cancer?" New Notes England journal of Medicine 314:1226-32. Baines C. J., A. B. Miller, and A. A. Bassett. 1989. "Physical Examination: Its We are indebted to Richardf Doll, Kenneth Stanley, andjan Stjernsward for Role as a Single Screening Modality in the Canadian National Breast their carefll comments tin earlier drafts. We have also benefited from disclis- Screening Study." Cancer63:1816-22. sions of this material at the World Health Organization and the United States Barnum, HI. 1988. "Economic Issues in Planning the Use of Health Resources Centers for Disease Control. The authors remain responsible for errors and for Chronic Diseases in China." Background Doctiment for World Bank conclusions. In particuilar, the conclusions de) not necessarilv reflect the views Sector Review, Population, Health, and Nutrition Department. Washing- of the World Bank. ton, D.C. 1. The use of aggregate nunbers may obscTre important differences between Beasley, R. Palmer. 1988. "Hepatitis B Immunization Strategies." men and women. For most cancers, site specific incidence for men is ahout WHo/Epli/EN/88.5. World Health Organization/Expanded Programme on double that for women. Rates of cervical and breast cancer are sLfficiently high Immunization, Geneva. Cancers 557 Beasley, R. Palmer, Chia-Chin Lin. Lu-Yu Hwang, and Chia-Siang Chien. Altman, E. H. Braff, P. O'Malley, D. Hawkins, F. N. Judson, K. Penley, T. 1981. "Hepatocellular Carcinoma and Heparitis B Virus." Lancet Novem- Nylund, G. Christie, F. Meyers, J. N. Moore, Jr., A. Gardner, 1. L. Doto, J. ber 21, 1129-33. H. Miller, G. H. Reynolds, B. L. Murphy, C. A. Schable, B. T. Clark, J. W. Bhargava, M. Krishna. N.d. KarnatakaStareCancerConrrolProgramme. Kidwai Curran, and A. G. Redeker. 1982. "The Prevention of Hepatitis B with Memorial Institute of Oncology, Bangalore. Vaccine: Report of the Centers for Disease Control Multi-Center Efficacy Bjelke, E. 1982. "The Recession of Stomach Cancer: Selected Aspects." In K. Trial among Homosexual Men." Annals of Internal Medicine 97(3):362-66. Magnus, ed, Trends in Cancer Incidence. Washington, D.C.: Hemisphere Gambia Hepatitis Study Group. 1989. "Hepatitis B Vaccine in the Expanded Publishing. Programme of Immunization: The Gambian Experience." Lancet May Boyle, P.. [). G. Zaridze, and M. Smans. 1985. "Descriptive Epidemiology of 13:1057-60. Colorectal Cancer." InternationalJournal of Cancer 36:9-18. Ghaditian, P., J. P. Thouez, and A. Simard. 1988. "La g6ographie du cancer Bulatao-Jayme, J., E. M . Almero, C. A. Castro, T. R. Jardeleza, and L. A. de l'oesophage." Social Science and Medicine 27:971-85. Salamat. 1982. "A Case-Control Dietary Study of Primary Liver Cancer Gray,Nigel. 1985."CancerRisksandCancerPreventionintheThirdWorld." Risk from Aflatoxin Exposure." International Journal of Epidemiology In M. P. Vessev and M. Gray, eds., Cancer Risks and Prevennon. Oxford: 11:112-19. Oxford University Press. Bumgamer, J. Richard, ed. 1992. China: Long-Tetm Issues and Options in the Greenwald, P., and E. J. Sondik, eds. 1986. "Cancer Control Objectives for Health Transition. A World Bank Country Study. Washington, D.C. the Nation: 1985-2000." NCI Monographs 2. National Cancer Institute, Bvers, T. 1988. "Diet and Cancer. Any Progress in the Interim?" Cancer Washington, D.C. 62:1713-24. Hakulinen, T.,H. Hansluwka, A. D. Lopez, and T. Nakada. 1986. "Global and Chamberlain, I., N. E. Day, M. Hakama, A. B. Miller, and P. Prorok. 1986. Regional Mortality Pattems by Cause of Death in 1980." Internatonal "UICC (Union Intemational Contre le Cancer llntemational Union against Journal of Epidemiology 15:227-33. Cancer]) Workshop on Evaluation of Screening Programmes for Gastroin- Hanley, James A. 1986. "Evaluating Primary Prevention Programmes Against testinal Cancer." Internanonal Journal of Cancer 37:329-34. Cancer." Bullenn of the World Health Organization 64(2 ):3 11-20. Clayman,C. B. 1989. "MassScreeningforColorectalCancer:AreWeReady?" Hartunian, Nelson, Charles N. Smart, and Mark Thompson. 1980. "The JAMA 261(4):609. Incidence of Economic Costs of Cancer, Motor Vehicle Injuries, Coronary Correa, P. 1992. "Human Gastric Carcinogenesis: A Multistep and Multi- Hleart Disease, and Stroke: A Comparative Analysis." Public Health functional Process." Cancer Research 52:6735-40. 70(12):1249460. Cramer, D. W. 1982. "Uterine Cervix." In Schottenfeld and Fraumeni, Jr., Hodgson, Thomas. 1988. "Annual Costs of Illness Versus Lifetime Costs of eds., Cancer Epidemiology and Prevention. Philadelphia: W. B. Saunders. Illness and Implications of Structural Change." Drug Informantion Journal Crofton, J. 1984. "The Gathering Smoke Clouds: A Worldwide Challenge." 22:323-41. InternationalJournal of Epidemiology 13:269-70. Hsu, Hsu-Mei, Ding-ShinnChen, Cheng-Hua Chuang,]oseph Chih-FengLu, Cromwell, ]., and P. Gertman. 1979. "The Cost of Cancer." Lar'rngoscope De-Min Jebo, Chin-Chang Lee, Hsing-Chi Lu, Shih-Hsiung Chen. Yue- 89 3937"09. Fen Wang, Chinying Chen Wang, Kwang-Juei Lo, Chun-jen Shih, and Dahl,93 une, a v r . " v a a l l juen-Low Sung. 1988. "Efficacy of a Mass Hepatitis B Vaccination Program Dahl, June, and David Joranson. 1987. "Relieving Cancer Pain." World Health in Taiwan." JAMA5 260:2231-35. IARC (International Agency for Research on Cancer). 1986. Tobacco Smoking. Day, N. E.. C. J. Baines, J. Chamberlain, M. Hakama. A. B. Miller, and P. Monographs on the Evaluacion of the Carcinogenic Risk of Chemicals to Prorok. 1986. "ulcc (Union International Contre le Cancer llntemational Humans 38. Lyons. Union against Cancer]) Project on Screening for Cancer: Report of the y Workshop on Screening for Breast Cancer." InternationalJournalof Caticer iARC (Intemational Agency for Research on Cancer) Working Group on 38:303-8. EvaluationofCervicalCancerScreeningProgrammes. 1986."Screeningfor Squamous Cervical Cancer: Duration of Low Risk after Negative Results of EsDtliares of Adoadable Risks of Cancer in The UCted Soates TodaQy. Oxford: Cervical Cvtology and Its Implication for Screening Policies." Brinsh Med- Oxford Universitv Press. Eddy, David M. 1981. "The Economics of Cancer Prevention and Detection." India, Ministry of Health and Family Welfare. 1984. National Cancer Control Cancer 47:1200-09. Programme for India. Directorate General of Health Services, New Delhi. . 1986a. A Computer-Based Model for Designing Cancer Control Strate- ITFH (International Task Force on Hepatitis B Immunization). 1988. Notes on gies. NCI Monographs 64(2):74-82. Hepatitis B and Its Control. PATH (Program for Appropriate Technology in - 1986b. "Secondary Prevention of Cancer: An Overview." Bulletin of Health.) Seattle, Wash. Published report. the W'orld Health Organization 64(3):421-29. Kelsey, J. L., and N. G. Hildreth. 1983. Breast and Gynecologic Cancer Epide- Eddy, David M., and Michael Schwartz. 1982. "Mathematical Models in miology. Boca Raton, Fla.: CRC Press. Screening." In David Schottenfeld and Joseph Fraumeni, Jr., eds., Cancer Knight, K. K., ]. E. Fielding, and R. N. Battista. 1989. "Occult Blood Screening Epidemiology and Prevention. London: W. B. Saunders Company. for Colorectal Cancer." JAMA 261:587-93. Editorial Committee for the Atlas of Cancer Mortality, eds. 1976. Atlas of Knox, E. G. 1982. "Cancer of the Uterine Cervix." In K. Magnus, ed., Trends Cancer Mortalhty in the People's Republc of China. Beijing National Cancer In Cancer Incidence. Washington, D.C.: Hemisphere Publishing. Control Office of the Ministry of Health. Kristein, M. M., C. B. Amold, and E. L. Wynder. 1977. "Health Economics Feinstein, A. R., D. M. Sosin, and C. K. Wells. 1985. "The Will Rogers and Preventive Care." Science 195(4277):457-62. Phenomenon: Stage Migration and New Diagnostic Techniques as a Source Kurihara, Minoru. Kunio Aoki, and Suketami Tominaga. 1984. Cancer Mor- of Misleading Statistics for Survival in Cancer." Neut Enigland Journal of talirt Statistics in the World. Nagoya, Japan: University of Nagoya Press. Medicine 312:1604-8. Lea, E. A., and L. Garfinkel. 1979. "Variations in Mortality by Weight among Fleischer, D. E., S. B. Goldberg, T. H. Browning, J. N. Cooper, E. Friedman, 750,000 Men and Women."Journal of Chronic Diseases 32(8):563-76. F. H. Goldner, E. B. Keeffe, and L. E. Smith. 1989. "Detection and l uce, B. R. 1980. "Allocating Costs and Benefits in Disease Prevention Surveillance of Colorectal Cancer." JAIIA 261:580-85. Programs: An Application to Cervical Cancer Screening." Case Study 7. Francis D. P., S. C. Hadler, S. E. Thompson, J. C. Maynard, D. G. Ostrow, N. U.S. Congress, Office of Technology Assessment, Washington, D.C. 558 Howard Barnum and E. Robert Greenberg Lynge, E., M. Madsen, and G. Engholm. 1989. "Effect of Organized Screeninig Schweitzer. Stuart. 1977. "Cost Effectiveness of Early Detection of Disease." on Incidence and Mortality of Cervical Cancer in Denmark." Cancer Health Services Research 9(1):22-32. Research 49:2157-60. Scotto, Joseph, and Leonard Chiazze. 1976. "Third National Cancer Survey: Mackay, Judith. 1989. "Battlefield for the Tobacco War." I.AMA 261:28-29. Hospitalizations and Payments to Hospitals." DHEW Publication 76-1094. McM ichael, A. J. 1984. "Oral Cancer in the Third World: Time for Preventive Department of Health, Education, and Welfare, Washington, D.C. Intervention?" InternatonalJournal of Epidemiology 13:403-5. Segi, Mirsuo, Stiketami Tominaga, Kunio Aoki, and Isaburo Fujimoto, eds. Maynard, J. E., M. A. Kane, M. J. Alter, and S. C. Hadler. 1988. "Control of 1981. Cancer Mortality and Morbidity Statistics: Japan and the World. GANN Hepatitis B by Im munization." Viral Hepatitis ancd Liver Disease. New York: Monograph on Cancer Research 26. Tokyo: Japan Scientific Societies Press. Alan R. Liss. Published report. Miller.R.W.,andF.W.McKay. 1984."DeclineintheU.S.ChildhoodCancer Shapiro, S., W. Venet, P. Strax, L. Venet, and R. Roeser. 1982. "Ten to Mortality: 1950 through 1980."JAMA 251:1567-70 Fourteen Year Effect of Screening on Breast Cancer Mortality." Journal of Muir, C. S., and J. Nectoux. 1982. "Intemational Pattems of Cancer." In D. the National Cancer Insttute 69(2):349-55. Schottenfeldandj. F. Fratmeni,Jr.eds.,CancerEpide7nioLogyandPretvetinon. Silverberg, E. and J. Lubera. 1987. "Cancer Statistics 1987." Ca-A Cancer Pennsylvania: W. B. Saunders. JournalforClnicians. 37(1)2-19. Muir, C. S., J. Waterhouse, R. Mack, J. Powell, and S. Whelan. 1988. Cancer Simopoulos, A. P. 1987. "Obesity and Carcinogenesis: Historical Perspective." Incidence in Five Continents. Scientific Publication 88. Lyons: Internatlinal AmencanJournal of Clnical Numntion 45:271-76. Agency for Research on Cancer. Stanley, K. 1986. "Lung Cancer and Tobacco-A Global Proiblem." Cancer Nair, M. Krishnan. 1988. Ten-Year ActioT Plan for Cancer Control in Kerala. Detection and Prevention 9:83-89. Printed at St. Joseph's Press, Trivandrum. Approved by the State Cancer Stanley, K., J. Stjemsnward, and V. Koroltchouk. 1988. "Cancers of the ControlAdvis,)ryBoardofKerala.JointlypublishedbytheRegionalCancer Stomach, Lung, and Breast: Mortality Trends and Control Strategies." Centre, Trivandrum and the Cancer Unit, World Health Organization, World Health Statistics Quarterly 41:107-14. Geneva. StjernswrTd, J. 1988. "WHOc Cancer Pain Relief Programme." Cancer Surveys National Research Council, Committee on Diet, Nutrition, and Cancer. 1982. 7(1):195-208. Diet, Nutrition. and Cancer. Washington, D.C.: National Academy Press. . 1990. "National Training of Radiotherapists in Sri Lanka and NIH (National Institutes of Health). 1988. 1987 Annual Canicer Stanstics Zimbabwe: Priorities and Strategies for Cancer Control in Developing Review. NIH Publication 88-2789, National Cancer Institute, Division of Countries." InternationalJournal of Radiation Oncology and Biological Physics Cancer Prevention and Control. U.S. Departmenit of Health and Hunian 19(5):001-4. Services. Bethesda, Md. Stjernsward, J., K. Stanley, David M. Eddy, M. Tsechkovski, L. Sobin, 1. Koza, Parkin, D. M. 1985. "A Computer Simulation Model for the Practical Plani- and K.H.Notaney. 1985."CancerControl:StrategiesandPriorities."World ning of Cervical Cancer Screening Programmes." BntshJournal of Cancer Health Forum 6:160-64. 51:551-68. Sun Tsung-Tang, Chu Yuan-Rong, Ni Zhi-Quan, Lu Jlan-Hua, Huang Fei, Ni , ed. Cancer Occurrence in Developing Countries. Scientific Publication Zheng-Ping, Pei Xu-Fang, Yu Zhi-lan, and Liu Guo-Ting. 1986. "A Pilot 75. Lyons: International Agency for Research on Cancer. Study on Universal Immunization of Newbom Infants in an Area of Parkin, D. M., E. Laara, and C. S. Muir. 1988. "Estimates of the Worldwide Hepatitis B Virus and Primary Hepatocellular Carcinoma Prevalence with Frequency ofSixteen Major Cancers in 1980." IntenmtanonallournalofCancer a Low Dose of Hepatitis B Vaccine." Journal of Cellular Physiology 4(supple- 41:184-97. ment):83-90. Parkin, D. MI., and S. M. Moss. 1986. "An Evaluation of Screening Policies Szmuness, W., C. E. Stevens, E. A. Zang, E. J. Harley, and A. Kellner. 1981. for Cervical Cancer in England and Wales Using a Computer Simulatioin "A Controlled Clinical Trial of the Efficacy of the Hepatitis B Vaccine Model." Journal of Epidemiological and Community Health 40(2):143-53. (Heptavax B): A Final Report." Hepatology 1:377-85. Percy, C., and C. S. Muir. 1989. "The International Comparability of Cancer Tabar, L_. C. J. Fagerberg, A. Gad, L. Baldetorp, L. H. Holmberg, 0. Grontoft, Mortality Data: Results of an International DeathCertificate Study." Amer- U. Ljungquist, B. Lundstrom, J. C. Manson, and G. Eklund. 1985. "Reduc- ican Journal of Epidemiology 129:934-46. tion in Mortality from Breast Cancer after Mass Screening with Mammog- Peto, Richard. 1986. "Influence of Dose and Duration of Smoking on Lung raphy." Laticet 1(8433):829-32. Cancer Rates." In D. G. Zaridze and Richard Peto, eds., Tobacco- A Major Takenaga, Nobuyuki, Ichiro Kai, and Gen Ohi. 1985. "Evaluation of Three Internatonal Health Hazard. Scientific Publication 74. Lyons: International Cervical Cancer Detection Programs in Japan with Special Reference to Agency for Research on Cancer. Cost-Benefit Analvsis." Cancer 55(1 0):2514-19. Poirier, S., H. Ohshima, G. De-The, A. Hubert, NI. C. Bourgade, and H. Thom, Jennifer B., J. Elizabeth MacGregor, Elizabeth M. Russell, and Kath- Bartsch. 1987. "Volatile Nitrosamine Levels in Common Foods from Tuni- leen Swanson. 1975. "Costs of Detecting and Treating Cancer of the sia, South China, and Greenland: High-Risk Areas for Nasopharvngeal UterineCervix in North-East Scotland in 1971." Lancet March 22674-76. Carcinoma (NPC)." IntrnationalJournal of Cancer 39:293-96. USDHHS (U.S. Department of Health and Human Services). 1977. Cancer Portenoy, Russell K. 1988. "Practical Aspects of Pain Control in the Patient Patient Survival. Report 5. DHEW Publication NIH (National Institutes of with Cancer." Ca-A CancerJournalfor Clinicians 38(6):327-52. Health) 77-992. Department of Health, Education, and Welfare; Washing- Rice, D. P., andT. A. Hodgson. 1981. SocialanidEcononiiclInipLcanonsofCanicer ton, D.C. in the United States of America. Publication(PHs) 81-1404. U.S. Department . 1982. The Health Consequences of Smoking: C.ANCER. Report of the of Health and Human Services, Washington, D.C. Surgeon General. Public Health Service, Office of Smoking and Health, Rice, D. P., T. A. Hodgson, and A. N. Kopstein. 1985. "The Economic Costs Washington, D.C. of Illness: A Replication and Update." Health Care Financing Review' . 1988. 1987 AnniaLCancerStatistics Review, IncLudingCancerTrends: 7(1):61-80. 1950-1985. NIH (National Institutes of Health) Publication 88-2789. Na- Schottenfeld, D., and J. F. Fraumeni, Jr. 1982. Cancer Epidemiolog-s and Preven- tional Cancer Institute, Washington, D.C. tion. Philadelphia: W. B. Saunders. Van Rensburg, S. J., P. Cook-Mozaffari, D. J. Van Schalkwyk, J. J. Van Der Schwartz, M. 1978. "An Analysis of the Benefits of Serial Screening for Watt, T. J. Vincent, and 1. F. Purchase. 1985. "Hepatocellular Carcinoma Breast Cancer Based on a Mathematical Model of the Disease." Cancer and Dietary Aflatoxin in Mozambique and Transkei." British Journal of 51:1550464. Cancer 51:713-26. Cancers 559 Wainwright, R. B., B. J. McMahon, L. R. Bulkow, D. B. Hall, MI. A. Fitzgerald, . 1984. "Control of Oral Cancer in Developing Countries." Buletin of A. P. Harpster, S. c. Hadler. A. P. Lanier, and W. L. Heyward. 1989. the Vorld Health Organization 62(6):817-30. "DurationoflmmumogenicityandEfficacyofHepatitisBVaccineinaYupik . 1985. "Essential Drugs for Cancer Chemotherapy: Memorandum Eskimo Population." JAXiA 261:2362466. from a NWIHO1 Meeting." Bulletin of ihe World Health Organization 63(6): Wamakulasuriya, K. A. A. S., A. N. 1. Ekanayake, S. Stvayohani, J. 999-1002. Stjernsward, J. J. Pindborg, L. H. Sohin, and K. S. G. P. Perera. 1984. _ -. 1986a. Cancer Pain Relef. WHO Monograph. Geneva. "Utilization of Prinmary Health Care Workers lor Early Detection of Oral Cancer and Precancer Cases in Sri Lanka." Bulletin of the World Health - . 1986h. "Control of Cancer of the Cervix: A WHO Meeting." Bultin (- rr-A -,,, ar - ~i' ' 243-50. oftheWorldHealthOrganzation 64(4):607-l8. Warner, Kenneth. 1987. "Health and Economic Implications of a Tohacco . 1986c. "The Use of Quantitative Methods in Planning National Free Society." ]AMA 258:2080-86. Cancer Control Programmes." Bulletn of the World Health Organization Waterhouse,J.,C. S. Muir, P.Correa,andj. Powell,eds. 1976. Cancer ticidence 64(51:683-93. in Five Continents. Vol. 3. Scientific Publication 15. Lyons: International . 1988a. "General Strategies and Provisions for Cancer Control in Agency for Research on Cancer. Chile." CAN/88. 1. Geneva. Waterhouse, J., C. S. MUIT, K. Shanmugaratnam, and J. Powell. eds. 1982. - . 1988b. "Global Medium-Term Programme for the WHO Cancer Cancer Incidence in Five Continents. Vol. 4. Scientific Puhblcatioll 42. Lyons: Control Programme during the Period 1990-1995." CAN/MTP/88. 1. Geneva. International Agency for Research on Cancer. ----. 1990. "Diet, Nutrition, and Prevention of Chronic Diseases." Tech- Widjaya, A. 1988. "Cost Effective Analysis in Various Hepatitis B Virtis nical Report. Geneva. Control Programs in Indonesia." Department of Health, Centre for Laho- WHO (World Health Organization)/indonesia, Ministr' of Health. 1989. ratLiry Services, Jakarta. Nanional Cancer Connrol Program in Indonesia. Geneva. Willett, W. S., and B. MacMahon. 1984. "Diet and Cancer-An Overview Yeh, F.-S., NI. C. Yu, C.-C. Mo, S. Luo, NM. J. Tong, and B. E. Henderson. (Parts I and 2)." New Englatidjounal of, vMedicine 310:633-38. 1989. "Hepatitis B Virus, Atlaroxins, and Hepatocellular Carcinoma in World Bank. 1988. World Development Report 1988. Washington, D.C.: Ox- Southern Guangxi, China." CancerResearch 49:2506-9. ford University Press. Yu, NI. C., J. H. C. Ho, S.-H. Lai. and B. E. Henderson. 1986. "Cantonese-Style WHO (World Health Organization). 1983. "Prevention of Liver Cancer." Salted Fish as a Cause of Nasopharyngeal Carcinoma: Report of a Case- Technical Report 691. Geneva. Control Study in Hong Kong." Cancer Research 46:956-61. 22 Diabetes J. Patrick Vaughan, Lucy Gilson, and Anne Mills In recent years there has been growing concern that diabetes insulin-dependent diabetes mellitus (ID)DM) generally occurs mellitus is becoming more common, mainly in the more ur- among younger age groups (with 25 to 50 percent of patients banized and industrialized countries, where the prevalence presenting before the age of fifteen years), and it is nearly rates of the disease in the total population are often I to 3 always acute in onset. Sufferers require regular doses of insulin, percent or more. In these countries it is widely agreed that by injection or a similar process at least once per day, in order diabetes is a significant public health problem, particularly to sustain life and to avoid acute and more long-term compli- among people in the older age groups. There are also well- cations. Those with non-insulin-dependent diabetes mellitus documented populations in developing countries in which (NIMDNM) usually suffer from a less severe illness, which has a diabetes has become much more frequent in the past ten to sloweronsetandismostcommonintheolderagegroups(older twenty years. Given the changing age structures and health than forty years). People with NIDDM, however, may suffer from pattems of the populations of developing countries, what the same long-term complications as those with InMM, such as public health priority should be given to diabetes, now and for retinopathy, nephropathy, neuropathy, and ischemic heart dis- the next twenty years or so? ease. In this review we will focus on these two types of diabetes In order to answer this question, we summarize in this and will largely consider their epidemiology separately. chapter the information on the frequency and time trends for With regard to diabetes incidence by sex, IoIMi appears to be diabetes in developing countries and on the indirect and direct about equal in males and females (Rewers and others 1988), costs of the disease. We consider the evidence for prevention but NlDDh1 may be more frequent in females. A third type of and case management strategies and assess the feasibility and diabetes, now frequently called malnutrition-related diabetes cost of these strategies. Until recently little consistent infor- mellitus (MRDM), has been reported from many developing mation was available on diabetes in developing countries, countries. The patients are usually young and have a history of but under the leadership of the International Diabetes nutritional deficiency. This disease is believed to be clinically Federation and the World Health Organization (WHO), in- distinct, and therefore the separate grouping has been proposed terest in the subject has grown steadily during the past ten to (WHO 1985). It has been extensively reviewed (Abu-Bakare fifteen years. A great deal of this information has been well and others 1986), but the incidence rate of MRDh4 is still largely summarized in a World Health Organization Technical Report unknown. Although the etiology is not understood, it is possi- (wHo 1985). bly caused by toxins in cassava, other food toxins, or protein- Diabetes mellitus is a chronic and noncommunicable dis- energy malnutrition. The case management is similar to that ease which is largely irreversible. Although it can occur at any for IDrIN. We will not consider this form of diabetes separately age, its onset is most frequent among the young and older here because there is inadequate epidemiological evidence and persons. Diagnosis is based on finding an abnormally high level the subject clearly needs further research. of glucose in the blood, a condition caused by poorly function- Some healthy individuals have lesser degrees of tolerance to ing beta cells in the pancreas gland and an insufficient output glucose, and when challenged with a dose of 75 grams of of the hormone insulin. The actual Linderlying etiological glucose taken by mouth (wi-ic 1985) they cannot be classified mechanisms that lead to this pathological state, however, are as diabetics, but they are, nevertheless, at increased risk of still largely unknown. coronary heart and peripheral and cerebrovascular diseases. Despite the fact that all diabetes cases have been classified About one-third of these individuals with impaired glucose and reported under one code (number 2 50) in the International tolerance will revert spontaneously to a normal state but, as a Statistical Classification of Diseases, Injuries, and Causes of Death group, people with impaired tolerance are at higher risk of (WHO 1975), it is now generally accepted that epidemiologi- subsequently developing diabetes mellitus and are believed to cally' there are two main types of the illness, and this is to be make a significant contribution to total mortality (Bennett acknowledged in the forthcoming revision. The onset of 1985; Grabauskas 1988). 561 562 J. Patrick Vaughan, Luc-v Gilson, and Anne Mills Gestational diabetes usually presents as NIDDM and only of undiagnosed NIDDM for every one or two diagnosed diabetics rarely is insulin required. There is poor epidemiological data in the community. on this condition in developing countries, and, although its When interpreting information on the frequency of a importance is not denied, criteria for its diagnosis remain chronic and irreversible disease such as diabetes, it is crucial to controversial ("Glucose Tolerance in Pregnancy" 1988). be clear what the incidence and prevalence estimates may The classification of diabetes into three main types and the mean. For example, IDDM is a relatively rare disease with regard criteria for diagnosing impaired glucose tolerance have gained to incidence, but the prevalence rate can reach 0.5 percent, or wide acceptance only during the past five to ten years. This I in 200 people, because of the long duration of survival with causes considerable difficulty in interpreting much of the older good case management and medical care. Therefore, as the published literature and in comparing newer with older data. case management for individual cases improves in developing The new internationally accepted criteria for diagnosing dia- countries, the prevalence rates may rise without any significant betesandimpairedglucosetoleranceareshowninappendix 22A. change in the real incidence. This has implications in the assessment of altemative strategies for addressing IDLOM. Where The Significance of Diabetes to Public Health the incidence is low the costs per new case averted by a preventive strategy may seem relatively high. With regard to Before considering the distribution and time trends for diabe- case management, however, IDDMi may become a relatively tes in various parts of the world, it is important to understand common disease, and the cumulative costs per diabetic treated the problems encountered in interpreting the available infor- are considerable; thus, substantial savings on treatment could mation, particularly that on incidence. result from an effective preventive strategy. It is, therefore, important to take account not only of control costs but also of Limitations of the Morbidity and Mortality Information treatment savings when assessing preventive strategies. Since the duration of survival varies, it is important when making The diagnosis of new cases ofdiabetes depends both on clinical comparisons of the cost-effectiveness of treating chronic dis- symptoms and signs and on the detection of an elevated blood eases such as IDMI to standardize for illness duration, by using glucose level, or, where this is not possible, on the persistent a unit such as cost per year of life saved. presence of glucose in the urine and a satisfactory response to In general, industrial countries have good information on the appropriate treatment. The detection and reporting of new the epidemiology of IDMM and NIDDM, but for many developing cases, therefore, depends heavily on the availability and use of countries the national data are very scantv or do not exist. This health services, or on the results of large-scale population- is well illustrated by map 22-1, which represents data collected based surveys. Access to, and use of, health facilities is poor in by the World Health Organization up to the early 1980s (WHO many developing countries, and so the reported figures for the 1985) and shows unrealistic and low prevelance levels for frequency of cases (both for incidence and prevalence) of wluni many areas, especially in Latin America and Africa. and NIDDM must be highly suspect. The results of many of the older surveys are also suspect because of the use of non- Current Trends for Insulin-Dependent Diabetes standardized diagnostic criteria, different screening methods, and inadequate sample sizes. Such surveys were undertaken There is considerable discussion about the etiology of] DM, but before the present classification and diagnostic criteria were it is clear that both genetic and environmental factors are internationally accepted. Thiscautionarv note is equally ifnot involved (Krolewski and others 1987). Studies of identical more important for mortality data. The International Classifi- twins show an overall concordance rate for developing diabe- cation of Disease (IcD) statistics, tnow used by most national tes of over 50 percent, rising to about 70 percent if certain death registration and certification systems, is based on naming genetic markers are also included. These rates contrast with the "underlying" pathological process, but the ICD data do not the fact that 85 percent of newly diagnosed diabetics have no distinguish between the epidemiologically different forms of close relative with the same condition (Bennett 1985). Other diabetes (that is, between IDDM and NIDDM). Moreover, it is well evidence also suggests that although the clinical onset is acute recognized that in most developing countries the registration and severe, there is probably a long latent period before the of deaths is grossly inadequate, and even the certification of illness becomes apparent (Tarn and others 1988). the pathological causes of the registered deaths is often incor- The incidence of new cases of mlmi has been found to vary rect. In addition, diabetes mellitus is frequently not included considerably with the seasons, more new cases presenting by the certifying doctor on death certificates, and coding rules during the winter months (DERI 1988). This variation has been preferentially select cardiovascular diseases and cancers in linked to the possibility that IDDM may be caused by a viral favor of diabetes (WHO 1985). Studies in the United Kingdom infection, hut a number of extensive reviews have concluded and the United States suggest that up to 75 percent ofdiabetics that, apart from a few instances, no good epidemiological nay not be counted in the internationally published mortality evidence exists for this hypothesis (Gamble 1980; Barratt- data (Fuller and others 1983). This situation has led to the Connor 1985). It should also be remembered, however, that a following crude, but general, rule. In populations in which viral infection occurring early in a person's life may not be diabetes is relatively common, there is probably another case detected and hence not associated with the later onset of IDLD. Map 22- 1. Prevalence of Diabetes Mellitus in Some Countries Irsd7nl~~~~~~~~ 03 1 N 0 \\ Pima _ A Highland pop,.,ation N"~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ar A~f Urban' \~....v "'~Indians Caucasians .:Rua "Africans Caucasians- Aboric N,'s [Z1-2% 5 -10% > 20% / / , Caucasians N .'' n~7 2 - 5% P 10 - 23 Molynesians Note: The rates shown include both NIDDM and IDIIM, though the latter represents a small proportion of the total. Rates are derived from many sources, principally national and regional surveys but also hospital itatistics in somiie developing CoUntries. In most cases WHO standardized diagnostic criteria have not heen used. Source: Reproduced by permission of the World Health Organization, Geneva fromi Diabetes mellitus: Report of a wox- Study Group (Technical Report Series no. 727, 1985). 564 J Patrnck Vatughani, LucN Gilson, and Anne Mill In addition, the incidence of IDM does not appear to have steady, there is some evidence to suggest that recently it has fallen with the much wider use of childhood vaccines. On the been rising, particularly in Finland (Reunanen and others basis that an infective process may be involved in triggering 1982), Scotland (Patterson and others 1983), and Poland off the onset of diabetes, it has been suggested that large-scale (Rewers and others 1987). The authors of another analysis of immunization may even lead to an increase in IDDM in develop- standardized epidemiological data from sixteen population- ing countries (King 1987). based IL)t)M case registries (all in industrial countries) have The incidence of li)Dm appears to vary greatly between concluded that there has been a linear increase in incidence different ethnic groups, being commonest in white people and during the past two decades in Europe and the western Pacific in the northem temperate zone. lndeed, there may be as much but not in North America (Diabetes Epidemiology Research as a tenfold increase in incidenice rates from southern Europe International [DERI] 1990). The situation in developing coun- to Finland. which has one of the highest rates worldwide, and tries is inconclusive because of the very poor data base. Even there is a thirty-four-fold difference in the childhood incidence if the incidence is rising, IDDM is still, in general, considered to of the disease between the highest in Finland and Japan (DERI be a rare disease in most developing countries. 1987). The Finnish incidence of diabetes in children between the ages of zero to fourteen (in which about half of the new Current Trends for Non-Insulin-Dependent Diabetes nDM cases would be expected to occur) is estimated at 29 per 100,000 persons per year. Assuming an average duration of When considering the balance between genetic and environ- survival of fifteen to twenty years, the prevalence of IDNDM in mental factors in the etiology of NIDDM, we find that the Finland in the total population would be about 0.5 percent. evidence for a genetic susceptibility appears in some ways to In many developing countries, particularly in Africa, iDLWo be stronger than that for IDDM. The concordance rate in iden- is considered to be a rare disease, but no reliable estimates are tical twins is higher for NIDDM, but so far no genetic markers available. If, for the sake of illustration, we assume an inci- have been discovered. There is good evidence that the inci- dence rate for a developing country of between I and 5 new dence of NIDDM does vary considerably between different eth- cases per 100,000 children age zero to fourteen years per year nic and racial groups, such as Indians and Chinese (Zimmet (and that those children are 40 percent of the developing 1982). It is also generally agreed that the prevalence (and country population), we might expect between 4 and 20 new probably incidence) of NIDDM can rise in the relatively short cases of IDDi per year in children in a population of I million period of one to two decades as more people, such as the people, or approximately twice this number (that is, 10 to 40 population of Nauru (Schooneveldt and others 1988) and new cases) among all age groups per year. Assuming an average Australian aborigines (Cameron, Moffitt, and Williams 1986), dtiration of survival for ILM cases in developing countries of become urbanized and "westernized." Studies have shown that five years, we might expect the population prevalence to be increased food intake, obesity, and lack of exercise can all be 0.005 to 0.02 percent. Even at such low prevalence levels the associated with NIDDM. Considerable discussion still exists, widespread lack of diagnostic and treatment facilities could however, as to whether these are causal factors operating in lead to a high case-fatality rate within the year following the susceptible individuals or ethnic groups and whether reversing onset of the disease. Moreover, a high case-fatality rate means these trends in populations would lead to a reduction in the that even if the incidence rate were higher than predicted the incidence of NIDDM. disease would appear to be very uncommoTn. Eighty-five to 90 percent of all diabetics in industrial coun- Limited contact with the health system, scarcity of specialist tries suffer from NIDDM, and the rapid rise in the prevalence of staff, and the lack of regular supplies of insulin and the neces- diabetes in the United States from just under I percent of the sary equipment are bound to lower the prognosis for patients general population in 1960 to over 2.5 percent by the late with IDLM in developing countries. A report from a specialist 1970s was largely due to an increase in NIDDM (Zimmet 1982). clinic at a large teaching hospital in a capital city in Sub- Although some of this increase may have been the result of Saharan Africa put the case-fatality rate at 30 to 40 percent improved diagnosis and prognosis, there is good evidence that during the first four to six years following diagnosis (Lutalo and the incidence of NIDDM has also risen. In 1936 the incidence of Mabonga 1985). Even with appropriate care, patients under- diabetes in the United States was reported as 5 per 10,000 goinlg insulill therapy mav suffer from hypo- (low) or hyper- population per year and by 1973 it was 29.7 per 10,000, a glycemia (elevated blood glucose). They are also especially sixfold increase. Worldwide, the most spectacular rise in the vulnerable to the potential complications resulting from injec- incidence of NIDDm has been clearly documented in the Pima tions-such as sepsis, hepatitis, and acquired immunodefi- Indians of North America (Godhes 1986) and in certain ciency syndrome (AIDns)-if they do not strictly adhere to Melanesian, Micronesian, and Polynesian island populations sterilizationi procedures. Education programs need to stress in the Pacific (King and others 1984; King and Zimmet 1988). the importance of preventing or immediately combating such Such data from "special" populations has strengthened the problems. belief that genetic susceptibility to NIDDNM is unmasked as such There is an unresolved question about whether the inci- people undergo urbanization and modernization (Zimmet dence of lmDi is stable or rising, particularly in Europe and the 1982; Zimmet and others 1986). For example, studies compar- United States. Although incidence appears to be relatively ing Indian immigrants and indigenous people living in coun- Diabetes 565 tries as diverse as Fiji (Zimmet 1982), Singapore (Cheah and that the information available for Africa and Latin America Tan 1979), South Africa (Marine and others 1969), and is incomplete and that very little up-to-date information is Trinidad and Tobago (Poon-King and others 1968) have available in which the new international guidelines for the shown prevalence rates for diabetes in Indians of 14.0, 6.1, diagnosis of NIDDM have been used (King and Zimmet 1988). 10.4, and 4.5 percent, respectively, all of which were higher Most of the available information is for prevalence estimates, rates than for the indigenous peoples. Both forms of diabetes and very little data exists on incidence. The information for were previously thought to be uncommon in India, but a recent Latin America is sparse and poorly standardized (Seneday and survey of known diabetics showed a higher prevalence among Masti 1987). residents in a wealthy suburb of Delhi than in Indians living in London. The crude prevalence of diabetes in Delhi was Economic Costs of Diabetes found to be 3.1 percent, and as many as 16 percent were being treated with insulin (Mather and Keen 1985; Mather and Using epidemiological and financial data, the economic costs others 1987). Another community-based survey in Coventry, of an illness to society can be calculated by its indirect and United Kingdom, found age-adjusted prevalence rates for dia- direct costs. The former reflect the cost of morbidity and betes of 11.2 percent in Asian men and 8.9 percent in Asian mortality to the community as a whole, and the latter, the costs women in contrast to 2.8 percent and 4.3 percent in white men to the health sector of prevention, diagnosis, and treatment. and women, respectively. The difference was not explained by Unfortunately, virtually no published economic study has differences in body mass (Simmons and others 1989). differentiated between the two forms of diabetes. The few A rise in the prevalence of NIDDM has been fairly widely studies of the economic burden that are available assess indi- reported from other migrant groups. For instance, diabetes was recr costs with respect ro lost production. Only one study for a thought to be uncommon in mainland China (Shanghai Dia- developing economy has been found (Guam). A review of betes Research Cooperative Group 1980), but surveys in Sin- studies in industrial countries, however, can point to the gaporean Chinese revealed that prevalence had risen from 1.6 potential burden diabetes may represent for developing coun- percent in 1975 to 4.0 percent ten years later (Thai and others tries and the potential benefits to be gained from prevention 1987). A survey of people age forty years and older in urban and case management programs. In table 22-1 we summarize and rural Taiwan (China), carried out during 1985-86, re- the main studies that are currently available. These studies vealed age-adjusted prevalence rates of 7.6 percent and 4.7 inevitably suffer from the weaknesses of the epidemiological percent, respectively, and by 1984, diabetes, mainly NIDDM, and economic data on which they are based and the fact that ranked fifth as a cause of death in Taipei, Taiwan (Tong-Yuan IDIOM and NIDDM are combined as one disease in the analysis. Tai and others 1987). Two other recent surveys are worth Two particular problems with using estimates from indus- noting because they demonstrated surprisingly high preva- trial countries and projecting them for developing countries lence rates. In a rural population in Saudi Arabia a crude are that it is unclear what level of disability or death results prevalence for all ages and both sexes of 4.3 percent was found, from illnesses to which diabetes contributes and what the but the proportion rose to 13.4 percent for both sexes age likelihood is that either IDDM or NIDDM patients will develop fifty-five years or older and to 18.7 percent for the females in complications. In general, the only apparent pattem is that the this group (Fatani and others 1987). In Tunisia, surveys of an longer a patient survives with either form of diabetes, the more urban and a rural population showed the age-standardized rate likely it is that such complications will develop (wHo 1985). to be twice as high or higher in the urban population (4.6 A WHO study group reports that in industrial countries diabetic percent as opposed to 2.3 percent in men and 3.5 percent in kidney disease, for example, is present in one in six diabetics contrast to 0.6 percent in women). Within the urban sample and directly causes or contributes to premature death in 50 the prevalence rate was similar for those people born in the percent of those in whom IDDM began in youth (WHO 1985). urban area as compared with those born elsewhere in the Nearly all IDDM patients and many NIDDM patients will eventu- country (Papoz and others 1988). ally develop some form of eye disease, but only some of them Some evidence from Sub-Saharan Africa suggests that appear to be at risk of developing the severe life-threatening NIDDM may be increasing in some urban populations. For in- complications. The risk of coronary heart disease is two to stance, in a report from a chronic disease register compiled in three times higher in both IDDM and NIDDM patients older than Zimbabwe it was shown that diabetes was responsible for 12.4 forty years in industrial countries. In these countries the out- percent of the cases (Lutalo and Mabonga 1985), and in look for stabilized IDDNI patients is good and there is some population-based estimates from Tanzania the prevalence in evidence that good control of the disease can delay (and may adults was between 0.2 percent and 1.1 percent (Ahren and even prevent) the onset of long-term pathological complica- Corrigan 1985; McLarty and others 1989). Such estimates, tions (WHO 1985; Ward 1988). Indeed, after stabilization, however, contrast with a report from rural Nigeria, where no patients with diabetes in industrial countries may live for one with diabetes was found in a survey of more than 1,300 fifteen or more years before the onset of complications-al- villagers (Teuscher and others 1987). though life expectancy is generally reduced by up to one-third The authors of a most useful review of the world situation (WHO 1985). This is true for nephropathy and probably also on trends in the incidence and prevalence of NIDDM conclude for the other complications of diabetes, and individual 566 1. Patnck Vaughan, Lucy Gilson, and Anne Mills Table 22 1. Studies on the Economic Burden of Diabetes Indirect/direct cost Country Year Total economic burden (percent of total cost) Source Guam 1976 us$3 million 80/20a Kuberski and Bennet 1979 Sweden 1979 1,317 million ,kr 57/43 Jonsson 1983 United Kingdom 1979-80 144.3 million 42/58 Laing 1981 United States 1969 us$2.6 billion 62/38 SBMLICc United States 1973 us$4.0 billion 59/41 SBMLl(c United States 1975 us$5.3 billion 53/47 SBMLIC d United States 1979 us$15.7 billion 64/36 Platt and Sudovard United States 1980 us$9.7 billion 51/49 SBMLICc United States 1984 us$13.8 billion 46/54 SBMLlCc United States 1987 Ls$20.4 billion 53/47 American Diabetes Association 1988 a. Approximate. b. Excludes cost of 64,047 years lost through premature death. c. Statistical Buireau of the Metropolitan Life Insurance Company. Discussed in Songer, in press. d. Discussed in Songer, in press. Source: See last column. susceptibility interacts with levels ofdiabetes control to deter- because of premature death. This picture of the balance of mine the tissue response, rate of damage, and ultimate severity indirect costs differs from that in most industrial country in the organ concerned. Early recognition and treatment of studies, in which morbidity costs are dominant. The difference retinopathy and foot problems, for example, can reduce dis- may reflect the younger average age at death of diabetics in ability and prolong life. Correction of hypertension and hyper- Ghana; many of the deaths could occur before the onset of lipidemia are also important to prolong survival. complications and other disabilities. It may also reflect tech- nical differences between the studies: the monetary valuation INDIRECI COSTS. The broad trends shown by the studies of the costs in industrial country studies and the use of undis- suggest that indirect costs are of decreasing importance within counted days of life lost in the Ghana study. the overall economic burden of diabetes (for example, in the Overall it is difficult to suggest the likely level of the indirect United States they fell from about 62 percent in 1969 to 46 costs associated with diabetes in developing countries because percent in 1984). In most studies indirect costs are dominated of poor epidemiological data and the failure to separate the two by the cost of the disability caused by the disease, although the forms of the disease in the available cost data. Indirect costs most recent U.S. study (American Diabetes Association 1988) will also depend on the level and quality of the health care suggests that mortality costs are more important. This finding, available. Because IDDM is a rarer disease worldwide, it might however, probably results from the inclusion in the study of be expected that NIDDM would dominate the indirect costs deaths in which diabetes was a contributory cause rather than attributable to diabetes in developing countries. The former only those deaths directly due to the disease. Unfortunately disease, however, occurs in younger age groups and is likely to this study did not also consider the potentially important costs cause death, so years of life lost as a result of the disease may of disability in which diabetes was a contributory factor, in- be substantial even though it is an uncommon disease. In stead focusing only on the costs of disability directly caused by contrast, NIDDM is more common, occurs in older people, and diabetes. Taken together, the total of these disability costs is less likely to cause death if untreated; but it may lead to would probably exceed mortality costs and dominate indirect substantial disability as a result of the complications associated costs, even in this study. with diabetes. As the prevalence of NIDDM rises and the age of The only figures from developing economies that shed light onset falls (as some evidence appears to indicate is likely), on indirect costs are those reported from Guam (Kuberski and these disability costs will increase. Both forms of diabetes, Bennet 1979) and Ghana (Ghana Hiealth Assessment Project therefore, have serious but not directly comparable conse- Team 1981). Of the total costs attributable to diabetes in quences with regard to indirect costs. Guam, about 80 percent were indirect costs, but it is not clear whether mortality or morbidity costs were dominant. In DIRECT COSTS. Only Kuberski and Bennet (1979), in their Ghana, the average age at onset of the disease was estimated study in Guam, discuss the direct costs of diabetes in develop- to be forty years (suggesting that only NIDDM was considered), ing economies. Direct hospital costs alone exceeded $600,000 with a 50 percent case-fatality rate after fifteen years (average in 1976, including 5,352 disability days from 435 patients age at death being fifty-five) and 30 percent disablement admitted to the hospital (an average of 12.3 days per person before death. The total days of life lost were calculated as 217 admitted).' These costs, however, will probably not be repre- per 1,000 persons per year; 52 percent of these days were lost sentative of all developing economies but will reflect the Diabetes 567 relatively high level of care offered within Guam's health Prinmary Prevention system. Simlarly, the evidence of direct costs from industrial coun- Beliefsconcerningthecauseorcausesofdiabeteshaverecently tries reflects their more sophisticated health systems and so moved away from genetic and immunological explanations to cannot be directly transferred to situations in developing coun- a much greater emphasis on environmental factors, thus in- tries. Industrial country studies do, however, indicate the main creasing the relevance of primary preventive strategies. An influences on these costs and may suggest future cost levels for eminent international study group claimed in 1987 "that at developing countries. least 60 percent of IDDM worldwide, and perhaps over 95 The studies that are available show that the cost of diabetes percent, is environmentally determined and thus potentially treatment programs is substantial and is increasing in industrial avoidable" (DERI 1987). Even if this is true, however, the causal countries (for example, rising from $1.65 billion in the United factors in the environment have not been clearly defined, and States in 1973 to $7.4 billion in 1984). This trend reflects the although it may be possible to identify some high-risk individ- inflation of medical care prices, the increased prevalence of uals, the costs and technical difficulties would prohibit this diabetes, the increased use of medical care among diabetics, option, even in most industrial countries (Zimmet 1987). As and the development of new treatment technologies (Songer, a consequence, it is not possible to make specific recommen- in press). Direct costs are now equally important to, if not more dations for the prevention of IDDMI in industrial, let alone important than, indirect costs within the overall economic developing, countries, and given the apparently low incidence burden of the disease. A high portion of direct costs (usually rates in the latter, a preventive strategy would not appear to not estimated) are likely to be the result of the complications be a high priority for them. A far greater priority for diabetes of, and illnesses associated with, diabetes rather than of diabe- in developing countries lies with further international collab- tes itself. Indeed, one study demonstrated that patients with orative research to establish the true incidence and the deter- chronic complications of diabetes incurred health care costs minants of IDDM through the use of population-based studies. fourteen times as high as diabetic patients without any record Although diabetes has been studied using such registers in of complications (Gambert and others 1988). Prevention and numerous industrial countries, no successful such register yet case management programs, therefore, have the potential both exists in Africa or Latin America (DERI 1987). to reduce the indirect costs of the disease and to reduce the At present there are also no proven intervention strategies costs of caring for complications. Still, the cost of such pro- that reduce the incidence of NIDDM. Still, many authorities grams is itself dependent on the nature of the strategy adopted, believe there is now sufficient evidence that experimental and careful consideration of the cost and effectiveness of community or population intervention studies should be made treatment options is important in seeking to contain costs. of lowered dietary intake of carbohydrates (including reduc- Songer (in press) suggests, for example, that economic evalu- tions in fat and sugar intakes), reduced obesity, and increased ations techniques should be used in evaluating screening pro- physical exercise and activity (Zimmet 1987). Considerable grams, insulin treatment programs (multiple as opposed to attention would have to be given to establishing sound meth- single injections as opposed to insulin pump therapy), compli- ods for measuring both the successful implementation of the cations treatment programs (laser surgery, dialysis, transplants, interventions and for evaluating the possible changes in inci- and so on), and home blood glucose monitoring programs. dence of NIDDM over time. In most developing countries the existing level of direct The links between diabetes, coronary heart disease, hyper- costs associated with diabetes is likely to be low. Although the tension, and other noncommunicable diseases have led the true incidence of IDDM is not known, it is clear that many such World Health Organization to propose an integrated program patients probably do not survive long after the onset of illness for the prevention and control of noncommunicable diseases and so do not obtain medical care. Those people with diabetes (for example, by stressing good nutrition, avoidance of obesity, who do survive receive only such care as is available, which for increased physical activity, and reductions in smoking and many of these countries will be limited, leading to high case alcohol consumption). A parallel has been drawn with inter- fatality in the first few years following diagnosis. The costs in vention projectsforcoronaryheart disease (wHo 1985; Zimmet Guam may suggest the upper end of the cost range for devel- and others 1986). Because no specific and modifiable causes oping economies, and the lower end is probably suggested by are known to account clearly for the rise in diabetes, a broad the hypothetical costing for diabetes case management pre- strategy that tackles a wider range of emerging health problems sented later in the chapter. makes sense for many developing countries. The strategy would need to rely heavily on modifying individual human Lowering or Postponing Diabetes Incidence behavior, improvements in the health services, mass health education, and government regulation and legislation. A reduction in the incidence of new cases of diabetes will only be achieved by primary prevention strategies, whereas the Costs and Consequences of Preventing Diabetes incidence of diabetes complications may be reduced by im- provements in case management and through earlier case In assessing whether or not to undertake a primary preventive detection. strategy for diabetes, it is useful to consider the possible costs 568 J. PatTck Vaughan, Lucy Gilson, and Anne Mill and consequences of such a program. Both costs and conse- hypertension in a number of industrial countries may suggest quences will be influenced by the nature of the preventive that changes in lifestyle can be effective means of preventing strategy adopted. In particular, costs are influenced by the scale noncommunicable diseases, including diabetes. Yet studies of the educational program (and by the nature of other preven- have not clearly shown that patient knowledge and patient tive activities); the consequences, by the effectiveness of po- behavior in diabetes care are correlated (Marquis and Ware tential educational strategies. 1979). In general, the evidence on the effectiveness of educa- Given the existing inadequacy of the epidemiological un- tional programs remains limited and contradictory, and the derstanding of IDDM, such a preventive strategy is currently costs per case prevented of such programs are at present im- only a possibility for NItDM. Prevention of NIDDM would involve possible to evaluate. mass health education in order to change the behavioral Alternatively, or in conjunction with healthy lifestyle pro- patterns that increase the risk of diabetes. Unfortunately, little motion, educational messages could be targeted at the groups is known about either the costs or the effectiveness of such known to be at high risk of developingNID)DM. Zimmet (1987), education programs. Phillips, Feachem, and Mills (1987) re- however, suggests that a population strategy is preferable for port that the total costs of mass media campaigns have varied primary prevention purposes because a high-risk strategy substantially-from less than $20,000 for a Kenyan childcare would affect only a small proportion of all people who would program to more than $500,000 for programs involving foreign subsequently develop diabetes. Targeting educational mes- expertise, careful audience research, and prime-time broad- sages at specific people might, however, reduce costs and casting (for example, the Tanzanian "Man Is Health" program improve effectiveness (by permitting more precise messages to costs about $600,000). It would appear difficult to justify be delivered). Such a high-risk strategy might also go hand-in- similar programs for diabetes alone, because the relatively low hand with a screening program for these vulnerable groups. incidence results in a small potential target population and so The World Health Organization (WHO 1985) suggests that would generate high costs per capita of this population. The screening programs provide the opportunity for creating public promotion of healthy lifestyles through mass media programs awareness and educating health professionals. Target groups aimed at the entire population would be more justifiable and should include those at high risk of glucose intolerance (for would, in part, seek to prevent the development of diabetes. example, the obese) and those in whom even mild glucose On the basis of Phillips, Feachem, and Mills's hypothetical intolerance might be a risk factor (for example, pregnant costings for five possible education programs (varying from women). The costs and effectiveness of these programs are not cheap to "luxury"), the cost per capita might be between $0.04 currently known but will be influenced by the sensitivity and and $0.96 for a population of 500,000, or between $0.02 and specificity of screening methods, the definition and size of $0.54 for a population of 1 million. target populations, and the level of care provided for those A number of factors clearly influence the effectiveness of found to have diabetes. such programs-in particular, their coverage and the subse- The potential negative consequences of a preventive quent use of their messages by the general or targeted popula- strategy that are suggested in table 22-2 could be forestalled tion. The author of a review of fifteen mass media health and by greater investment in existing health services, but in the nutrition projects in developing countries concluded that al- short term such investment is rarely forthcoming in devel- though mass media programs can quickly reach large numbers oping countries. It is, therefore, important to consider the of people and up to half of those reached by the message total amount of resources that can be harnessed to provide remember it in the short term, there is only limited evidence health care in developing countries, how these resources that people actually adopt new behavioral patterns (Leslie should be allocated among the health needs of a country, 1987). Other studies (for example, Foote 1985 on the promo- and how to allocate responsibilities for the provision of tion oforal rehydration therapy in the Gambia) present a more health care among different providers. This approach will optimistic picture of the influence of broad educational pro- ensure more efficient use of currently available resources grams on some health-related behavioral patterns. The success and will provide the basis for efficient use of investment of programs aimed at the control of cardiovascular disease and funds available in the long term. Table 22-2. Costs and Consequences of Undertaking a Preventive Diabetes Strategy Bearer of costs Examl___ _Emes ____ Consequences Government Fducation program, required strengthening of health Reduction in morbidity and mortality from diabetes and its infrastructure, screening program complications; indirect and direct cosr savings Household Increased visits, more medications, better diet Household savings (for example, reduced time and monetary costs because less treatment and less loss of earnings) Note: In addition, positive spin-offs include reduction in associated diseases, leading to indirect and direct cost savings (such as improved general knowl- edge and behavior). Negative spin-offs incluide lower quantity and quality of care for other conditions because of emphasis on diabetes. Source: Authors. Diabetes 569 Case Management of a urine sample for glucose, or during the investigation of another illness or diabetic complication. Once diagnosed, The case management of IOL)M and NILDDM are very different and most Nl)DMI patients can be stabilized as outpatients and do not are, therefore, considered separately here. require more than brief treatment with insulin. Their blood glucose or urine glucose needs to be monitored less frequently Insulin-Dependent Patients than that of 1DnL* patients, and they often respond well, at least initially, to a modified diet, weight reduction, and increased Case management varies between iDDM and NIDLDI. For ILuDM, exercise and physical activity. the main requirements are the establishment of the diagnosis, Only if these methods fail are drugs that lower oral blood stahilization of patients on daily insulin therapy, and the glucose (hypoglycemic drugs) required. Such drugs have been training of these patients to inject their own insulin and to available for the treatment of NIDDM for nearly thirty years. monitor their own control on a regular basis. Patients using Data from a twenty-two-year analysis (1964-86) of the use of insulin must also lower their dietary carbohydrates and sugar such drugs in the United States showed that chlorpropamide intake, eat regular meals, and increase their physical activity. (a sulfonylurea) was then the most widely used, two new Good control of diabetes requires the regular monitoring of preparations introduced in 1984 gaining 41 percent of the blood glucose or urinary glucose levels, or both, and the early market by 1986. Patients age sixty years and older received oral detection of the long-term clinical complications. Because of hypoglycemic drug prescriptions at the rate of 478 per 1,000 the high incidence of diabetic complications, it is essential to visits in 1986, and 35 percent of all diabetic patients were maintain good case follow-up and monitoring procedures for taking suchdrugs (Kennedy and others 1988). The inadequacy all patients with diabetes. This follow-up can be the responsi- of primary-level services, poor dietary and general advice, and bility of primary-level care but shared appropriately with sec- the lack of patient supervision in developing countries is likely ondary and tertiary levels. to result in greater use of these drugs. For instance, evidence It is important to point out that insulin has been available from a number of Pacific countries suggested that about 80 for the treatment of diabetes for more than fifty years and that percent of NrrIDI patients were using oral hypoglycemic drugs it is a highly effective treatment in saving and prolonging the (South Pacific Commission 1978). If oral hypoglycemic drugs lives of patients with IMM. It is ironical, therefore, that the fail to control the diabetes, insulin is required and is costly for control of IDD&1 can be so difficult for patients and health the health services and for patients. In industrial countries workers in many developing countries (Serantes 1985). In the number of NJDDM patients who finally require insulin ther- particular, the regular supply of a suitable insulin preparation apy may be more than the number of IDMM patients regularly and appropriate syringes may be difficult to ensure, despite using insulin. their obvious importance to survival (WHO 1985). Most of the It appears, therefore, that providing adequate case manage- world's insulin is produced by a few manufacturers based in ment in developing countries is a much more feasible option industrial countries, and although there is a move to standard- for NIDNM than IDDM patients, particularly if planned primary- ize both the insulin (to 100 international units per milliliter) level health care strategies are adopted. It is important to note, and the syringe, there are many different strengths and however, that even though patients are classified as IDDM and types available (Bloom 1985; WHO 1985). In addition, insulin NIDL)M, insulin is important in the treatment of both groups. is a biological product with a limited storage life that re- quires appropriate cold-chain conditions. It is also an expen- Costs and Consequences of Diabetes Case Management sive drug, which is not available through the United Nations Children's Fund (UNIPAC) in Copenhagen, although it is in- The case management strategies presented in the previous two cluded in the World Health Organization's list of essential sections have implications for the care ideally available at all drugs (WHO 1988). Because insulin availability is such a prob- levels of the health care system. The World Health Organiza- lem in many developing countries, particularly in Africa, the tion (1985) recommends that at the primary level the compo- International Diabetes Federation has been organizing the nents of diabetes care offered should include self-care, home collection and intemational transportation of unwanted insu- care, basic care, screening for complications, and health edu- lin vials (IDF 1987). cation (WHCO 1985). Support for families should be provided by a primary care physician, a nurse, and other health profession- Non-Insulin-Dependent Patients als. Health workers must know the diagnostic, therapeutic, and preventive aspects of care. A list of essential items required for People with NIDDM suffer from a less severe illness day to day ooivi and NIDOM management, as recommended by the World and so, if the necessary equipment and specialist advice is Health Organization, is shown in table 22A-3. Referral from available, can largely be cared for by general medical practi- the primary level to the secondary or tertiary level will be tioners and trained nurses in industrial countries (Howe and necessary when specialized assistance is required in the man- Walford 1984; Burrows and others 1987). Because of the agementofthediseaseoritscomplications.Laboratoryservices insidious onset of NIDDM, patients are commonly diagnosed will also be needed. At the tertiary level special clinics should only incidentally by screening procedures, such as the testing be organized to provide diagnostic and management skills for 570 J. Patrick V aughan, Lucy Gilson, and Anne Mills Table 22-3. Costs and Consequences of Diabetes Case Management Strategies Bearer of costs Examples Consequences Government Case management (diagnosis, monitoring, treatment), Redtuction in morbidity and mortality from diabetes and its patient education complications as result of improved treatment; indirect and direct cost savings Household Increased visits, more medications, better diet Household savings of time and money from improved treatment Note: In addition, positive spin-offs include strengthened health services. Negative spin-offs include clinical side effects (for example, AIDS and hepatitis, ef- fects of inappropriate use of hypoglycemic drugs and insulin), impact on existing heaith services. Soutce; Authors. the treatment of diabetic retinopathy, end-stage renal disease, 1983) and a 75 percent decrease in below-knee amputations and vascular disease. (Assal and others 1982). In contrast, the long-term (more than Case management must include a patient-oriented educa- one year) effect of nutrition education on weight loss was tional strategy, focusing on face-to-face education of known disappointing (Foreyt and others 1981; Wing and others 1985). diabetic patients concerning the dietary and behavioral Studies show that education provided in the outpatient setting changes necessary to maintain optimal metabolic control and can be effective for diabetes (for example, intensive outpatient to prevent and reduce the severity of diabetic complications. education was associated with lowered plasma glucose levels As with preventive strategies, it is useful in decisionmaking [Mazzuca and others 19861), but it is also often undermined by to outline the costs and consequences of case management attrition rates. A review of such studies showed that in the five strategies (see table 22-3). Given the potentially large costs in which attrition rates were reported, up to 90 percent of the and unknown effectiveness of preventive strategies, the first patients failed to complete the educational program (Kaplan option should be for improved case management. Developing and Davis 1986). This emphasizes the importance of motiva- countries will need to improve case management to ensure tion and of linking education to follow-up visits. appropriate and cost-effective care and to reduce the likeli- The key resource required for such an education program is hood of acute and chronic complications. appropriately trained personnel, although literature, equip- In broad terms, diabetes management aims to preserve the ment, and facilities are also necessary (WHO 1985). Education life of the diabetic patient, to relieve the symptoms of the can take place in the hospital or in the outpatient setting; disease, and to avoid its associated complications. The studies unfortunately there is little cost data for either strategy. The of direct costs indicate that there is substantial variation be- authors of a study in Australia in which the outpatient initia- tween countries in the treatment patterns adopted and that tion of insulin therapy was assessed showed that this strategy these differences also influence costs. The technology used is is feasible where the facilities for education about diabetes an especially important influence, and the trend toward more exist; they also showed that it is safe, achieves satisfactory expensive care within industrial countries is one that develop- metabolic control, is acceptable to most patients, and, com- ing countries can ill afford. Developing countries must assess pared with inpatient care, reduces costs by Aus$1,857 per new the cost-effectiveness of standard case management practices. patient stabilized (Bruce and others 1987). For example, the authors of a study in the United Kingdom In order to estimate the annual costs of treating IDDM and stress the need for a structured clinic recall system for diabetics NIDDM patients in a low-income developing country, data from in order to improve clinical surveillance (Burrows and others Malawi on inpatient and outpatient costs and from interna- 1987). There is, however, considerable uncertainty about ap- tional essential drugs lists (WHo and the nongovemment propriate practices-some patients are monitored frequently Dutch organization [IDA]) were used. In the absence of devel- and others infrequently. In another study in the United King- oping country data on hospitalization rates for diabetes pa- dom, Jones and Hedley (1986) showed that if follow-up times tients, the analysis used estimates based on data from the were increased by 30 percent (for example, from six to eight United States. Despite the limitations of the available data, it months) an additional 2,000 known nonattenders could be is clear that the cost of treating IDDM patients will be dominated seen for a cost increase of less than 5 percent of the existing by their need for insulin and the equipment for its administra- annual cost. Recommended practices must aim to be as cost- tion, at an estimated cost of $191 per diabetic per year (1987 effective as possible in order to ensure that best use is made of prices), approximately 90 percent of total costs. The find- limited available resources. ing reflects the high cost of insulin in developing countries More is known about the effectiveness of patient education in relation to that of hospitalization, and the relatively low as a part of case management than the effectiveness of other probability of hospitalization. It is more difficult to estimate educational strategies, but the evidence is not conclusive. In the costs associated with NIDDM patients, but it seems likely onehospital,patienteducationledtoareductioninoccupancy that the biggest cost is for oral hypoglycemic drugs at an by diabetic patients from 5.6 days per year to 1.4 days (Miller estimated $20 per diabetic per year. If we assume no hospital- and Goldstein 1972). Similarly, patient education in self-care izations, this represents nearly 90 percent of the total cost per led to a 78 percent decrease in hyperglycemic coma (Davidson patient. Diabetes 571 It is possible that as case management improves and case Priorities fatality is reduced, the cost of hospitalizations for complica- tions will rise and eventually dominate the total cost of insti- In any consideration of future priorities, there is an inevitable tutional care for diabetic patients. These rising treatment costs trade-off between investing resources in primary prevention theoretically increase the potential benefits (with respect to and in better management of patients. In addition, there are cost savings) of primary and secondary preventive strategies major equity and ethical questions to be considered. and also stress the importance of improved case management of the complications of diabetes to reduction of the direct costs Priorities for Resource Allocation associated with them. In practice, diabetes case management (and preventive) Resource allocation plans need to take into account questions strategies must balance what is feasible in each developing of public health policy and equity. country situation against the potential benefits. Feasibility is related to the cost of the strategy, to the coverage and quality PUBLIC HIEALTH CONSIDERATIONS. The current evidence of care achieved, to the availability of trained health workers about the public health significance of diabetes worldwide is in existing health systems, to the pool of available health limited. It would seem that IDDM is numerically the lesser resources, and to the range of other compelling health needs problem, but its recognition and management requires the use that exist. For example, although recall systems are a relatively of relatively specialized and more costly techniques. By con- cheap and very cost-effective method of protecting the health trast, NIDDM is relatively more common and with changing of diabetics, existing recall systems (for example, those for conditions is of growing significance, particularly in the Pa- tuberculosis patients) often do not function effectively: health cific, China, Asia, and the Middle East. This is probably also staff have too many other "priority activities" to give adequate true for Latin American and African countries, although there attention to patient monitoring, and patients either do not is less clear evidence for these two continents. In the United understand the benefits of monitoring systems or judge that States more deaths are attributed to diabetes than to lung the costs of regular check-ups outweigh the potential benefits. cancer, breast cancer, motor vehicle accidents, cirrhosis of the In addition, record systems are frequently poorly maintained liver, or infant mortality (WHO 1985). Improved case recogni- in health facilities, and health cards are not always kept by tion and management is, therefore, important and can be patients. Remedying this situation requires patient and pro- justified for industrial countries simply on the grounds of its vider education and additional resources (additional health or potential to reduce both the future indirect (morbidity and clerical personnel, card index systems, and so on). mortality) costs of the complications associated with the dis- The broad case management strategy presented here must ease and the future direct costs of caring for patients suffering therefore be adapted to the country-specific situation through with such complications. It is also probable that emphasizing consideration of the needs ofexisting diabetic patients and the the preventive strategy could reduce the costs of the disease level of available resources. Important issues will include and its complications. Given this situation and the changing whether basic care can be offered at health centers or only at health pattems of developing countries, it seems clear that the local hospitals on an outpatient basis and whether drugs or burden of diabetes and the predictable potential costs of the insulin can be made widely available, possibly dispensed disease for developing countries cannot be ignored as they through the private sector. A growing trend is to recommend undergo development. district diabetes centers, which can offer routine care and The position faced by developing countries conceming the education and ensure the adequacy of supplies and drugs. broad policy options of primary prevention and case manage- In many developing countries, a diabetes case management ment is as follows: strategy will be based on primary care facilities, the regular supply of essential drugs and equipment, and the setting of * In general, primary prevention has the potential to re- appropriate clinical priorities, such as standard diagnosis and duce both the indirect and direct costs of diabetes but only treatment protocols. After these have been achieved, consid- at substantial expense eration should then be given to the additional resources needed A Primary prevention is not, currently, a realistic option for the development of secondary and tertiary levels of care. for IDDM The cost of treating an IDDM patient can be used to estimate * Primary prevention has more potential for NIDDM, but the cost per disability-adjusted life-year gained by a case man- prevention would need to rely on interventions and educa- agement strategy. A similar estimate is not made for NIDDM tional programs of unknown effectiveness patients because of the uncertainty about the number of years of life saved by treatment. * In general, case management has the possibility of reduc- Treatment of an IDDM patient is lifesaving; thus each yearns ing the potentially large indirect and direct costs of the treatment saves a year of life. There is marginal reduction in complications of diabetes, except in its most developed treamen saes yer oflif. Tereis argial edutio in forms, when medical care may be beyond the level afford- the quality of life because of the inconvenience of treatment able by many developing countries and lifestyle limitations. Assuming that 0.9of a disability-adjusted year of life is obtained for the $213 annual cost of treatment, * Only a small number of IDDM patients will require case then the cost per disability-adjusted life-year gained is $237. management, but it will be potentially expensive because 572 J. Patrick Vaughan, Lucy Gilsoni, and Anne Mills such patients need regular insulin throughout their lives and country-specific situation must obviously be assessed before most patients eventually suffer from at least one complica- the relative status of diabetics can be judged. tion. Still, because treatment postpones death for many What would these characterizations suggest about the effect years, it may well be cost-effective in comparison with on equity of diabetes prevention and case management pro- interventions for some other chronic diseases grams? One argument might be that late-onset NIDDM patients * Case management is very important for NIDDM patients are not only more wealthy but probably in the productive years and should be based on patient education, behavior modifi- of life, possibly in responsible and influential jobs, and that cation, and appropriate use of oral hypoglycemic drugs. they are an important economic asset which must be protected. It can also be argued that the existing bias of many developing The diversity of manifestations of diabetes makes it difficult country health systems is already set against lower socioeco- to suggest universally appropriate policy strategies. Each coun- nomic groups and in favor of wealthier patients and further try needs to determine for itself how it will tackle the potential state expenditure on upper-income groups cannot be justified. problems of the disease, within the primary care framework This latter argument does not imply that little or no provi- proposed by wHo (1985). In table 22-4 we summarize some policy sion should be made for diabetes but rather that it cannot be options for discussion and indicate that the options differ viewed as a priority for state health services alone. It might be between developing countries as a result of the differing in- possible to consider, for example, the provision of drugs and come levels, the differing health infrastructure, and the differ- insulin on a means-tested basis in the public sector or on a ing relative significance of diabetes as a health problem. Within fee-for-service basis in the private sector. Such a strategy has policies, the differences between IDDM and Ni[)DM must also be many practical difficulties, however, and the possible benefits recognized. In the table we assume the incidence of both forms may exceed the costs of implementation. For example, it would ofdiabetes taken together is either high or low. The distributional promote the development of dual health standards: high- issues discussed next must also be considered and should feed quality private care for the more wealthy diabetics and low- into policy discussions about how to finance diabetes case quality or inadequate public care for the less wealthy. This management activities and what levels ofcare can be afforded. strategy would also probably conflict with ethical considera- tions. Since insulin has been available for more than fifty years EQUITY CONSIDERATIONS. Considerations of distribution and and it is an essential, life-sustaining drug for IDDM1 patients, equity can have an important influence on investment priori- many people would consider it unethical to adopt a health ties. Discussion of the equity implications of diabetes preven- policy which fails to guarantee adequate supplies of insulin. tion and case management programs must be based on who This is the implicit policy in many countries, however; because suffers from the disease and therefore who will gain from they cannot afford to provide the drug, the way is left open to prevention and treatment, as well as on the sources of finance the private sector to supply unregulated care for those who can for the programs. afford to pay. Inequities already exist, and countries must Non-insulin-dependent diabetes mellitus is often character- consider the effect on them of possible policy changes. ized as a disease of the rich, because it is clearly associated with Resolution of distributional problems is not easy, especially environmental changes reflecting increased wealth. Within when resources are scarce and policy choices often have developing countries, for example, the urbanized population is unwanted consequences. Future health care investments in generally deemed to be more affluent than the rural popula- developing countries, however, must seek to reflect both ex- tion and may be more at risk of developing NIDDM. Insulin- isting resource constraints and such distributional concerns. dependent diabetes mellitus, however, is less clearly associated The production and distribution of insulin as a life-preserving with such environmental factors, and so patients suffering from drug is an important part of the broader considerations involv- this type of diabetes may be of high or low income. The ing its usage, such as patient monitoring and health education. Table 22-4. Policy Options for Control of Diabetes Mellitus Incidence of DMI Higher-incotne developing countries Lower-income developing countries Low Focus on primary and secondary care only, especially Do nothing face-to-face education of known diabetics Consider minimum case management requirements Develop standard case management protocols for IDDM and NIDDM Have appropriate drugs available High Consider tertiary provision Focus on primary and secondary care only, especially Screen high-risk patients for NIDMM Strengthen preventive and case management efforts Face-to-face education of known diabetics (for example, through broader education programs) Develo.p standard protocols Secure regular supplies of insulin and oral hypoglycemic Make insulin and oral hypoglycemic drugs available drugs, and ensure delivery to patients on demand only Stimulate epidemiologic, clinical, and laboratory research Form links with international research efforts Source: Authors. Diabetes 573 There is a clear need to bring together the interested parties Sub-Saharan Africa, and Latin America, through a greater worldwide, including the manufacturers, the Intemational support to epidemiological studies Diabetes Federation, the World Health Organization, and * Large-scale evaluation studies of possible interventions other international agencies, to develop more effective and and strategies to prevent NIDDM, including studies of the cost acceptable strategies for developing countries, and effectiveness of a broad strategy of noncommunicable disease control Priorities for Operationlal Research * Assessment of the costs and effectiveness of alternative Specific intemational actions can assist the development of case management procedures (IDDM/NIDDM) appropriate strategies, and the International Diabetes Federa- * Case studies to assess what resources are currently con- tion and the World Health Organization are active in these sumed by IDDM and NIDDM areas (King and Mitrofanov 1988). For example, there is an * Operational research to improve the quality of case man- urgent need for operational research to find ways of facilitating agement within existing health services, such as the develop- the purchase of insulin and oral hypoglycemics at reasonable ment of standard treatment protocols for primary care prices (for example, through UNIPAC) and to guarantee their * Consideration of appropriate financing mechanisms availability in developing countries in order to improve case for expanding the management of diabetes patients, management strategies and to reduce the cost per case. More- particularly within secondary and tertiary health care over, international support for diabetes research is necessary, facilities. and more international collaborative research centers are needed in the developing world. Such research should aim to clarify the importance of diabetes and the options available for Appendix 22A. Diagnosis and Self-Care of Diabetes its prevention and case management. The following are the preliminary research priorities that we have identified in this The tables that follow show the blood glucose levels of both chapter for national and international action: diabetics and nondiabetics and the glucose concentrations * Assessment of the incidence and prevalence of both iDDM they produce during the glucose tolerance test, as well as the and NIDDM in developing countries, particularly in Asia, equipment essential to self-care for diabetics. Table 22A- 1. Blood Glucose Levels for Diagnosis of Diabetes Mellitus DM likely DM uncertan DM unlikely Sample mmol/l mg/dl mmol/l mg/dl mmol/l mg/dl Whole blood Venous > 10.0 > 180 4.4-10.0 80-180 < 4.4 < 80 Capillary > 11.1 > 200 4.4-11.1 80-200 < 4.4 < 80 Plasma Venous > 11.1 >200 5.5-11.1 100-200 < 5.5 < 100 Capillary > 12.2 > 220 5.5-12.2 100-220 < 5.5 < 100 Note: Unstandardized (casual, random) blood glucose values. Source: WHO 1985. Table 22A-2. Diagnostic Glucose Concentration Values in the Oral Glucose Tolerance Test [mmol/l (mg/dl)] Diabetes mellitus Impaired glucose tolerance Sample Fasting Two hours after glucose load Fasting Two hours after glucose load Whole blood Venous 6.7 (120) 10.0 (180) < 6.7 ( <120) 6.7-10.0 (120-180) Capillary 2 6.7 (120) 11.1 (200) < 6.7 (< 120) 7.8-11.1 (140-200) Plasma Venous 2 7.8 (140) 11.1 (200) < 7.8 (< 140) 7.8-11.1 (140-200) Capillary 2 7.8 (140) 2 12.2 (200) < 7.8 (< 140) 8.9-12.2 (160-220) Note: For epidemiologic or population screening purposes, the two-hour value after 75g oral glucose may be used alone or with the fasting value. The fasting value alone is considered less reliable because true fasting cannot be assured. Source: WHO 1985. 574 J. Patrick Vaughan, Luc-y Gilson, and Anne Mills Table 22A-3. Basic Equipment for Self-Care of Diabetics Self-management of IDDM Self-nmanagement of NIDDM Prinmary health care center Urine testing materials for glucose and Urine testing materials for glucose and UIrine testing materials for glucose and ketone bodies and/or blood glucose ketone bodies and/or blood glucose testing ketone bodies and/or blood glucose testing materials materials testing materials Book or chart and pencil for recording Book or chart and pencil for recording test Book or chart and pencil for recording results results and body weight results Insulin as prescribed and cool place for Oral hypoglycemic agents, when applicable Insulin, plus cool place for storage storage Sugar lumps or other readily absorbed Syringe and needles Syringe, needles, and carrying case carbohydrates Sterilization facilites Sterilization facilities Cotton wool Cotton wool Cleansing agent Sugar lumps or readily absorbed Sugar lumps or readily absorbed carbo- carbohydrates hydrates Oral hypoglycemic agents Materials for testing the presence of protein in urine Weighinig machine Blood glucose monitors or meters and test- strips Glucose for intravenous use (glucagon. if available) Simple printed education materials and teaching aids Place for storing patients' records Source: WHO 1985. Notes Bmuce, D. G., E. M. Clark, G. A. Danesi, C. V. Campbell, and D. J. Chisholm. 1987. "Outpatient Initiation of Insulin Therapy in Patients with Diabetes This chapter has benetited greatly fronm all the most helpful comments and Mellirus.".Meducal]oumalof Azstralia 146:19 22. criticisms that we received on early drafts, but we would particuilarly like to Burrows, P. J., P. . Gray, A.-L. Kinmouth, D. J. Payton, 0. A. Walpole, R. 3. thank the following: Dr. H. Keen, Guy's Hospital, London; Dr. H. King, World Walton, D. Wilson, and G. Woodbine. 1987. "Who Cares for the Patient Flealth Organization, Geneva; Dean Jamison, University of Califomia at Los with Diabetes? Presentation and Follow-Up in Seven Southampton Prac- Angeles; Dr. R. E. LaPorte and Dr. T. J. Songer, Diabetes Epidemiological tices." Journal of the Royal College of General Practinoners 37:65-69. Research Intemational, Pittsburgh; Dr. R. Williams. Departrnent of Commu- Cameron, W. t., P. S. Moffitt, and D. R. R. Williams. 1986. "Diabetes Mellitus nity Medicine, Cambridge; and Dr. P. Zinsmet, Lions International Diabetes in the Australian Aborigines of Bourke, New South Wales." Diabetes Institute, Melboume. In addition, our colleagues within the Department of Research and Clinical Pracnce 2:307-14. Public Health and Policy at the London School of Hygiene and Tropical Cheah, J. S., and B. Y. Tan. 1979. "Diabetes amongst Different Races in a Medicine gave us many new insights and happily commented on our ideas. Smilar Environment." In l w Waldhouse, ed., Diabetes Amsterdam I. All dollar amounts are current U.S. dollars unless otherwise indicated. Excerpta Medica. Davidson.). K. 1983. "The Grady Memorial HIospical Diabetes Programme." In J. Mann, K. Pyorala, and A. Teuscher, eds., Diabetes in Epidemiological References Perspective. London: Churchill Livingstone. Diabetes Epidemiological Researcl International (DERi). 1987. "Preventing Abu-Bakare, A.. R. Taylor, G. V. Gill, and K. G. M. M. Alberti. 1986. Insulin Dependent Diabetes Mellitus: The Environmental Challenge." "Tropical or Malnutrition-Related Diabetes: A Real Svndrome?" Lancet BrtishMedical]oumal 295:479-81. 1:1135-38. - 1988. "Geographic Pattems ofChildhood Insulin-Dependent Diabe- Ahren, B., and C. B. Corrigan. 1985. "Prevalence of Diabetes Mellitus in tes MelliLus." Diabetes 37:1113-19. North-Western Tanzania." Diabetologia 26:333-36. -.1990. "SecularTrends in Incidence of lOMi in 10 Countries." Diabetes American Diabetes Association. 1988. Direct and Indirect CGsts of Diabetes in 39:858-64. the United States in 1987. Alexandria, Va. Fatani, H. H., S. A. Mira, and A. G. El-Zubier. 1987. "Prevalence of Diabetes Assal,J. P.,R.Gseller,andJ.-M.Ekoe. 1982. "Patient Education in Diabetes." Mellitus in Rural SaudLi Arabia." DiabetesCare 10:180-83. In H. Bostrom, ed., Recent Tretds in Diabetes Research. Stockholm: Almqvst Foote, D. R. 1985. The Mass Media and Health Pracnces Evaluation in The and Wicksell. Gambia: A ReporL of Ma;or Findings. Stanford University, Stanford, Calif. Barratt-Connor, E. 1985. "Is lnsulin Dependent Diabetes Mellitus Cauised by Foreyt, J. P., D. G. Goodrick, and A. M. Gotto. 1981. "Limitations of Coxsackie Virus B Infection'A Review of the Epidemiological Evidence." Behavioural Treatments of Obesity: Review and Analysis." Journal of Reviews of Infectious Diseases 7:207-15. Behatioural Medicine 4:159-74. Bennett, P. H. 1985. "Changing Concepts of the Epidemiology of Insulin Fuller, J. H., J. Elford, P. Goldblatt, and A. Adelstein. 1983. "Diabetes Dependent Diabetes" Diabetes Care 8:29-33. Mortality: New Light on an Underestimated Public Health Problem." Bloom. A. 1985. "Syringes for Diabetics." British Medicaljournial 290:727-28. Diabetologia 24:336-41. Diabetes 575 Gambert, S., N. Fox, and J. Jacobs. 1988 "Oral Hypogivcaemic Therapy and Marquis, K., and J. E. Ware. 1979. New Measitres of Diabenc Patient Knowledge, Rates of Health Care Utilisatioi in Tvpe 2 Diabetes (NIODMA)." In Direct and Behaviou, and Attitude. Santa Monica, Calif.: Rand. Indirect Costs of Diabetes in the United States in 1987 American Diabetes Mather, H. M., and H . Keen. 1985 "The Southall Diabetes Survey: Prevalence Association Inc. Alexandria, Va. of Known Diabetes in Asians and Europeans." Bnntsh Medical Journal Gamble, D. R. 1980. "The Epideemiology of Insulin Dependant Diabetes, with 291:1081-84. Particular Reference to the Relationship of Virus Infection to Its Etiology." Mahtler, H. M., N. P. S. Verna, S. P. Mehta, S. Madhu, and H. Keen. 1987. Epidemiologic Re'teus 2:49-70. "The Prevalence of Known Diabetes in New Delhi anid London."Jourtiahlo Ghana Health Assessment Project Tvarnt. 1981. "A Quantitative Methosd .-f the Medical.Associaton of Thailand 70(supplement 2):54-58. Assessimg the Health Impact of Different Diseases in Less Developed Ma-zuca.S. A., N. H. Moorman, M. L. Wheeler J. A. Norton, N.S. Fineberg, Countries." Inreratiowul jourral of Epidemiolog! 10:73-80. F. Vinicir, S. J. Cohen, and C. M. Clark, Jr. 1986. "The Diabetes Education "GIuicose Tolerance in Pregnancy-The Who and How of Testing." 1988. Studv: A Controlled Trial of the Effects of Diabetes Education." Diabetes LUncet (editorial) 2:1173. Care 9:1-10. Godhes, D. M 1986. "Diabetes in American Indians: A Growing Problem." Miller, L. V.. and G. Goldstein. 1972. "More Efficient Care of Diabetic Diabetes Care 9:609-13. Patients in a Country-Hospital Setting." Netu England Medical Journal Grabauskas, V. J. 1988. "Glucose Intolerance as a Contributor to Non-coim- 286:1388-94. municdble Disease Morbidity and Mortality." Diabetes Care 11:253-57. Papoz, L, F. Ben Khalifa, E. Eschwege, and H. Ben Ayed. 1988. "Diabetes Hnwe. P., and S. Walford. 1984. "Diaberes Care. Whose ResponsibiIiti?" Mellitus in Tunisla: Description in Urban and Rural Populations." Interna- Bnrnsh Medical]ournal 289:713- 14. tona JoumalofEpidemioig- 17:419-22. IDF (International Diabetes Federation). 1987. "Finland Begins Insulin-Redis- Patterson, C. C., NI. Thorogood, P. G. Smith, M. A. Heasman, J. A. Clarke, tribLItion Programi." Jnrernatioril Diabetes Federatiotii Bulletin 32 193. and 1. I. Mann. 1983. "Epidemiology of Type I (Insuilin-Dependent) Dia- betes in Scotlan1d, 1968-1976: Ev idence of an Increasing Incidence." Jones, R. B., and A. J. Hedley. 1986. "Adjusting Follow-Up Intervals in a bDiaetologia 74 2384s3. Diabetic Clinic- Implications for Costs and Quality of ('are." lournal of the Royal College on Physicians of London 20: 36-39. Phillips, M. \R. R.Feachem. and A. Mills. 1987.Opnonsfor Dtarrhoeal Diseases ContrTl. The Cost and Effectiveness of Selected Interve7tions for the Prevention Jonssen, B. 1983. "Diabetes-The Cost of ltlniess and Control: An Estimate of Diarrhoea. Evaluation and Planning Centre. Publication 13. London for Sweden, 1978." Acta Medica Scandmavica 671 isupplement): 19-27. School f Hygiene and Tropical Medicine. Kaplan, R. MN., and W. K. Davis. 1986. "Evaluating the CiOsts and Benefits of Poon-King, T., M. V. Henrv, and F. Rampersad. 1968. "Prevalence and Outpatient Dia.hetes Education and Nutrition Counselling." Diabetes Care NatLral History ofDiabetes irn Trinidad." Lancer 1:155-60. 9:81-86. Reunanen, A., H. K. Akerblom, and M. L. Kaar. 1982. "Prevalence and Ten Kennedy, D. L., J. NM. Piper, and C. Bawm. 1988. "Trends in the Use of Oral Year (1970-1979) Incidence of Instilin Dependent Diabetes Mellitus in Hypoglycaemic Drugs, 1964-1986." Diabetes Care 11:558-62 . ChdIdren and Adolescents in Finland." Acta Paediamca Scandinavia 71: King, H. 1987. "Preventing Insu1lin Dependent Diabetes Mellitus: The Envi- 891-99. ronmental Challenge." British MedicalJouiral 295:92 3. Rewers, M., R. E. LaPorte, Fl. King, andJ. Tuomilehto. 1988. "Trends in the King, H., and MN. P. N'litrofanov. 1988. "World Health Organiation Activities Prevalence and Incidence of Diabetes: Insulin-Depenident Diabetes Melli- in the Field of Diabetes Mellitus." World Health Statistical Quarterl' tus in Childlhood." Wnrld Health Statistical Quarterly 41:179-89. 41:197-99. Rewers, MN.. R. E. LaPorte, M. Walczak. K. Dmochowski, and E. Bogaczynska. King, H., and P. Ziinmet. 1988. "Trends in the Incidence and Prevalence of 1987."An Apparent'Epidemic'ofYouth Onset, Insulin Dependent Diabe- Diabetes: I 1. Non-insulin Dependent Diabetes MellitUs." World Health tes Mellitus in Midwestern Poland." Diaberes 36:106-13. Statistical Quarterly 41:190-96. Schooneveldt, M., T. Songer, P. Zimmet, and K. Thoma. 1988. "Changing King, H., P. Zimmet, K. Pargeter, L. R. Raper, and V. Collins. 1984. "Ethnic Mortality Patterms in Nauruans: An Example of Epidemiological Transi- Differences in Susceptibility to Non-insulin Dependent Diabetes: A Csmpar- tion." Journalof Epidemilogy and Community Health 42:89-95. ative Study of Two Urbanised Nitcronesian Populations." Diabetes 33:1002-7. Seneday, M., and M. L. Masti. 1987. "Diabetes in Latin America." Jounial of Krolewski, A. S., J. H. Warrem, 1. R. Lawrence, and C. R. Kahn. 1987. the Medical Associanon ofThaiiland 70(supplement 2):77-78. "Epidemiologic Approach to the Etiology of Type I Diabetes MellitLIs and Serantes, N. A. 1985. "The Problem of the Diabetic Patient in Developing Its Complications." New EnglandoJusuial of Medicine 317:1 390-98. Countries." Diabetologia 28:597-601. Kuberski, T., and P. Bennet. 1979.The Status ofDiabetes Mellitus in the Territory Shanghai Diabetes Research Cooperation Group. 1980. "Diabetes Mellitus of Guam. South Pacific Commission Information Document 47. Sura, Fiji. Stirvey in Shanghai." Chinese MedicalJournal 93:663-70. Laung, W. 1981. "The Cost of Diet-Related Disease." In M. Turner, ed., Simmons, D., D. R. R. Williams, and M. J. Powell. 1989. "Prevalence of Pretentative Nutrition and Society. London: Academic Press. Diabetes in a Predominantly Asian Community: Preliminary Findings of Leslie, J. 1987. "The Use of Health Statistics as an Effectiveness Measure in the Coventry Diabetes Study." Bntsh MedicalJournal 298:18--21. Certai Development Communications ProJects in UNESCO." In The Eco- Songer, T. J. In press. "The Economics of Diabetes Care." In K. G. G. M. nomnics of New Educaioti iMedia. Vol. 2, Costs and Effectiveness. P'aris: Alberti, R. A. de Frunzo, H. Keene, and P. Zimmet, eds., International UtNISCO. Textbook of Diabetes Mellitus. London: John Wiley and Sons. LutaLo, S. K., and N. Mabonga. 1985. "Some Clinical and Epidemiological South Pacific Commission. 1978. Diabetes, Gout and Hypertension in the Pacific Aspects of Diabetes Mellinrs on an Endemic Register in Zimbabwe." East Istanids. Information DocLument 43. Sura, Fiji. Afncan MedicalJourial 26:433-45. Tarn, A. C., . M. Thomas, B. NI. Dean, D. Ingram, G. Schwar., G. F. Bottazzo, tucLarty, D. G., H. Ni. Kitange, B. L. M tinangi, W. J. Niakene, A. B. NI. Swat, and E. A. M. Gale. 1988. "Predicting Insulin-Dependent Diabetes." Lancet G. Masuki, P. M. Kilima, L. M. Chawa, and K. G. M. Ni. Alberti. 1989. 1:845-50. "Prevalence of Diabetes and Impaired Glucose Tslerance in Rulral Tanza- Teuscher, T., P. 13ailol, J. B. Rosman, and A. Teuscher. 1987. "Absence of nia." Lincet 1:871-75. Diabetes in a Rural West African Population with a High Carbohy- Nirine, N., 0. Edelstein, W. P. W. )ackson, and A. 1. Yinik. 1969. IDiabetes drate/Cassava Diet." Lcincet 1:765-68. HvpergIvcaemic and Glycosuria among Indians, Malays, and Africans Thai, A. C., P. Yeo, K. Lun, K. Hughes, K. Waugh, S. Sorhy, K. Liini, W. Ng. (Bantu) in Cape Town, South Africa." Diabetes 18:433-45. J. Cheah, W. Phoon, and P. Lim. 1987. "Changing Prevalence of Diabetes 576 J. Patnck Vaughan, Lucy (ilson, and Anne MilLs Mellitus in Singap(ore over a Ten Year Period." Journal of the Medical Williams. D. R. R. 1985. "Hospital Admissions of Diabetic Patients: Informa- Association of Thailand 70(supplement 2):63-67. tion from Hospital Activity Analysis." Diabetic Medicine 2:27-32. Tong-yuan, Tai, Chih-liang Yang, Chih-jen C(gang, Shu-tneilha, Yuh-huev Wing, R. R., L. H. Epstein, M. P. Norwalk, R. Koeske, and S. Hagg. 1985. C'hen, Boniface Juisiang Ln, Kiang-Shi Kt. Muy-Shy Chen, and Chien-Jen "Behaviour Change, Weight Loss, and Physiological Improvements in Type Chen. 1987. "Epidemiology of Diabetes Mellitus in Taiwan, R.O.C.-Com- 2 Diabetic Patients." Journal of Consultant Clincal Psychologists 53:111-22. parison between U rban and Ruiral Areas."Journal oJ- ihe Medical Association Zimner, 1'. 1982. "Type 2 (Non-insulin-dependent) Diabetes-An Epidemi- of Thailand 70(supplement 2):49-53. ological Overview." Diabetologia 22:399-411. Ward., 1. . 1988. "Preventing the Longterm Complications of Diahetes." - 1987. "The Prevention and Control of Diabetes-An Epidemiologi- Proceedings of the Royal College of Physicians of Edinburgh 18:146--53. cal Perspective." Journal of the Medical Associaton of Thailand 70 (supple- WHO (World Health Organization). 1975. Internanonal Statistcal Classification ment 2):30-35. of Diseases. Injunes, and Causes of Death. 9th rev. Geneva. Zimmet, P., H. King, and S. Bjorntorp. 1986. "Obesity, Hypertension, Carbo- - . 1985. Diabetes kMelltus Report of a vsio Study Group. Technical hydrate Disorders, and the Risk of Chronic Diseases: Is There Any Epide- Report Series 727. Geneva. miological Evidence for Integrated Prevention Programmes?" Medical I1988. Essential Drugs List. Geneva. JounalofAustralia 145:256-62. 23 Cardiovascular Disease Thomas A. Pearson, Dean T. Jamison, and Jorge Trejo-Gutierrez Disorders of the circulatory system cover a broad range. We of these more traditional patterns of cardiovascular diseases in focus in this chapter on four chronic vascular diseases for Africa. which atherosclerosis and/or hypertension are the defining characteristics. These are coronary heart disease (CHD), Atherosclerosis and Hypertension strokc, peripheral vascular disease, and hypertensive heart disease. The reasons for this are several. First, these condi- We begin this chapter with a brief discussion of the etiology tions are the most important cardiovascular diseases (CVDs) and pathogenesis of atherosclerotic and hypertensive cardio- in industrial countries, accounting for half of mortality in vascular disease, emphasizing the role of modifiable risk factors North America and Europe. Second, they are already quite as targets for preventive strategies. We then review trends in prevalent in developing countries, contributing approxi- these diseases, making projections into the twenty-first cen- mately 16 percent of deaths (WHO MONICA Project 1989). tury. Finally, we discuss preventive and case management Indeed, it has been observed that the majority of the world's strategies as appropriate to countries with limited health care cases of cardiovascular disease no longer occur in industrial resources. nations but, rather, in developing countries (figure 23-1); demographic analyses, summarized in this chapter (table Etiology and Pathogenesis of Atherosclerosis 23-3), confirm this view. Third, they have some risk factors in common and, thus, share preventive strategies. Indeed, Atherosclerosis is a chronic vascular condition characterized in those industrial countries experiencing declining cardio- by focal accumulations of smooth muscle cells, collagen, and vascular mortality rates, most of the declines in total mor- lipids in medium and large arteries. The condition may begin tality could be attributed to declines in CVD. Fourth, a great as early as childhood, initiated by injury to the endothelial deal is known about the pathogenesis, risk factors, preven- lining of these major arteries, exposing the subintimal smooth tion, and treatment of these conditions, allowing priorities muscle cells and macrophages to serum lipoproteins, platelets, to be set regarding preventive and therapeutic interventions. and other constituents (Ross 1986). These in turn stimulate Several important cardiovascular diseases have not been the proliferation of the smooth muscle cells and the focal included in this chapter. These include rheumatic heart dis- accumulation of lipids. Initially, small, flat fatty streaks are ease, cor pulmonale, congenital heart diseases, and cardio- observed. Previous studies have noted rather high prevalence myopathies. Rheumatic heart disease remains a worldwide of these lesions in autopsystudies even in developing countries problem, especially in developing nations; Chapter 10 of this (McGill 1968). If cellular proliferation continues, there may collection (Rheumatic Heart Disease) provides a review of the be growth into the lumen of the artery, forming the lesion main issues concerning this condition. Cor pulmonale results pathognomonic of atherosclerosis, the fibrous-capped plaque. principally from chronic obstructive pulmonary disease and is, Tracey and Toca (1975) found that, at least in the 1960s, these therefore, discussed in the chapter on that subject by Bumgar- lesions were much less prevalent in developing than industrial ner and Speizer (chapter 24, this collection). Little is known countries. Evidence has long suggested that fatty streaks can aboutthecausesandpreventabilityofcongenitalheartdisease, undergo regression; some recent evidence has suggested that with the exception of rubella-related disease, and, therefore, even fibrous-capped plaques can regress, though this point is these conditions are not dealt with in this collection. Finally, controversial (Blankenhorn and others 1987). cardiomyopathies are regionally, rather than globally, import- ant (for example, Keshan disease in China, Chagas disease in CLINICAL SYNDROMES RELATED TO ATHEROSCLEROSIS. Angina South America, West African cardiomyopathy). These condi- pectoris, myocardial infarction, sudden death, transient isch- tions are beyond the scope of the chapters in this collection. emic attacks, atherothrombotic stroke, aortic aneurysm, and Hutt (1991) provides an overview, with extensive references, intermittent claudication are all related to atherosclerosis and 577 578 Thomas A. Pearson, Dean T. Jamison, and orge Trejo-Gtnerre: Figure 23-1. Estimated Distribution of Causes of Death, 1980 World Total Perinatal Accidents diseases / 6 percent 5 percent \ All other dIseases 24 percenit diseaspss - ' 1 ~~~~~~PCVD / ~sl04pis*10ry |23 perent/ Developing Countries Industrial Countries Accidents Respiratory diseases 5 percent (including TB) Accidents Cancer 7.5 percent 7 percent 6 percent \ Perinatal diseases e- _,p All other 7 percent f diseases . - 27 percent Cancer diseases 19 percent 18.5 percent CVD | 16 percent mIMUMM ,t)/ CVD/ \ mr1R / \ 48 percent/ diseases 18 percent TB = Tuberculosis Note: Of the total deaths, 78 percent are in developing countries. Source: Authors' compilation; WHO MONICA Project, 1989. its complications. When the atherosclerotic plaque occludes Hypertension-Related Diseases: Etiology and Pathogenesis the lumen to the extent that blood flow is impaired, any increased demand for blood flow (such as that brought about The end organs affected most by elevated blood pressure in- by exertion) will lead to deficiency of oxygen and nutrients to clude the arteries, the heart, and the kidneys. Hypertension is the organ supplied. When the coronary arteries are narrowed, a risk factor for atherosclerosis. Furthermore, hypertension is chest pain (angina pectoris) occurs; when the peripheral arter- thought to play a significant role in weakening the arterial wall ies are narrowed, intermittent claudication occurs. Plaques in the cerebral circulation, leading to cerebral hemorrhage. may undergo complication when the fibrous cap ulcerates, Though the sequence of events is controversial, cardiac hyper- ruptures, or forms a clot. Subacutely, this can result in unstable trophy is strongly associated with hypertension. Severe cardiac angina in the coronary circulation or transient ischemic at- hypertrophy impairs the heart's ability to fill with blood, re- tacks in the cerebral circulation. Complete occlusion of the sulting in congestive heart failure. The increased myocardial coronary artery with thrombosis then leads to myocardial mass interacts with atherosclerotic disease to predispose to infarction; occlusion or embolism in the cerebral circulation infarction and arrhythmia. The arteries of the kidneys are leads to atherothrombotic stroke. Myocardial oxygen depriva- susceptible to medial and intimal hypertrophy, known as tion from arterial stenosis or occlusion can also cause a variety nephrosclerosts, a common cause of chronic renal failure. of cardiac arrhythmias, including those causing cardiovascular collapse and sudden death. Myocardium damaged by infarc- The Role of Risk Factors tion may be unable to maintain cardiac output, leading to congestive heart failure. Atherosclerotic involvement of the The pathogenesis of vascular disease is summarized in figure aorta can lead to a weakening of the arterial wall, resulting in 23-2 . These so-called risk factors have been associated with the aortic aneurysm and aortic dissection. clinical cardiovascular syndromes, usually in epidemiologic Cardiovascular Disease 579 Figure 23-2. The Pathogenesis of Atherosclerotic As demonstrated in table 23-1, most of the important car- Vascular Diseases diovascular diseases share important risk factors. An example of this is the powerful effect of elevated blood pressure in Normal Atherosclerotic increasing relative risk for both stroke and coronary heart artery plaque * Complications disease, as is illustrated in figure 23-3. (Notice that risk for both conditions rises over the range of even "normal" values of t t diastolic blood pressure [DBPI, suggesting, thereby, the limited usefulness of specific cutoff values to define "hypertension.") Risk Risk Thus, reduction in one or more risk factors may prevent several factor factor cardiovascular diseases. Several general comments are in order. First, the multifactorial nature of the etiology of cardiovascular Source: Authors'compilation. disease makes complex the development and evaluation of preventive strategies; multifactorial interventions may be nec- studies. The other point illustrated in figure 23-2 is that some essary to maximize the effect of preventive efforts. Second, the risk factors act to cause atherosclerosis, whereas others act after relative strength of association differs between diseases such the formation of the atherosclerotic plaques to cause the that control of a risk factor may be more important for the complications (thrombosis, hemorrhage, and so on) which control of one vascular disease than another. This may also be present as clinical syndromes. true for specific cardiovascular diseases within different racial There are several practical implications of these roles for and ethnic groups. For example, the relative risk of hyperten- risk factors. First, risk factors causing the formation of ath- sion as in cases of stroke appeared higher in American blacks erosclerotic plaques would be logical targets for primary than American whites. Third, some of the risk factors are prevention efforts. Those risk factors acting after the forma- modifiable, whereas others are not. The physiological risk tion of atherosclerosis should be modified in secondary factors are only indirectly modifiable, through change in the prevention efforts. Second, it is conceivable that in certain behavioral ones; they thus serve principally as indicators to populations which do not develop atherosclerotic plaques, spur behavior change (or medical intervention) and as mea- certain risk factors may not be as important as in other suresofprogress. Fourth, incontrasttomanyotherconditions, atherosclerosis-prone groups. there appear to be no important environmental or infectious Table 23-1. Association between Risk Factors and Cardiovascular Diseases Degree of association Coronars- heart Atherothrombotic Peripheral vascular Hemorrhagic Hypertensive heart Riskfactor disease stroke disease stroke disease Nonmodifiable Age ++.++ .... .... ... ++++ Male sex .... .... ... +.+++ + Black race + + + ? +++ +++ Family history +... + + + + + + + + Modifiable physiological Elevated LDL(low-densitylipoprotein) ++++ + + +++ - O Decreased HD)L(high-density lipoprotein) +.++ + + + 0 0 Hypertension +++ + + + + +.+.+.. . Diabetes ... +++ +...+ 0 0 Obesity + + + Behavioral Smoking .. ++ .++++ 0 0 Dietary cholesterol and saturated fat + + + ++ - - Salt consumption 0 + 0 + + + + Alcohol consumption - + - + + Sedentary lifestyle + + 0? + Key. + +++ Strong association between disease and risk factor. +++ Moderately strong association between disease and risk factor. ++ Moderate association between disease and risk factor. + Weak association between disease and risk factor. O No association between disease and risk factor. /- Association varies with the level of the risk factor. Unclear association between disease and risk factor. - Inverse association between disease and risk factor. Source: Authors. 580 Thnmas A. 1'earson, Dean T. Jamison. and Jurge Trejo-GunerrcZ Figure 23-3. Diastolic Blood Pressure and the Risk risk factors, with the possible exceptions of water hardness for Stroke and Coronary Heart Disease (WHO 1982, p. 33) and cytomegalovirus infection (Melnick, Adam, and DeBakey, 1990). Fifth, males are at substantially Relative risk of stroke higher risk than females, although to different extents for different diseases. In table 23A- I we summarize MONICA (Mon- 4.00 - itoring Trends and Determinants in Cardiovascular Disease Project) mortality data that suggest a very strong effect for ischemic heart disease and a much weaker one for stroke; 2.00 - indeed, as a proportion of overall mortality, males appear less affected by stroke than females. Although race, sex, and age are obviously unmodifiahle, their importance suggests the need for specific modeling of other risk factor effects for these 1.00 / characteristics. Ti-E ORIGIN OF RISK FACTORS. A useful concept is that of the 0.50 - "proximal" risk factor; that is, the factor which causes the development of the risk factors (table 23-2). Several points deserve emphasis. First, age is related to most risk factors. 0.25 - Second, several risk factors are transmitted genetically, render- ing a subgroup of the population at high risk regardless of their lifestyles. Third, other risk factors may cause the development 0 l8 l l lof certain risk factors (for example, smoking is related to low 76 84 91 98 105 high-density lipoprotein [HDLI cholesterol), and some of the effect of one risk factor may be mediated through another risk Approximate mean usual DBP (mm Hg) factor. Fourth, as reflected in table 2 3-1, a number of the risk factors are physiologic, which, in turn, appear to be related to behavioral or lifestyle factors (for example, smoking and sed- entary lifestyle are related to low HDL cholesterol). Thus, alteration of these behaviors may control the risk factor, even Relative risk of CHD if it has never been clinically identified. Finally, risk factors 4.00 - tend to cluster within individuals as a result of behavioral or genetic mechanisms. Thus, the treatment of risk factors must often take these coexisting risk factors into account. The multifactorial nature of etiology of cardiovascular dis- 2.00- , ease and the interrelated nature of the most important risk . factors makes for a rich and complex range of possibilities in the development of preventive strategies. In practical applica- 1.00 tion, both the alteration of human behaviors leading to the risk factors (the population-based strategy) and the clinical treatment of those with high-risk factor levels (the high-risk 0.50 strategy) may be more straightforward strategies to reduce cardiovascular risk. When the population-based strategy is implemented before risk factors become highly prevalent in a 0.25 population, it is called primordial prevention (Dodu 1984, 1988). The preceding discussion can, perhaps, best be summarized in a diagram (figure 23-4) that shows a (simplified) scheme of 0 76 84 91 98 105 the relationships among groups of risk factors and end points. Polcy instruments per se are not shown in the figure, but they Approximate mean usual DBP (mm Hg) divide naturally into those that operate through the behavioral risk factors (for example, antismoking campaigns and taxes on fatty meat products) and medical interventions that operate Note: Stroke risk based on seven prospective observational studies with 843 directly on physiological risk factors (such as medication that events. CHo risk based on nine prospective observational studies with 4.856 lowers cholesterol). Emphasis on one policy or the other may events. The live baseline DBP categories are estimated from remeasurements in differ between population subgroups, depending on the prev- the Framingham study. Source: MacMahon and others 1990. alence and relative risk of thie risk factors. Cardiovascular Disease 581 Table 23-2. Association between Nonmodifiable and Behavioral Risk Factors and Development of Physiological Risk Factors Degree of association Nonmodifiable and behavioral risk factors Elevated LDL Decreased HDL Hypertension Diabetes Obesiot Nonmodifiable Age + +++ + + Male sex 4 ++- Black race - P .+. ++ a Family history +... + + ... + + + Behavioral Smoking -+ + Dietary saturated fat +.+ + - + Dietary calories + + + + + + + .... Salt consumption - + + + Alcohol consumption P + +- + Sedentary lifestyle + + + + + + .++ a. Women only. Key: + + + Strong association. +++ Moderately strong association. ++ Moderate association. + Weak association. - No association. P Protective association. Source: Authors' compilation. The breadth and sophistication of current epidemiologic 1970s, since which time they declined-a trend continuing knowledge of the etiology of cardiovascular disease is a scientific today. The data are much different for Eastern Europe, where triumph of the past two decades. This knowledge provides the cardiovascular disease rates have increased steadily without evi- tools for prevention that have been applied, with remarkable dence of stabilization or decline (Feinleib 1984; Thom and others success, in several high-income countries. Applying this knowl- 1985; Uemura and Pisa, 1988; Thom 1989; see table 23-3). edge in a developing country remains a central challenge. Rates in the developing world are much less well docu- mented. Urban China, for example, appears to be experiencing Clinical Presentation and Mortality Rates a similar rise in coronary heart disease (Wu and others 1984; Tao and others 1989). Other developing countries appear to This section reviews the available data for the specific diseases. have similar rises in coronary heart disease rates (the reasons are discussed later). A significant exception is the recently CORONARY HEART DISEASE. As mentioned previously, coro- documented experience of the city of Sao Paulo, Brazil, which, nary heart disease can present as angina pectoris, myocardial during the period 1970-83, experienced a 28 percent decline infarction, or sudden cardiac death. If worldwide proportions in mortality from ischemic heart disease and a 16 percent are similar to those in the United States, approximately 40 decline in deaths from stroke for the age group forty through percent of coronary disease presents as angina pectoris, 40 sixty-nine (de Lolio and others 1986). These reductions were percent as myocardial infarction, and 20 percent results in from quite high initial levels, so the pattern is like that of many sudden cardiac death (death within one hour of onset of of the industrial market economies. symptoms [Kannel and Feinleib 19721). Because as many as 20 percent of the myocardial infarction victims will die in the STROKE. Because the presentation of hemorrhagic and hospital (in addition to the victims of sudden death) and a atherothrombotic stroke appears similar clinically and devel- significant portion will remain disabled after infarction, pri- oping countries may not have the technology to estimate their mary prevention is an obviously important objective. The occurrence separately, their importance will be discussed to- reduction in case-fatality rates following infarction and the gether. In the United States, approximately 70 percent of prevention of infarction or sudden death in angina patients strokes are atherothrombotic, with an additional 12 percent would be important goals in secondary prevention. It should hemorrhagic (Schoenberg 1979). An additional 10 percent are be pointed out that any reduction in case-fatality rates for made up of subarachnoid hemorrhage and other stroke syn- myocardial infarction would yield prevalent cases of coronary dromes. In Asia, the occurrence of intracranial hemorrhage is disease, requiring great expenditure of health resources. This higher, with 50 percent of strokes being atherothrombotic and again emphasizes the need for primary prevention strategies. 40 percent being hemorrhagic. Coronary heart disease mortality rates increased in North The type of stroke is an important distinction, in that the America and Westem Europe until the late 1960s and early mortality and prognoses differ (Marquardsen 1986). No more 582 Thomas A. Pearson, Dean T. Jamison, andJorge Trejo-GuCierrez Figure 23-4. Relationships among Risk Factors for Cardiovascular Disease Nonmodifiable risk factors Endpoints Physiological risk factors MAle sex Hypertensive Mage sex _ heart disease Family history _ * [ Hypertension Hemorrhagic stroke Elevated LDL cholesterol Coronary heart disease Decreased LDL cholesterol Atherothrombotic Behavioral risk factors stroke Sedentary lifestyle Diabetes Diet 10Peripheral - Saturated fat vascular disease -Salt - Cholesterol - Total energy content Heavy alcohol consumption Smoking _ Source: Authors' compilation. than 30 percent of those who suffer atherothrombotic stroke The Burden of Cardiovascular Disease will die in the following three weeks; the prognosis for those with hemorrhagic stroke is worse, with case-fatality rates of 60 This section reviews the effects of increased incidences of to 85 percent. Overall, 20 to 30 percent of stroke victims suffer cardiovascular disease on current mortality levels. major disability; only 50 percent to 75 percent are able to walk unaided. Again, primary prevention appears to be the logical Current Levels of Mortality target for intervention. In table 23-4 we show rough demographic estimates, globally HYPERTENSIVE HEART DISEASE. Estimates of the prevalence of and by region, of the burden of mortality, in 1985, due to hypertensive heart disease have been difficult because of the vascular diseases. In the table in appendix 23A we provide aggregation of hypertensive heart disease with coronary heart more detailed information on sex differences. It is evident from disease in the Intemational Classification of Disease. Because table 23-4 that death rates, and even proportion of deaths, due congestive heart failure and sudden death can be due to a to circulatory disease are already very high indeed in Africa variety of causes, including coronary heart disease, estimates and Latin America and quite significant in Asia. Although the of this condition are imprecise. Yet, in some countries, espe- demographic assessments in table 23-4 are strongly suggestive cially those with low atherosclerotic prevalence (for example, of current levels of the mortality burden from CVD in different China), this condition appears to he an important cause of regions of the world, it is essential to bear in mind the substan- morbidity and mortality (Nissinen and others 1988, Hutt and tial uncertainty surrounding the numbers in that table. They Burkitt 1986). Blacks residing either in Africa, North Amer- rely,formostofthedevelopingcountries,notongoodnational ica, or Europe consistently appear to have earlier onset, higher cause-of-death data but, rather, on extrapolations to the de- levels, and more severe cardiac sequelae than do whites veloping countries today of the experience of other countries, ("Hypertension in Blacks and Whites" 1980; Falase 1987). mostly European, from a time when their mortality levels were Blood pressure control appears to be the main thrust of lower. It is therefore useful to supplement those extrapolations primary prevention; growing evidence suggests potential for with examples from developing countries for which data do regression of left ventricular hypertrophy with control of exist, and in table 23-5 we summarize such data from a number blood pressure. of sources. Cardiovascular Disease 583 Table 23-3. Change in Death Rates from Coronary This will very likely be true even where age-specific death rates Heart Disease, Selected Countries, 1969-78 and from CVD are declining, as they already are in parts of the 1979-85 developing world (for example, Sao Paulo [see de Lolio and (percent) others 1986]); where age-specific rates are rising-as they are Men Women for ischemic heart disease in Mexico (Lozano-Ascencio and Country 1969-78 1979-85 1969-78 1979-85 others 1990)-the increasing prominence of CVD will be more pronounced. This will be particularly true of the managerial Australia -24 -25 -25 -26a and professional classes in those countries. The trend toward Austria 4 -II' -6 14 Aegumstria6 -64 -7b the increasing importance of these diseases is dramatically Belgium -13 -164 21 4 22 illustrated in table 23-7. Canada -14 -24 -21 -22 Czechoslovakia 8 12 2 16 Denmark 14 -14 1 -16 EFFECT OF INCREASE IN LIFE EXPECTANCY. The crude incidence England and Wales 5 -12 9 -6 and prevalence of cardiovascular diseases must increase as life Finland -8 -16 -18 -23 expectancy increases. This so-called "epidemiologic conver- France 4 -4 -21 -2 sion" predicts such a rise in chronic disease rates as a function Former Fed.3Rep. of life expectancy (Dodu 1988). Because coronary disease Hungary 37 27 21 2 occurs even in people in their forties, it could represent signif- Ireland 12 -7 -10 3 icant burdens from those who are younger. Israel -22 -20 -19 -28 Another effect of increased age will be an increased preva- ltaly 13 -10C -7 -11' lence of risk factors which may require control. Hypertension Japan -22 -20 -37 -27 and elevated cholesterol are both examples of risk factors that Netherlands 1 -3 -2 -9 become increasingly prevalent as age increases in most indus- New Zealand -13 _?0 -16 -21 Norway -5 -16 -11 -6 trial countries, although these age-related increases may not Northem Ireland 18 -18 28 -16 be inevitable. Poland 77 21a 71 26a Portugal 2 -15 a -d -5a EFFECT OF INCREASE IN STANDARD OF LIVING. An increased Scotland 3 -14 2 -7 affluence of population, if it were to occur, would have several Spain 53 -- 29 - possible effects on cardiovascular disease rates. First, an im- Sweden 19 -12 -12 -19 proved diet may also increase saturated fat, dietary cholesterol, Switzerland 10 -13 -7 -17 United States -25 -22 -24 -15 Yugoslavia 32 15- 11 2 Table 23-4. Estirnated Mortality from Circulatory - Not available. System Diseases, by Region, 1985 Note: Death races are for men and women aged 45-64. - a. 1980-86. Age-standardized death rate b. 1979-84. (per 100,000 population)' c. 1979-83. Total Cerebro- d. 1971-78. Deaths deaths Ischemic vascular Region (thousands) (percent) Total disease disease Trends to 2015 Industrial market economies 3,355 46 235 99 59 It is likely that all the important cardiovascular diseases will Industrial increase sharply in their morbidity and mortality between now nonmarket economies 2,220 47 357 164 106 and the years 2000 and 2015. Assumed In these conclusions is Latin America the control of major childhood and infectious diseases, with and the concomitant increase in population, life expectancy, and stan- Caribbean 691 22 222 69 57 dard of living. The epidemiologic transition from infections to Sub-Saharan degenerative diseases as the main causes of death may be Africa 756 10 273 85 74 divided into at least four stages (table 23-6). A progressive Middle East and North Africa 602 14 250 82 68 change in risk factors and concurrent change in cardiovascular Asia 3,841 17 195 46 91 diseases is suggested by international comparisons of risk-factor prevalence and cardiovascular mortality in industrial countries Total 11,466 23 243 84 81 (see table 23-3 and Knuiman and others 1980; Masironi and Note: Constructed from the figures used by Bulatao and Stephens in tables Rothwell 1988; Nissinen and others 1988; Uemura and Pisa 3 and 4 for "ischemic heart disease," "cerebrovascular disease," and "other 1988; WHO MON ICA Proj ect 1988). I t is predic ted that coronar), cardiovascular." The total deanhs in this table refer to the sum of these three 1988; HONIOICA Poject1988) It ispredited tat coonary categories, nor to Bulatao and Stephens' "circulatory" category. disease, stroke, or hypertensive heart disease will be the leading a. Rates are standardized using the 1985 world age structure. cause of death for many developing countries by the year 2000. Source: Bulatao and Stephens 1990. 584 Thomas A. Pearson, Dean T. Jamison, and Jorge Trejo-GunerTez Table 23-5. Age-Standardized Mortality frmn tion (chapter 21), Barnum and Greenberg report enormously Vascular Diseases in Men in Selected Industrial high income elasticities of consumption of tobacco products, and Developing Countries particularly among the countries with the lowest incomes. Annual mortality rate Another, but favorable, effect of increased standard of living (per 100,000 population) may be the improved cold storage of food, which has been lschemic Cerebro- Ratio of suggested as the cause for the decline in strokes since 1900. heart vascular ischemic heart The reduced requirement for smoked and salted foods may Country disease disease disease to stroke reduce the sodium content, resulting in reduced hypertension. Indust-rial An increase in medical care services may also provide op- Canada 276 61 4.5:1 portunity for reduction in cardiovascular disease rates. The Japan 53 122 0.43:1 disbursement of funds for preventive as opposed to acute care Portugal 108 240 0.45:1 services, however, may determine the extent to which morbid- United States 274 59 4.6:1 ity and mortality are reduced. Former U.S.S.R. 486 245 2.0:1 Developing The Econormic Burden of Cardiovascular Disease Brazil (Sao Paulo)a 310 198 1.6:1 China(Beijing) ~ 124 - 1:5' Analyses of the economic burden of disease in developing China (Guangzhou)b 42 1: 5' countries are currently unavailable. It may, however, be sug- Mauritius 123 94 1.3:1 gestive to summarize findings on the burden of circulatory Costa Rica 65 26 2.5:1 Cuba 168 65 2.6:1 system disease in the United States based on one recent analysis (Rice, Hodgson, and Kopstein 1985). The authors of - Not available. this analysis found that the direct cost of illness (that is, the a. Age group forty through sixty-nine. b. Age group thirty-five through seventy-four. cost of providing care) was $211 billion in 1980; given a gross c. For China as a whole. national product (GNP) of approximately $2.6 trillion, this Source: WHO 1988b, 1989; Tao and others 1989: de Lolio and others 1986. amounted to 8.1 percent of GNP. Of direct costs, about 15.4 percent were for circulatory diseases, so the cost of caring for calories, sodium, and alcohol, leading to increases in blood these diseases was about 1.25 percent of GNP. The indirect costs cholesterol, body weight, and blood pressure. For example, of all disease (productivity loss due to morbidity and premature increases over time in dietary cholesterol, dietary fat, meat mortality) were slightly more, about 9.4 percent of GNP; for consumption, and cigarettes smoked per adult all correlated circulatory disease, the indirect costs were about 2 percent of with increases in coronary heart disease in industrial countries GNP. Two points are notable: first, at a cost of more than 3 (Byington and others 1979; Blackbum 1989; Epstein 1989). percent of GNP, circulatory disease imposes an enormous eco- Sodium consumption correlates significantly with blood pres- nomic burden. Second, because deaths due to circulatory sys- sure levels in a recent study of fifty-two centers worldwide tem disease were about 47 percent of total deaths, and its cost (Intersalt Cooperative Research Group 1988). Second, in- was less than 20 percent of the total costs of illness, the burden creased use of transportation and increased mechanization of of circulatory illness is relatively much less measured in eco- industry may increase the prevalence of sedentary lifestyles. nomic terms than mortality ones. Third, increased affluence may increase cigarette consump- Analyses of this sort are sensitive to alternative assumptions tion; indeed, in the companion piece on cancer in this collec- and methodologies, and transfer of results from the U.S. situ- Table 23-6. Circulatory System Disease at Phases of the Epidemniologic Transition Deaths from circulatory Phase of epidemiologic transition disease (percent) Circulatory problems Risk factors Age of pestilence and famine 5-10 Rheumatic heart disease; infectious and Uncontrolled infection; deficiency deficiency-induced cardiomyopathies conditions Age of receding pandemics 10-35 As above, plus hypertensive heart disease High-salt diet leading to hypertension; and hemorrhagic stroke increased smoking Age of degenerative and man- 35-55 All forms of stroke; ischemic heart disease Atherosclerosis from fatty diets; sedentary made diseases lifestyle; smoking Age of delayed degenerative Probably Stroke and ischemic heart disease' Education and behavioral changes leading diseases under 50 to lower levels of risk factors Note: Omran 1971 introduced the concept of epidemiologic transition with discussion of phases 1, 2, and 3. Olshansky and Ault 1986 added the concept of a fourth phase. a. At older ages. Represents a smaller proportion of deaths. Source: Omran 1971; Olshansky and Ault 1986. Cardiovascular Disease 585 Table 23-7. Ratio of Deaths from Circulatory ment of high-risk individuals (Kottke and others 1985; System Diseases to Deaths from Infectious Fries, Green, and Levine 1989; Gunning-Schepers and oth- and Parasitic Diseases, by Region and Year ers 1989), although arguments to the contrary have been Region 1970 1985 2000 2015 presented (Oliver 1984; Lewis and others 1986; McCormick and Skrabanek 1988). The current policy of the World Industrial market 5.63 6.64 9.61 11.88 Health Organization, which is to emphasize primordial pre- countries vention in developing countries, is predicated on a broad countries consensus among experts that a community-based strategy Latin America and the 0.68 1.09 2.46 4.74 is feasible (WHO 1982). Finally, all the community-based Caribbean strategies might be carried out in clinical settings, so that the Asia 0.42 0.60 1.55 2.75 community-based and clinic-based strategies are overlap- Middle East and Norch 0.37 0.41 0.81 1.26 ping, rather than exclusive. Sub-SAfaric Africa 0.23 0.27 0.38 0.54 The high-risk, or clinic-based, strategy does include Sub-Saharan Africa 0.23 0.27 0.38 0.54 several effective interventions, however. Lipid-lowering Total 0.69 0.89 1.62 2.39 drugs, for coronary disease and peripheral vascular disease, Source: Bularao and Stephens 1990. and antihypertensive drugs, for stroke, hypertension, and coronary disease, are clearly effective (MacMahon and oth- ers 1990). With the exception of antihypertensive thera- ation can, at best, be suggestive. That said, until country- pies, it has been difficult to attribute recent declines in specific analyses from developing countries are available, these coronary disease in industrial countries to the medical man- estimates do probably serve as a reasonable first approximation agement of risk factors (Pearson 1989). These strategies will of the cost of CVD (in percentage of GNP) for middle-income often require large numbers of well-trained workers, how- developing countries. ever, as well as technology in the form of laboratories to monitor effects and side effects of therapy (for example, Strategies for Preventing Cardiovascular Disease lipid laboratories to monitor serum potassium levels). Drugs can also be relatively expensive, although there are a number The preventive strategies, along with their cost in manpower, of important antiplatelet, antidysrhythmic, and antihyper- technology, and drugs, are listed for coronary heart disease, tensive drugs (for example, verapamil, propranolol, and stroke, and hypertension in table 23-8. The preventive strate- hydrochlorothiazide) available through the World Health gies are divided into public health (community-based) and Organization/United Nations Children's Fund (WHO/UNICEF) high-risk (clinic-based) strategies. Several conclusions can be Essential Drugs program at costs of under $5 per patient per reached. First, several of the public health interventions are year (UNICEF 1989). It should be emphasized, however, that effective, including smoking cessation. The recent reviews by a wide range of costs exists for drugs, from inexpensive Warner (1989, 1990a) and the Surgeon General ("Reducing lipid-lowering (for example, nicotinic acid) and antihyper- the Health Consequences of Smoking" 1989) of antismoking tensive agents (for example, reserpine) to expensive drugs campaign effectiveness in the United States are encouraging (HMNIC0.-A [3-hydroxy-3 methylglutaryl coenzyme A] reduc- in this regard. The Surgeon General's report, for example, tase inhibitors, angiotensin converting enzyme inhibitors). summarizes evidence on the potent effect of taxes on tobacco Aspirin, which is extremely inexpensive and widely avail- products, particularly in the young and particularly with re- able, plays a potentially central role through its antiplatelet spect to initiation of use. The estimated price elasticity of effects. Antiplateler drugs may be effective in the primary demand is a substantial -0.42 for adults and a dramatic -1.2 prevention of coronary disease but not stroke (Hennekens for twelve- to seventeen-year-olds. In what is perhaps the first and others 1989). study of the price (actually, tax level) elasticity of tobacco It should be pointed out that the relationship between the consumption from a developing country, Chapman and Rich- population-based and high-risk strategies remains to be as- ardson (1990) found stronger elasticities in Papua New Guinea sessed; nonetheless, the European Atherosclerotic Society than in the United States. Efforts to promote low-fat diets and (1987) has noted that it seems highly probable that the two active lifestyles are also probably effective. Second, the public strategies would be mutually supportive. Along these lines, health interventions are often not costly. Much of the cost for Rothenberg, Ford, and Vartiainen (1990) have undertaken community-based cardiovascular interventions may constitute a multirisk-factor modeling exercise that suggests that over- theuseofmanpower,dependingonthenatureoftheinterven- all preventive strategy may well focus on high-risk groups tion. Many developing countries have a relative abundance of for some risk factors and whole populations (probably age workers, including those with some training, which can be targeted) for others, depending on specific characteristics of applied to these tasks. The technology and drug requirements available interventions. for these strategies are low. Third, the ability to shift the risk It is not likely that new technologies available for public of an entire population is calculated to have more promise in health interventions will become available. Knowledge re- the prevention of the majority of deaths than does the treat- garding the beneficial effects of lipid-lowering diets, exercise, 586 Thomas A. Pearson, Dean T. Jamison, and Jorge Trejo-Gunerrez Table 23-8. Efficacy and Cost of Preventive Strategies for Major Cardiovascular Diseases Effectiveness Costs Preventive strateg-v Compliance Clinical efficacy Manpower Technology Drugs Coronary heart disease Community-based Smoking cessation + +.++ +/- 0 0 Low-salt diet + /- 0 Modified far diet + .++ +/- 0 0 Exercise + . +/- 0 0 Diabetic diet + - +/- 0 0 Obesity control ?/- 0 0 Clinic-based Antihypertensive drugs + + + ++ + + + Lipid-lowering drugs +/- ++++ + + + + + + Diabetic drugs + - ++ ++++ ++ + Antiplatelet drugs + + + + + + + Stroke Community-based Smoking cessation + .+ +/- 0 0 Low-salt diet + + +/0 0 Obesity control - + +/- 0 0 Clinic-based Antihypertensive drugs + .+.+ + + + + + Anriplatelet drugs 4+ .++++ + + + Hypertension Community-based Low-salt diet + + +/- 0 0 Exercise + + + + 0 0 Obesicy control - +/- 0 0 Alcohol restriction + 0 0 0 Clinic-based Antihypertensive drugs +/++ ++++ + 4 + + Key: .+.+ Highly favorable. +++ Moderately favorable. ++ Favorable. + Minimal. +/- Variable. - Poor. - Not effective. O Not required. ? Unknowln. Source: Authors' compilation. and sodiuin restriction, however, should continue to increase. being implemented but are probably too costly for most devel- For example, results are now being published from the study opingcountriesat presentprices (Weinstein and Stason 1985). undertaken by the IntersaltCooperative ResearchGroup (1988) Diabetic drugs, including new insulin delivery systems, are in thirty-two countries, includingtwelvedevelopingcountries, being evaluated for efficacy. Again, these are high-technology of the prevalence of hypertension and its association with instruments, usually with high unit costs. sodium, potassium, and alcohol intake. Community interven- tion strategies that will probably be generalizable to developing The Gap between Good Practice and Actual Practice countries are now being implemented in the United States and Europe (Blackburn 1983). Many, or perhaps most, countries, industrial or developing, A number of changes will likely occur with physician-based have not enacted national policies intended to prevent the use preventive strategies (Pearson 1989). New antihypertensive of tobacco products. Further, few countries encourage low-fat, drugs with better treatment schedules and fewer side effects are low-cholesterol, low-salt diets; indeed, few countries have a constantly being developed. It is hoped that inexpensive, low national nutrition policy of any kind. Exercise and fitness are side-effect, and once-a-day dosing drugs will become available. variably encouraged in schools and work sites. Thus, a number New techniques for detection of high blood cholesterol are of public health or regulatory policies ieed to be implemented. Cardiovcascular Disease 587 Physicians in industrial countries remain oriented toward case, point more sharply in one direction than another; rather, acute care, rather than preventive care. Many health messages they suggest adjustments of treatment cutoff points (hyperteni- are therefore not delivered at the time of the acute care visit sion cutoffs down, cholesterol cutoffs up) to maximize health (for example, an antismoking message). Thus, risk factors in gains. Further, for developing countries, these costs are high; the high-risk patient are often not detected or treated. later we suggest that care taken in implementation could lower In jLidging the potential for improving pracrice, assessments costs in developing countries. of the cost-effectiveness of alternative strategies for prevention provide potentially useful guides. Weinstein and Stason Suggested Intervention in a Standardized Population (1985) provide a valuable introduction to the relevant litera- ture that was available at the time of their writing; that A concerted effort through public education, including that in literature is limited, unfortunately, to studies from high- the schools, work sites, and media, should be carried out to income countries. Further, we have been unable to find any reduce the prevalence of smoking. Regulations to restrict studies assessing community-based interventions; the closest smoking from work sites, public places, and the like might be was a study of screening of hypertensives followed by their enacted. The sale of cigarettes might be heavily taxed. Nutri- treatment in North Karelia, Finland. The authors of the North tional programs to provide low-fat, low-cholesterol sources of Karelia assessment estimated the cost per (disability-adjusted) protein should be enacted. The use of nonsaturated fat in year of life gained to be about $3,600 (Nissinen and others cooking oils should be encouraged. Facilities and encourage- 1986). Weinstein and Stason reported that in earlier studies ment for exercise should be made policy at schools, work sites, on hypertension control rather more costly, although not and public places. The use of personal transportation might markedly higher, results had been found. Naturally, the cost emnphasize expenditure of calories for the nonpoor population per year of life gained falls as the cutoff level for treatment of subgroups. blood pressure rises-btit with concotmitant loss of some lives As health care strategies, all adults might have their blood that would have been saved were the cutoff level lower. In a pressure taken, and those in whon it is found to be elevated recent study, Hatziandreti and others ( 1988) assessed the cost- might begin a weight loss, salt-restricted diet with appropriate effectiveness of exercise as a preventive measure for ischemic physician follow-up, including the use of inexpensive antihy- heart disease, finding costs of about $1 1,000 per disability- pertensive agents. Cholesterol screening might be restricted to adjusted year of life saved. The main cost is that of the time persons with relatives who have developed cardiovascular taken for exercise and that "cost" is itself highly dependent on disease at age fifty years or less. Follow-up and treatment would whether exercise is a pleasure; the authors assumed that, on require dietary restrictions offats and cholesterol, and possibly average, it was not. low-cost medicines to lower cholesterol. Much of the work on the cost-effectiveness of prevention We have made several tentative calculations to estimate has addressed the attractiveness of efforts to reduce serum probable cost and effect, in the hypothetical population of I cholesterol. Taylor and others (1987) put forward an often- million, of several preventive interventions-a general public refuted model linking changes in cholesterol levels to changes preventive package (education, screening, counseling, refer- in life expectancies for varying initial ages and cholesterol ral), a program to control hypertension, a program to control levels; the results provide the effectiveness measures that can hvperlipidemia, and a 20 percent tax on tobacco products. Our be used in cost-effectiveness analyses and suggest very little preliminary estimates, which assume low-cost medications benefit (several moniths' gain in life expectancy) for lifelong (such as propranolol purchased at prices on the Essential Drug programs to lower cholesterol by typical dietary means. More List [UNICEF 1989]), are as follows: substantial gains result from the potent effect of cholesterol- * For hypertension control, perhaps 90 deaths per year lowering drugs, but an early analysis of the cost per year of life cou avertedsatn cost of abu $000 per year gained from the use of cholestyramine resulted in high esti- d mates, in the region of$125,000 (Weinstein and Stason 1985). gained In a still earlier assessment of a pediatric screening program * For hN'percholesterolemia control, somewhat fewer followed by dietary management, Berwick, Cretin, and Keeler dealths could e averted at a cost of about $4,000 per year of (1980) found much more modest costs-in the region of lifegained $10,000. Kinosian and Eisenberg (1988) assessed several * Assuming that a public preventive package could be agents in their investigation into the medical managemenit of implemented for about half the cost as in the United States cholesterol. Their findings on cholestryamine were similar to (tthat is, for ahout $0.75 percapita per year) perhaps 250 lives those of Weinstein and Stason, but they found that the use of per year could be saved at a cost of about $150 per year of oat bran resulted in much lower costs-about $18,000 per year life gained of life saved, which is still expensive in relation to the hylper- * A 20 percent tobacco tax would reduce consumption by tension control efforts assessed. about 20 percent and free resources for other uses. Assuming The purpose of these assessments is to help guide resources that there were 200,000 smokers in the population of I toward uses that buy the greatest possible gain in disability- million and that the health effect was as though 40,000 adjusted life-years for the money available. They do not, in this people stopped smoking and others did not change behav- 588 Thormas A. Pearson, Dean T. Jamison, and jyrge Trejo-GuneTrez ior, perhaps 40 deaths per year from coronary artery disease Gaps between Good Practice and Actual Practice would be averted. It appears likely that some of the new technology will improve Calculations of this sort will, obviously, give quite different case-tatalitv rates. Many institutions, however, even in the results in different epideiniiological and health service environ- industrial countries, cannot afford its implementation. There iments; and even these estimates for a hvpothetical population also appears to be a lack of attention to some of the rather should be viewed as tentative. Nonetheless, these estimates do simple chronic interventionis (smoking cessation, cardiac re- give a rough sense of the cost-effectiveness that might he hahilitation, beta-blocker therapy) which have been shown to expected for several key interventions. be related to improved chance of survival in patients with CHD. Thus, an improved use of these interventions may increase Case Management of Cardiovascular Conditions survival rates. Cost-effectiveness studies for several of these strategies have been attempted, including bypass surgery, use The potential elements of case managenment strategies are of beta blockers after myocardial infarction (mi), coronary care listed in table 23-9 and are separated into acute and chronic units, and cardiopulmonaryresuscitation programs(Weinstein interventionis. Several conclusions can be drawn. First, the and Stason 1985). More recently, Steinberg and others (1 988) acute interventions are often not extremely effective when examined the use of thrombolytic therapy in treating acute Mi. tested in clinical trials (for example, antiarrhythmic drugs Most of these strategies are expensive (upward of $30,000 per such as lidocaine, coroniary care units, cardiopulmonary year of life saved), although CABS for left main coronary artery resuscitation, calcium channel blockers; see MacMahon disease and three-vessel coronary artery disease is a relatively and others 1988; Yusuf, Wittes, and Friedman I 988a; Held, cost-effective treatment in high-income countries, at costs Yusuf, and Finberg 1989). Second, these acute interveni- well under $10,000. Risk stratification appears able to identify tions are costly with regard to manpower, technology, and patients who have especially attractive cost-benefit ratios; drugs. Third. the effectiveness of some chronic interven- Goldman and others (1988) estimate beta-blocker therapy to tions is perhaps better, with more widespread application cost from $23,400 per year of life saved in low-risk patients to feasible. Again, several interventions suchi as angioplasty $3,600 per year of life saved in high-risk patients. and bypass surgery are costly; even so, targeted interven- tions (for example, coronary artery bypass surgery [CABS] for Suggested Intervention in a Standardized Population left main coronary artery disease only) can be cost-effective, at least in industrial countries (Williams 1985). Certain Four interventions appear worthy of serious consideration- interventions, such as smoking cessation, lowering of serum risk reduction in post-mi patients, risk management in post- cholesterol, and exercise in cardiac rehabilitation, show stroke patients, angina control, and low-cost management of promise of being effective at low cost (Oldridge and others acute Mi. These are discussed below. 1988; Yusuf, Wittes, and Friedman 1988h; "The Surgeon For both the post-Mi and poststroke patients, emphasis on General's 1990 Report on the Health Benefits of Smoking the modificationi of risk factors should be made, including the Cessation" 1990). It is further logical that chronic risk control of hypertension in the stroke victim; the cessation of factor intervention may be more effective in populations smoking in the coronary, cerebrovascular, and peripheral vas- also in communiity-based programs. cular patient; and the lowering of blood cholesterol by diet or It is further predicted that the classes with greater education inexpensive drugs. For the patient with chronic cardiovascular and income may demand this high-technology care, obtained disease, rather inexpensive drugs might be used for the treat- in their home countries or by traveling to an industrial country. ment of angina or following myocardial infarction (for exam- Because it is these classes that will develop cardiovascular ple, beta blockers; see Julian 1989). Antiplatelet agents, disease first in a country, extreme pressures will likely be particularly aspirin, may be helpful in patients with coronary exerted to develop high-technology health care. It is probable artery disease, and atherothrombotic but not hemorrhagic that for this reason a small amount of this high technology w ill stroke. These low-cost secondary preventive measures for this be needed in all countries. Nonetheless, a key element of high-risk group might cost $3 per patient per year, if carefully overall strategy will be to make low-cost but effective alterna- implemented. They might be expected to reduce the annual tive interventions regularly available. In many cases, these probability of death by 0.01 or 0.02 for an age group in which options do exist. a death would result in loss of, perhaps, fifteen discounted (at The field of interventional cardiology is rapidly changing, 3 percent) life-years for the post-Mi patient and ten years for with the advent of new therapies for the patient with angina the poststroke patient. This would result in a cost per disabil- pectoris, myocardial infarction, stroke, or peripheral vascular ity-adjusted life-year gained of $155 for intervention in the disease. Most invasive strategies (for example, laser angiopla- post-mi group and $230 in the poststroke group. sty) and those involving new drugs (new thrombolytic agents, Of particular importance in case management is the cost-ef- free radical scavengers, and so on) are likely to be costly in fectiveness of angina control medication, which is inexpen- manpower, technology, and drugs. A full description of poten- sively available from the WHOI/UNICEF Essential Drugs program. tial advances is beyond the scope of this review. It is estimated that the standard population would have about Cardiovascular Disease 589 Table 23-9. Effectiveness and Cost of Case Management Strategies for Cardiovascular Disease Costs Case management strategy ___ ____ Efficacy Manpower Technology Drugs CoronaTy heart disease Acute mi/unstable angina Cardiopulmonary resuscitation +/-. .+++ + + Coronary care units + ++++ ++.++ ++ Antiarrhythmic drugs + + + + +++ Thrombolysis + + .... ++++ ....+ Nitrates + + +/.+++ + +I++ Antiplatelet/anticoagulant drugs + + + + + Chronic angina/stable post-MI Beta-blocking agents +.. .+++ + .... Calcium channel blockers 0 +..++ + .... Angioplasty ? +.-.+ +.+.+.+ + + Coronary artery bypass surgery + +.... ... + + Antiarrhythmic drugs - +++ + . .+.++ Cardiac rehabilitation (including smoking cessation) +.++ .+++.- + 0 Antiplatelet agents + + + 0 + Cholesterol-lowering drugs + + + + +++ ++++ Congestive heart failure Drug therapy + +. ... + 4++- +.+.+.+ Heart transplantation + + +... ++++ .++++.. S troke Acute stroke General support +++ + + 0 Stable poststroke/transient ischemic attack Antiplatelet agents + + + 0 + Antihypertensive drugs +.. + + + +++ Carotid surgery ?+++ +++ + + Peripheral vascular disease Acute Surgery + + ... +... + + Chronic Surgery/angioplasty + + +.+.+.+ +.+.+.+ + + Smoking cessation +.++- + 0 0 Hypertensive disease Congestive heart failure Drug therapy +.- ++++ . ..... Hypertensive renal disease Dialysis + + +... +++.+ + + Key: ++++ Verv high. +++ Hligh. ++ Moderate. + Minimal. - Not effective. +/- Variable. 0 Not required. ? Unknown. Source: Aurhors' compilation. 6,000 cases of moderate to severe angina at any given time, of life gained from angina control would be only $100 to which could be medically controlled for perhaps $150,000 to $200. $200,000 per year. If, consistent with Weinstein and Stason, Table 23-9 contains a range of options for management of we use a Q factor of 0.7 to 0.9 for angina (that is, between 0. I acute Nil and unstable angina. Many of these are highly costly, and 0.3 years of healthy life are assumed to be lost per angina- and some are of limited or unproven effectiveness. One strat- affected person per year), the cost per disability-adjusted year egy of medical intervention, however, would be relatively 590 Thomas A. Pearson, Dean T. Jamison, and Jorge Trejo-Gutierrez easily managed and involve only very low cost drugs-nitro- cardiovascular mortality occurs. Likewise, although cardiovas- glycerin and aspirin (as an antiplatelet agent). Pitt (1989) cular disease already accountsfor lOto 35 percent ofall deaths suggested that at least some of the reduced case-fatality rates in developing countries (and, soon, will account for a much for Mi observed in industrial countries during the past two higher percentage), the substantial literature on epidemiology decades may be attributed to these agents. Patients presenting and health planning for developing countries concentrates with acute mi would be provided these medications (and sup- almost exclusively on communicable diseases, particularly the portive care) on an inpatient basis for several days. The drug communicable diseases of childhood. Nonetheless, for a few cost would be close to negligible; depending on local circum- developing countries, and in a preliminary way, explicit con- stances, the cost of hospital stay and physician time might be sideration has begun of the implications for health policy of in the range of $100 to $250. Intervention of this sort might the increasingprominenceofchronicdiseases; forexample, for reduce the short-term mortality risk from 20 to 15 percent, China (Jamison and others 1984), for Malaysia (Harlan and resulting in a gain of 0.75 disability-adjusted life-years (again others 1984), for Mexico (chapter 3, this collection), and for assuming that a death averted at about this age will result in a Brazil (de Lolio and others 1986, and ongoing World Bank gain of fifteen disability-adjusted life-years). The resulting work). Our purpose in this chapter has been to take stock cost-effectiveness would, then, be in the range from $135 to of what literature does exist concerning cardiovascular dis- $335 per disability-adjusted life-year gained. eases in developing countries as a starting point for further The above cost-effectiveness estimates can be considered work. only approximate; much more careful analyses could (and The appropriate range of policy instruments to consider, as should) be done in country-specific circumstances. The esti- well as research priorities, will vary, of course, depending on mates ofcostand effectiveness are, nonetheless, within reason- the epidemiological and economnic conditions in a particular able ranges; the resulting cost-effectiveness estimates serve to country. Although many of those conditions are quite country give a realistic (if approximate) sense of what is possible. specific, we have nonetheless found it useful to characterize In table 23-10 we summarize the range of case management the evolving nature of circulatory problems in a way that options for vascular disease with regard to their objectives parallels the epidemiological transition; in table 23-6 we sum- (secondary prevention or rehabilitation) and the sophisti- marized this characterization. During the pretransition phase cation of the probable venue for the intervention. Priority of high mortality, circulatory problems are of modest relative interventions tend to be those that can be delivered in less importance (although, perhaps, of substantial absolute im- sophisticated environments, such as those described in more portance); they are, in this phase, dominantly conditions of detail in the preceding paragraphs. infectious origin, which we have not discussed in this chap- ter. As mortality declines, there appear to be, first, rising Conclusions and Priorities problems associated with the hypertension-related circula- tory diseases; this stage is followed, perhaps substantially We have attempted to summarize available data and analyses later, by diseases that are principally atherosclerotic in ori- concerning the epidemiology of cardiovascular diseases in gin. (Indeed, the low levels of atherosclerotically related developing countries, strategies of prevention, and methods of diseases in Japan today suggests the general possibility that case management. Imbalances in the available literature are a major epidemic of these diseases could be avoided in other striking: there is a vast medical and epidemiological literature parts of the world.) This phasing has important implications on cardiovascular disease, but almost none of it deals with for the timing of the introduction of both preventive and problems of developing countries, where well over half of case management interventions. Table 23-10. Objectives and Venue for Case Management of Cardiovascular Disease Venue Objective __ ___ _ Household Prinry facility __ _ Secondarfacili-ty _ < Tertaryfaciity Secondary prevention Angina pectoris Behavioral change: risk Simple diagnosis: Complex diagnosis;b Invasive diagnosis; surgical Myocardial infarction factor management; prescription of first- second-line drugsc therapy; complex drugs and Stroke drug compliance line drugs' technologies Rehabilitatioon Myocardial infarction Habits of daily living; Monttored outpatient Supervised inpatient Specialized rehabilitative Stroke long-term, unsupervised program program; physical and occupational therapy; exercise progiam occupational therapy prostheses, patient-assist devices a. Propranolol, verapamil, aspirin. b. Established through exercise electrcardiogram, echo-dolppler examination, and similar technologies. c. For example, long-acting nitrates, anni-arrhythmic drugs. Source: Authors' compilation. CaTdiovascular Disease 591 Operational Priorities for Developing Countries CASE MANAGEMENT. The range of cost-effective case manage- ment interventions is much narrower in developing countries The following steps might be appropriately emphasized as ones than in industrial ones. They include the following: that developing countries might take to forestall (primordial * Post-MI and poststroke care. Improvement is needed in prevention) or reduce the forthcoming epidemic of cardiovas- such care, with emphasis on modification of risk factors (for cular disease and to deal with its consequences. These inter- example, smoking) and inexpensive drug treatment (for ventions tend also to be appropriate for other chronic diseases, example, beta blockers, aspirin). and the suggestions here are generally consistent with those of e beta blo pranolo idc WHO's "Interhealth" program (Integrated Programme for Non- * Low-cost drugs (propranolol, aspirin) and modification communicable Disease Prevention and Control, Shigan of risk factors for treatment of angina. Risk factor modifica- 1988). The interventions are listed in order of priority for tion should include lipid-lowering diet or drugs, smoking implementation. cessation, and the like. Treatment of angina appears partic- ularly cost-effective. PREVENTION. The range of appropriate preventive interven- * An inexpensive (and cost-effective) protocol to treat tions in developing countries is similar to that for industrial acute MI and unstable angina. Such a protocol should em- ones and includes the following: phasize the use of aspirin and nitroglycerin. Regulation, taxation, and education to reduce the pro- * Strict limitations on provision for (locally or abroad) duction and use of tobacco products. These form perhaps invasive diagnostic (coronary arteriography) and treatment the single most important cluster of preventive interven- (coronary care units, coronarybypass surgery) facilities, with the inge mot iportnt lustr o preentve iteren- their use limited to the young, highest-risk subgroups of tions. The potential benefits of this strategy are reviewed in t patients, and then only after the more cost-effective strate- more detail In appendix A of this collection, and Warner (1990b) emphasizes the particularly important role that gies just described have already been tried. taxation policy can play in limiting tobacco use in develop- Operations Research in Developing Countries ing countries. * National nutrition policies, including substantial use of The following are areas of priority for operations research, taxes, to improve nutrition without the use of excess satu- along the lines that Tanzania has initiated with the assistance rated fat, dietary cholesterol, salt, calories, and alcohol. of WHO (WHO 1986). Given the public health significance of These policies should be targeted at those segments of a cardiovascular disease in developing countries, and its rising society in whom malnutrition (that is, undernutrition) is absolute and relative importance, the current neglect of ap- not a problem. The prevention of obesity should be a major plied research is striking. target of these programs. * Screening for (probably in the context of other screen- EPIDEMIOLOGY. Additional data on the distribution, causes, ing) and nonpharinacologic treatment of hypertension. Use and national history of CVD is needed in the following areas: of even relatively inexpensive antihypertensive drugs in thos hig-ris patentsnot espodingto nnphama- * Better data must be sought on the incidence, prevalence, c tho erayisk probably not cost-ffeti excpt form case-fatality rates, and prognosis of cardiovascular diseases cologic therapy is probably not cost-effective, except , . to determine the circumstances under which development individuals at very high risk. The drugs themselves are now and increasing affluence result in increased age-specific quite inexpensive (UNICEF 1989); the cost issue is that of cardiovascular disease rates. monitoring of response, side effects, and the compliance ovasculr datase rates. of patients with their treatment regimens. The large num- * Better data must be sought on the prevalence of modifi- ber of compliant individuals required per death averted is able risk factors, especially in those countries with increas- the limiting factor on cost-effectiveness. ing cardiovascular disease rates. * Better data must be sought on the relative risk and the * National fitness programs. Such programs should be em- phasized for those segments of the population that have attributable risk of risk factors in developing countries. occupations or lifestyles which reduce their levels of physical * Developing country participation must be greatly ex- activity, or those subgroups prone to obesity. panded in international programs of cardiovascular epide- * Toward later stages of the epidemiologic transition, miologic investigations and surveillance, such as MONICA, to limited introduction of inexpensive antihypertipidemics. provide standardized measurement techniques and numer- These might be considered for individuals with very ad- ous industrial countries for comparison. verse lipid profiles; the WHdO/UNICEF Essential Dug List verse lipid profiles; the WH,/UNICEF Essential D.ug Lit ,RISK MODIFICATIONS. Little information is available on the should, therefore, be expanded to include an effective, effectiveness of interventions to lower the risk of CVD. Addi- inexpensive lipid-lowering agent (such as nicotinic acid). tional research is needed in the following areas: As with control of hypertension, however, but even more so, the large number of compliant individuals required to * Researchmustbeundertakenincommunity-specificand avert a death sharply constrains cost-effectiveness. culture-specific strategies to use mass media, community 592 Thomas A. Pearson, Dean T. Jamison, and Jorge Trejo-Gurierrez resources, and other modifiers of human behavior to alter developing countries; we have reviewed the range of available the level of risk factors. options for prevention and case management; and we have * The effectiveness of price and taxation policies in mod- assessed the cost-effectiveness of a range of the most attractive if ying the distribution of risk factors must be studied. intervention options. A number of broadly relevant inter- ifyig Stuedie stribeundertakn ofrisk fcths ratie effectu iven ventions have been identified that, potentially, have cost- * Studies must be undertaken of the relative effectiveness efetvns meSUe frm$5 o$5 e iaiiy and cost-effectiveness of both nonpharmacologic and drug effc ivens eares;from $150 to $350 per disability- inevnin* ordc ris factr in iniiul of that adjusted life-year gained; although these figures are not the patervticu c ture. most attractive that Jamison (chapter 1 of this collection) has summarized for interventions addressing health problems of CASE MANAGEMENT. The management of established CVD also adults, they do fall at the low end of the range and should needs further invstia manaen. become integral to the range of services most countries pro- vide. Interventions to reduce smoking are even more attrac- * Studies must be conducted of the relative effectiveness tive, in part because they simultaneously reduce the risk of lung and cost-effectiveness of both nonpharmacologic and drug cancer and other conditions. Finally, it is worth stressing that interventions to reduce case-fatality rates and to improve many interventions of very low cost-effectiveness have also prognosis in patients with cardiovascular disease. been identified; considerations of both efficiency and equity * Given that only very small numbers of expensive inva- suggest that their use should be curtailed. sive procedures will be undertaken, research must be carried out concerning the economics of referral for these pro- cedures and optimal location for them (including at cen- Appendix 23A. Cardiovascular Mortality ters abroad). Differentiated by Sex In this chapter we have assembled evidence that points to MONICA data suggest a very strong effect for ischemic heart the large and growing significance of cardiovascular disease in disease and a much weaker one for stroke among males. Table 23A- 1. Sex Differences in Age-Standardized Cardiovascular Mortality, Selected Populations, 1984 Cerebrovaoscular disease Ischemic heart disease Male/female Male/femnale Male mortalit-y Female ratio of Male mortality Female ratio of rate mortalit rate Male/female proportional rate mortality rate Male/female proportional Location (10-m) Orl) rao mortality' 5) (10 ratio mortality' Perth, Australia 26 24 1.1 0.57 161 38 4.2 2.1 Bejing, China 98 82 1.2 0.96 40 28 1.4 1.3 Finland 60 42 1.4 0.46 374 57 6.6 2.1 Italy: Latina area 68 33 2.1 0.92 96 20 4.8 2.3 California 25 20 1.3 0.59 221 40 5.5 2.7 Note: Mortality Tates are annual, for a population of I00,000 age thirtv-five through sixty-four. a. Obtained by dividing the percent of total mortality in males by the same percentage in females. Source: MONICA Principal Investigators 1987. Notes Blackbum, H. W. 1983. "Research and Demonstration Projects in Community Cardiovascular Disease Prevention." Journal of Public Health PolicN 4:398- The authors wish to acknowledge their gratitude to Robert Beaglehole. 421. Jeanne Bertolli, John Briscoe, Elaine Eaker, Jon Eisenberg, John Evans, . 1989. "Trends and Determinants ofcHD Mortality: Changes in Risk Richard Feachem, William Harlan, Millicent Higgins, Tord Kiellstrom, Jeffrey Factors and Thelr Effects." Interrationai Journal of Epidemiology 18(supple- Koplan, Matthew Longnecker, Adetokunbo 0. Lucas, Anthony R. Measham, ment I ):S210-S2 15. Richard Morrow, P. Nordt, Kenneth Powell, and P. Tatsanavivat for their Blankenhorn, D. H., S. A. Nessim, R. L.Johnson, M. E. San Marco, S.P. Azen, assistance and comments on earlier drafts of this chapter. Richard Peto and L.Cashin-Hemphill. 1987."BeneficialEffectsofCombinedColestipol- provided particularly detailed and valuable critical comments. Niacin Therapy on Coronary Atherosclerosis and Coronary Venous Bypass Grafts."JAMA 257:3233-40. Bulatao, R., and P. Stephens. 1992. "Global Estimates and Projections of References Mortality by Cause, 1970-2015." PRE Working Paper 1007. Population, Health, and Nutrition Department, World Bank. Washington, D.C. Berwick, D. M., S. Cretin, and E. B. Keeler. 1980. Cholesterol, Children, and Byington, R., A. R. Dyer, D. Garside, and others. 1979. "Recent Trends of Heart Disease: An Analysis ofAlternatives. Oxford: Oxford University Press. Major Coronary Risk Factors and CHD in the United States and Other Cardiovascular Disease 593 Industrialized Countries." In R. J. Havlik and M. Feinleib, eds., Proceedings Kottke, T. E., P. Puska, J. T. Salonen, J. Tuomilehto, and A. Nissinen. 1985. of the Conference on the Decline in Coronary Heart Disease Mortality NIH "Projected Effects of High-Risk versus Population-Based Presentation Strar- 79-1610. U.S. Department of Health, Education, and Welfare, Public egiesinCoronaryHeartDisease." AmericanjournalofEpidemiology_121:697- Health Service, National Institutes of Health. Washington. D.C. 704. Chapman, Simon, and Jeff Richardson. 1990. "Tobacco Excise and Declining Lewis, B., J. I. Mann, and M. Mancini. 1986. "Reducing the Risks of Coronary Tobacco Consumption: The Case of Papua New Guinea." AmericanJournal Heart Disease in Individuals and in the Population." Lancet 1:956-59. of Public Health 80:537-40. de Lolio, C. A., J. M. P. de Souza, and R. Laurent. 1986. "Decline in Dodu, S. R. A. 1984. "Coronary Heart Disease in Developing Countries: The Cardiovascular Disease Mortality in the City of S. Paulo, Brazil, 1970 to Threat Can Be Averted." World Health Organization Chronicle 38:3-7. 1983." Revista de Saude Publica 20:454-64. . 1988. "Emergence of Cardiovascular Diseases in Developing Coun- Lozano-Ascencio, R., and others. 1990. "Tendencia de la mortalidad por tries." Cardiology 75:56-64. cardiopatia insquemica en Mexico, de 1950 a 1985." Salua Piblica de Mdxico Epstein, F. H. 1989. "The Relationship of Lifestyle to Intemational Trends in 32:405-20. CHLD." InternationalJoun7al of Epidemiology 18(supplement I):S203-S209. McCormick, James, and Petr Skrabanek. 1988. "Coronary Heart Disease Is European Atherosclerosis Society Study Group. 1987. "Strategies for the Not Preventable by Population Interventions." Lancet 2:839-41. Prevention ofCoronary Heart Disease: A Policy Statement of the European McGill, Henry. 1968. "Fatty Streaks in the Coronary Arteries and Aorta." Atherosclerosis Society." European HeartJournal 8:77-88. Laaoratory Investigation 18:560-64. Falase, A. 0. 1987. "Are There Differences in the Clinical Pattern of MacMahon, S., R. Collins, R. Peto, R. W. Koster, and S. Yusuf 1988. "Effects Hypertension between Africans and Caucasians?" Tropical Cardiology of Prophylactic Lidocaine in Suspected Acute Myocardial Infarction: An 13:141-5 1. Overview from the Randomized, Controlled Trials."JAMA 260:1910-16. Feinleib, Manning. 1984. "The Magnitude and Nature of the Decrease in MacMahon, S., R. Peto, J. Cutler, R. Collins, P. Sorlie, J. Neaton, R. Abbott, Coronary Heart Disease Mortality Rate." American Joumal of Cardiol)gy J. Godwini, A. Dyer, and J. Stamler. 1990. "Blood Pressure, Stroke, and 54:2c-8c. Coronary Heart Disease. Part I-Prolonged Differences in Blood Pressure: Fries,James F.,Lawrence W.Green,andSolLevine. 1989. "HealthPromnotion Prospective Observational Studies Corrected for the Regression Dilution and the Compression of Morbidity." Lancet 1:481-82. Bias." Lancet 335:765-74. Goldman, L., S. T. B. Sia, E. F. Cook, J. D. Rutherford, and M. C. Weinstein. Marquardsen, J. 1986. "Epidemiology of Strokes in Europe." In H. J. M. 1988. "Costs and Effectiveness of Routine Therapy with Long-Term Beta- Bamett, B. M. Stein, J. P. Mohr, and F. M. Yatsu, eds., Stroke: Pathophysiol- Adrenergic Antagonists after Acute Myocardial Infarction." New England ogy, Diagnosis, and Management. London: Churchill Livingstone. Journal of Medicine 319:152-57. Masironi, R.. and K. Rothwell. 1988. "Tendances et effects du tabagisme dans Gunning-Schepers, L. J.. J. J. Barendregt, and P. J. van der Maas. 1989. Ie monde." World Health StatisticalQuarterly 41:228-41. 'Population Interventions Reassessed." Lancet 1:479-80. Melnick, J. L., Evin Adam, and Michael E. DeBakey. 1990. "Possible Role of Harlan. W. R, L. C. Harlan. and W. L. Oii. 1984. "Changing Disease Pattems Cytomegalovirnus in Atherogenesis." JAMA 263:2204-7. in Developing Countries: The Case of Malaysia." In P. Leaverton and L. MONICA (Monitoring Trends and Determinants in Cardiovascular Disease) Massi, eds., Health Infornation Systems. New York: Praeger Scientific. Principal Investigators. 1987. "WHOMONICA Project: Geographic Variation Hatziandreu, E. I., J. P. Koplan, M. C. Weinstein, C. J. Caspersen, and K. E. in Mortality from Cardiovascular Disease." World Health Stanstics Quarterly Wamer. 1988. "A Cost-Effectiveness Analysis of Exercise as a Health 40:171-84. Promotion Activity." AmencanJournalof Public Health 78:1417-21. Nissiineni, A., S. Bothig, H. Granroth, and A. D. Lopez. 1988. "Hypertension Held, P. H.. S. Yusuf. and C. D. Furberg. 1989. "Calcium Channel Blockers in in Developing Countries." World Heakh Statistical Quarrerby 41:141-54. Acute Myocardial Infarction and Unstable Angina: An Overview." British Nissinen, A., J. Toumilehto, T. E. Kottke, and P. Puska. 1986. "Cost- MedicalJoumal 299:1187-89. Effectiveness of the North Karelia Hypertension Program, 1972-1977." Helinekens, C. H., J. E. Buring, Peter Sandercock, Rory Collins, and Richard Medical Care 24:767-80. Peto. 1989. "Aspirin and Other Antiplatelet Agents in the Secondary and Oldridge, N. B., G. H. Guyatt, M. E. Fischer, and A. A. Rimm. 1988. "Cardiac Primary Preventi(on of Cardiovascular Disease." Circulation 80:749-56. Rehabilitation after Myocardial Infarction: Combined Experience of Ran- Hutt, M. S. R. 1991. "Cancer and Cardiovascular Disease." In R. G. Feachem domized Clinical Trials." JANA 260:945-50. and D. T. Jamison, eds., Disease and Mortality in Sub-Saharan Africa. New Oliver, Michael. 1984. "Coronary Risk Factors: Should We Not Forget about York: Oxford University Press. Mass Control?" World Health Forum 5:5-18. Hutt, M. S. R.. and D. P. Burkitt. 1986. The Geography of Non-infectious Olshansky, S. J., and A. B. Ault. 1986. "The Fourth Stage of the Epidemiologic Disease. New York: Oxford University Press. Transition: The Age of Delayed Degenerative Diseases." Milbank Memonal "Hypertension in Blacks and Whites." 1980. Lancet (editorial) 2:73-74. Fund Quarterly 64:355-91. Intersalt Cooperative Research Group. 1988. "Intersalt: An International Omram, A. R. 1971. "The Epidensiological Transition: A Theory of the Study Of Electrolyte Excretion and Blood Pressure: Results for 24 Hour Epidemiology of Populationi Change." Milbank Memonal Fund Quarterly Urinary Sodium and Potassium Excretion." Brinsh MedicalJournal 297:319-28. 49:509-38. Jamison, D. T., J. R. Evans, T. King, I. Porter, N. Prescott, and A. Prost. 1984. Pearson, T. A. 1989. "Influences on CHD Incidence and Case Fatality: Medical China: The Health Sector. A World Bank Country Study. Washington, D.C.: Management of Risk Factors." InternationalJoumal of Epidemiology 18(sup- The World Bank. plement l):S217-S222. Julian, D. 1989. "Treatment fli Chroiiic CHD." Internarinal-Journal of Epide- Pitt, Bertram. 1989. "Therapy for Myocardial Infarction: Effect on Trends in miology 18(supplement 1):S228-S230. Coronary Disease Mortality." International Journal of Epidemiology 18(sup- Kannel, W. B., and Manning Feinleib. 1972. "Natural History of Anginia plement l):S223-S227. Pectoris in the Framingham Study." AmencanJournalofCardiology 29:154- Reducing the Health Consequences of Smoking: Thirty-five Years of Progress: A 63. Report of the Surgeon General. 1989. cLrc-89-841 1. Washington D.C: U.S. Knuiman, J. T., R. J. J. Hermus, and J. G. A. J. Hautvast. 1980. "Serum Total Department of Health and Human Services. and High-Density Lipoprotein (HDL) Cholesterol Concentrations in Rural Rice, D. P., T. A. Hodgson, and A. N. Kopstein. 1985. "The Economic Costs and Urban Boys from 16 Countries." Atherosclerosis 36:529-37. of Illness: A Replication and Update." Health Care Finance Review 7:61-80. 594 Thomas A. Pearson, Dean T. Jamison, and Jorge Trejo-Gutierrez Ross, Russell. 1986. "The Pathogenesis of Atherosclerosis: An Update." New Vaughan, J. P., and W. E. Miall. 1979. "A Comparison of Cardiovascular England Journal of Medicine 314:488-500. Measurements in The Gambia, Jamaica, and the United Republic of Tan- Rothenberg, R.. E. Ford, and E. Vartiainen. 1990. "Cardiovascular Disease zania." Buletin of the World Health Organizanon 57:281-89. Prevention; EstimatingthelmpaccrofInterventions."Paperpresentedatthe Wamer, K. E. 1990a. "Behavioral and Health Effects of the Antismoking International Symposium on Health. Environment, and Social Change, Campaign in the U.S." Paper presented at the Intemational Symposium on July, Taipei, Taiwan. Health, Environment, and Social Change. July, Taipei, Taiwan. Schoenbetg, B. S. 1979. "Epidemiology of Cerebrovascular Disease." Southern . 1990b. "Tobacco Taxation as Health Policy in the Third World." MedicalJournal 72:331-37. American Journal of Public Health 80:529-31. Shigan, E. N. 1988. "Integrated Programme for Noncommunicable Diseases Weinstein, M. C., and W. B. Stason. 1985. "Cost-Effectiveness of lnterven- Prevention and Control." W'ord Health Statistical Quarterly 41:267-73. tions to Prevent or Treat Coronary Heart Disease." Annual Review of Public Steinberg, E. P., E. J. Topol, J. W. Sakin, S. N. Kahane, L. J. Appel, N. R. Health 6:41463. Powe, G. F. Anderson, J. E. Erickson, and A. D. Guerci. 1988. "Cost and WHO (World Health Organization). 1982. Prevention of Coronary Heart Dis- Procedure Implications of Thrombolytic Therapy for Acute Myocardial ease. Technical Report 678. Geneva. Infarction."Journal of the Amencan College of Cardiology 12:58L-68A. _____. 1986. "The WHO Programme on Cardiovascular Diseases, 1985- "The Surgeon General's 1990 Report on the Health Benefits of Smoking 1986." Division of Noncommunicable Diseases. Geneva. Cessation." Morbidity and Mortality Weekly Report 39(RR- 12), pi-xv, 1-12. - - ' ' . ~~~~~~~~~1988a. Appyropriate Diagnostic Technology in the Managerent of CaTdio Tao, Shouchi, Zhendong Huang, Xigui Wu, Beifan Zhou, Zhikui Xiao, vascular Diseases. Technical Report 772. Geneva. Jiansheng Hao, Yihe Li, Runchao Cen, and Xuxu Rao. 1989. "CHD and Its Risk Factors in the People's Republic of China." International Journal of . 1988h 1988 Worl Health Statstcs Annual. Geneva. Epidemiology 18(supplement 1):5159-S163. . 1989. "Mortality in Developed Countries." Weekly Epidemiological Taylor, W. C., T. M. Pass, D. S. Shepard, and A. L. Komoroff. 1987. "Choles- Record 14:103-7. terol Reduction and Life Expectancy." Annals of Internal Medicine 106:605- WHO (World Health Organization) MONICA (Monitoring Trends and Deter- 14. minants in Cardiovascular Disease) Project. 1988. "Geographical Variation Thomn, T. J. 1989. "International Mortality from Heart Disease: Rates and in the Major Risk Factors of Coronary Heart Disease in Men and Women Trends." IniternationalJoumal of Epidemiology 1 8(supple ment I ):S20-S28. Aged 35464 Years." Word Health Statistical Quarterly 41:115-40. Thom.s S..S F. H. Epstein,J.J. Feldman, and P. E. Leaverton. 1985. "Trends . 1989. "WHiO MONICA Project: Objectives and Design." International in Total Morbidity and Mortality from Heart Disease in 26 Countries from Joumnal of Epidemiology 18(Supplement 1 ):S29-S37. 1950 to 1978." InternationalJournatofEpidemiology 14:510-20. Williams, Alan. 1985. "Economics of Coronary Artery Bypass Grafting." Tracy, R. E., and V. Toca. 1975. "Relationship of Raised Atherosclerotic Bntsh MedicalJournal 291:326-29. LesionstoFattyStreaksinl9Locauon-RaceGrouips."AAtheroscLerosts 21:21-36. Yusuf, Salim, Janet Wittes, and Lawrence Friedman. 1988a. "Overview of Uemura, K., and Z. Pisa. 1988. "Trends in Cardiovascular Disease Mortality Results of Randomized Clinical Trials in Heart Disease. 1. Treatment in Industrialized Countries since 1950." World H.ealth Statistical Quarterly Following Myocardial Infarction."JAMA 260:2088-93. 41:155-78. - 1988t. "Overview of Results of Randomized Clinical Trials in Heart LINICEF (United Nations Children's Fund). 1989. "Essential Drugs Price List." Disease. 2. Unstable Angina, Heart Failure, Primary Precautions with UNICEF Supply Division, Copenhagen. Aspirin, and Risk Factor Modification." JAMA 260:2259464. 24 Chronic Obstructive Pulmonary Disease J. Richard Bumgarner and Frank E. Speizer Chronic obstructive lung (pulmonary) disease refers to several affect the heart, leading to right-sided heart failure. There is disease manifestations and is known by many names-cor no cure for emphysema-once tissues have been degraded the pulmonale (heart disease with an underlying pulmonary defi- loss of ventilatory function is irreversible. Diagnosis is nor- ciency), right-sided heart disease, asthma, emphysema, mally accomplished through measurements of abnormally pre- chronic bronchitis, and peripheral airways disease are among mature and permanent declines in lung function, but clinical the most common. Often these names refer to one or another confirmation has been dependent upon the pathologic exam- of the pathological manifestations which are here grouped as ination of lung tissue, generally obtained by autopsy. Newer, chronic obstructive pulmonary disease (COPD) and its allied but expensive, imaging techniques-computerized tomogra- conditions. The many and varied names indicate in part the phy (cT) scanning-now make it possible to confirm diagnosis uncertainty in the clinical diagnoses of these conditions as well of emphysema while the patient is alive. Cor pulmonale is as the fact that the respiratory tract has a limited number of characterized by clinical evidence of right-sided heart failure ways to respond to injury: the obstructive pattem is the unify- with edema (fluid retention in the limbs). This condition is ing manifestation of these conditions. generally superimposed on severe obstructive airways disease and long-standing respiratory insufficiency with hypoxia and Description of the Diseases hypercapnia. Chronic obstructive lung disease not otherwise spec- ified. more recently reported in many health systems, generally Because all the conditions of COPD are characterized by means irreversible obstructive disease by physiologic testing airways obstruction, the clinical manifestations and defini- but does not specify the type of disease. A great deal of this tions overlap considerably, and diagnosis and measurement disease category is probably chronic bronchitis with obstruc- of their progression have not in the past, unfortunately, tion or emphysema. Collectively, these diseases are referred to been obtained in a uniform manner. For example, chronic here as COPD. bronchitis involves inflammation and narrowing of the large Persons suffering from one or more manifestations of COPD bronchial passageways to the lung. It is accompanied by have abnormally rapid rates of decline in lung function to severe coughing caused by the hypersecretion of mucus in levels which are severely disabling by middle life and fatal in the inflamed passages. The cough sounds terrible and is severe cases. In addition, COPD contributes to the severity and discomforting to the patient, but in itself it is seldom seri- eventual fatal outcome of cardiovascular diseases, including ously disabling. The condition is not normally fatal, provid- coronary and rheumatic heart disease, and other respiratory ing there is good access to medical care to control primary diseases, such as pneumonia and pulmonary tuberculosis. inflammation and to avoid the consequences of secondary Other diseases affect and can also complicate declines in infection. Usually the condition abates when the cause of respiratory function. Asthma is induced by a wide variety of the inflammation is removed. Clinical information about allergic and nonallergic agents and produces widespread in- this condition is usually obtained from a standard set of flammation and narrowing of the airways. It is characterized questions about cough and sputum production. Emphysema by a history of recurrent episodes of wheeze, with or without is best characterized by a history of shortness of breath shortness of breath, but with reversible airways obstruction. Its resulting from progressive destruction of lung tissue. In onset can be sudden and severe. Treatment with broncho- sequence there is a loss of elastic structure followed by the dilators or anti-inflammatory agents can provide rapid return destruction of alveolar walls and the collapse of smaller of normal lung function. In many patients lifestyle and produc- airways. This eventually results in the loss of ability to tivityremainsnormaluntilthenextepisode.Otherpulmonary transport oxygen from the airways to the blood and the conditions, such as pneumonia, pneumoconiosis, silicosis, and reduced rate of exchange of carbon dioxide from the blood byssinosis can complicate or exacerbate underlying COPD. These to the airways. This change in diffusion may eventually diseases, however, are distinct from COI'D. 595 596 J. Richard Bumgamer and Frank E. Speize-r Etiology of COPD together the approximate natural history of lung function in a healthy individual. They have found that it increases with age The pathogenesis ofCOPD is not yet fully understood. Two main until about the mid-twenties, when it begins a slow, natural mechanisms have been postulated (airflow obstruction and decline. The lungs have a large ventilatory reserve, and the mucus hypersecretion), and these are believed to be indepen- decline does not normally become evident as a significant limit dent but overlappingdisease processes (Peto and others 1983). on lifestyle or work capacity (curve A; see figure 24-1). Most First, airflow obstruction can be caused by the presence of persons who fall between 35 percent and 50 percent of the excessive amounts ofelastase, an enzyme that is responsible for predicted value of FEvl become short of breath on minimal degrading elastin in the lung and destroying alveolar walls, exertion, and to a variable degree this may lead to a complaint resulting in emphysema. The absence of alphal-antiprotease, to their health care provider. Because decreases of function to a protein that acts as an elastase inhibitor, is one genetic model this level generally occur only gradually, the definition of the for the occurrence of emphysema. In addition, it is known that onset of disease, when such a complaint is made, is also the inhalation of particulates in smoke results in an inflamma- variable. This is in sharp contrast to the definition of onset of tory response in the lung, which increases production of a myocardial infarction or the diagnosis of a cancer. For CPD elastase. At the same time, it has been shown that cigarette it is thus difficult to define the transition between health and smoke acts by oxidizing alphal-antiprotease, resulting in the disease. removal of the natural control on elastase production (U.S. Only a relatively small group (approximately 20 percent) of Surgeon General 1984). These processes and other unknown the population (mostly, but not all, smokers) reach a level of enzymatic processes may interact and lead to the resulting lung function associated with disability. The remaining non- destruction of lung tissue and the condition of emphysema. smokers and smokers simply do not live long enough to become Second, the muco-ciliary apparatus of the respiratory tract disabled from lowered levels of lung function. Smokers whose is a natural defense against particulate matter which may be rate of decline of lung function is rapid and who stop smoking inhaled. The cilia, tiny hairlike projections, are coated with a in early mid-life do not regain a substantial amount of lung thin layer of mucus that envelops entering foreign particles. function, but their rate of loss becomes more like that of a The cilia beat in waves about 1,000 to 1,500 times each nonsmoker (curve D and/or curve E in figure 24-1); if they stop minute, propelling foreign particles upward to the trachea. smoking soon enough they may not reach the disabling level Many pollutants, including cigarette smoke, cause transient of lung function in a normal lifetime. This does not mean that paralysis of the cilia. Over a long period, the cilia may be such a smoker will be protected from the other ravages of permanently injured by such pollutants. In addition, chronic smoking. irritation increases production of mucus by the bronchial mucus glands. The thick, excess mucus not only overwhelms Figure 24-1. Theoretical Curve Representing the cilia but may also plug the bronchioles, resulting in the Varying Rates of Change in FEV by Age development of chronic bronchitis with recurrent lower respi- ratory tract infections and increased morbidity from airways FEV1 (percent of value at obstruction (Carnow and others 1970). Protection of the alve- age 20-25) oli from the particular matter and pollutants is consequently 100 reduced, and production of elastase and greater oxidization of alphaj-antiprotease is likely to be increased. Lung Function and the Development of COPD Lung function can be clinically measured and recorded as 50 - B various indexes of the ability of the lungs to take in and expel air. A simple physiologic test is performed by having patients Disabiity _ ---_ -\\ C_ take as big a breath as they can and blow out as fast and as hard 25 - D as they can into a recording device (a spirometer) that mea- - Death - \\ E sures the volume of air expelled in a specified time. The forced I I I G I expiratory volume in 1 second (FEVi) is a common measure. 5 10 15 20 30 40 50 60 70 80 By comparing the FEVl of individuals with standards measured in large population groups and adjusting for body size (height Age (years) or height squared) and age for each individual, one can calcu- late a percent-predicted (percentage of norm) value for FEVl. Note: Curve A represents normal decline in FEV1. Curve B shows less than Because no single study . has been able.to follow apopulation optimal d evelopment of normal lung function. Often, the disability-related Because no single study has been able to follow a population decline continues as a variable rate curve (C). Curve D shows effect of from early adult life on into the development of COPD in smoking cessation; also seen in disabled individuals (curve E). Curve F is a disability-related decline continuing at variable rate. Curve G represents mid-life to late mid-life, investigators have relied upon rela- the accelerating decline in FEV1 with cigarette smoking and continuing rapid decline until death as a consequence of respiratory failure. tively short-term prospective studies of several years to piece Source: Speizer and Tager 1979. Chronic Obstructive Pulmonary Disease 597 For those patients who do become disabled and do not give sibly with common indoor environmental exposures. The roles up smoking, COPD is a devastating disease that often kills in less and importance of these factors are unknown at present. than five years. Death from respiratory insufficiency would occur at about FEVI at 15-2 5 percent of predicted value. Even Air Pollution before being disabled those subjects with lower levels of lung function at approximately age forty-five have almost a fifty- Both indoor and outdoor air pollution have long been rec- fold higher risk of mortality in a twenty-year period (Peto and ognized to be potentially exacerbating factors for COPD. others 1983) than subjects whose level of function is better Illness in patients with preexisting disease symptoms clearly than average. worsened in association with daily changes in peak levels of Maximal lung capacity is different for each individual. One smoke (Lawther 1970), and overall levels of air pollution question of current research interest in the industrial world is have been recognized to have a short-term acute effect on how to identify which persons, and in particular which smok- persons with the disease. During dramatic episodes of air ers, are at greatest risk of developing COPD. Authors of longitu- stagnation and pollution in London, New York, Japan, and dinal studies in children have observed a high degree of Dublin, substantial excess mortality occurred among the tracking of lung function, leading to the suggestion that risk elderly and in those with preexisting disease who were factors that put children on tracks of lower growth rate of lung exposed to the high concentrations of smoke and sulfur function, such that these children never quite reach their dioxide (Holland 1983). The morbidity resulting from short- maximum predicted level of lung function in early adult life, term exposure could not distinctly be attributed to pollution may explain those persons with more decline in lung function. rather than adverse meteorological (cold temperature and Factors that reduce the rate of lung growth in children are high moisture) conditions. discussed below. The causal role of air pollution in COPD has been examined in several British studies done in the 1950s and 1960s (such as Risk Factors Holland and Reid 1965; Holland and others 1965). Both these and American studies (Reid and others 1964; Deane, Gold- The main risk factors for COPD are several and diverse. smith, and Quma 1965; Wynder, Lemon, and Mantel 1965; Densen and others 1967) have to date failed to demonstrate Cigarette Smoking conclusively a causal link between air pollution and the onset of COPD. Conversely, smoking behavior alone has not been The best-documented cause-and-effect relationship in the eti- sufficient to explain the geographic differences in the preva- ology of COPD is cigarette smoking (Palmer 1954; U.S. Surgeon lence of symptoms in England (Lambert and Reid 1970). General 1984). Numerous studies have confirmed cigarette Prevalence rates for symptoms were found to increase with smoking as a primary cause of COPD. There is a clear dose- increasing levels of air pollution independently of cigarette response relationship between the prevalence of chronic consumption, indicating that atmospheric and indoor smoke mucus hypersecretion and obstructive airways disease and the pollution may account for the urban-rural differences in the quantity of cigarettes smoked (Anderson, Ferris, and data on respiratory morbidity in Britain (Reid and Fairbairn Zickmantel 1964; Ferris 1973). Prospective studies in a number 1958). of countries show much higher COPD mortality in smokers than In developing countries also COPD appears to be unexplain- in nonsmokers. The onset of symptoms associated with COPD able solely by cigarette smoking, because mortality and preva- may occur at an early age and at relatively low levels of lence rates often appear to be much higher than in industrial cigarette smoking (Peters and Ferris 1967). countries and to have more equal sex ratios. For example, the frequency of chronic bronchitis in northern India would ap- Individual Susceptibility pear to be explainable by tobacco use among men, but its prevalence in women may be indicative of the effect of chronic Individual susceptibility must play a role and the multiple exposure to fumes produced during cooking with cow dung, factors involved are not yet fully understood. A hereditary wood, and coal (Malik 1977). In a study of nonspecific lung cause of emphysema was first described by Eriksson (1965). disease in Delhi in which similar studies in London and Chi- A deficiency in alphal-antiprotease is a recessive genetic cago were compared, Saha and Jain (1970) found that 16 trait that in its severest form predisposes subjects to the percent ofDelhi patients were nonsmokers, whereas nearly all development of emphysema even without exposure to ciga- patients in the other two studies were smokers. In addition, rette smoke.2 Those who are heterozygote for the putative gene cigarette smoking and occupational air pollution (including appear to be more susceptible to cigarette smoke. Fortunately exposure to steel, coal, cotton, and other dust) act additively the gene frequency of this condition is relatively low, and the in causing chronic lung disease (Commission of the European condition cannot account for more than a few percentage Communities 1975). Exposure to multiple risk factors in de- points of the total number of cases of emphysema. Other veloping countries may be much higher than in industrial familial factors include increased frequency of allergies, possi- countries, and the effect of these multiple exposures has not bly associated with increased airways responsiveness, and pos- been fully evaluated. 598 J. Richard Bumgamer and Frank E Speizer Childhood Respiratory Tract Infections The Public Health Significance of the Condition Samet, Tager, and Speizer (1983) revieuwed suggestive but not From an epidemiological perspective COPD must rank as a maj or conclusive evidence on the relationship between lower respi- public health problem. ratory tract infections in childhood and the subsequent devel- opment of coPD. Respiratory illness in early childhood has been Mortality and Morbidity shown to be associated with lower levels of lung function in children six to ten years old (Tager 1983; Gold 1989). Among Death from COPD as the primary cause can occur directly Chinese children, passive exposure to cigarette smoke from because of respiratory failure or because of right-sided heart fathers who smoked nearly doubled the relative risk of severe failure. Mortality rates from COPD in many industrial countries respiratory infection in the first eighteen months of life (Chen have consistently been higher for men than for women because 1986). Studies in Papua New Guinea, where from birth people of the longer and heavier smoking experience among men. are exposed to wood smoke in unventilated huts, show high rates of respiratory infection and chronic bronchitis (Colley Figure 24-2. Age-Specific Mortality Rates for COPD and Reid 1970). Other studies support the notion that respira- in United States and China, by Sex tory disorders in children predispose them to later disease (Cooreman and others 1990); and in general, children's lung Rates (per 100,000) function tracks uniformly from early childhood (Dockery 800 1983). As shown on curves A and B in figure 24-1, lower lung function in early adult life would be anticipated to result from Males impaired respiratory functions in childhood. The effect of this 600 / reduced maximally attained lung function may determine the plateau from which further declines during adulthood can be expected. This hypothesis remains unconfirmed, however, 400 because no studies have yet followed individuals long enough to determine whether it is the effects of cigarette smoking or / other factors that put those who have slightly lower lung 200 function by early adulthood at greater risk of developing COPD _-.' . in later life. .._. =-. Occupational Dust Exposure 3 55-64 65-74 Age (years) Becklake (1988) and other researchers have documented an association between other specific and nonspecific occupa- Rates (per thousand) tional dust exposure and excess chronic mucus hypersecretion B00 and obstructive airways disease. Dust exposure appears to exacerbate (but not cause) COPD, and the effects can often Females be managed by improved ventilation or protective respirator 600 - equipment use at work. Socioeconomic Status 400 - Low socioeconomic status may be a surrogate factor for a number of less fully understood risk factors for the development 200 _ / of COPD and has been investigated in a number of studies ._- (Colley andReid [1970] list manystudies). Factors may include _._ . _._.------ - -: higher prevalence of cigarette smoking, poorer nutrition, 0 higher levels of indoor smoke exposure, and poor housing 35-44 45-54 55-64 65-74 conditions, all associated with increases in the frequency of Age (years) other respiratory illnesses, occupational exposure, difficulties in reaching health care services, and less contact with health China 1986 United States 1968 education. Rapidly changing external and domestic factors ----- UnitednSates1985 86 UnitedStates1960 have made long-term cohort studies among such groups diffi- --------- United States 1977 cult and expensive to conduct. Firm conclusions related to these associations remain few even after more than two de- Source: National Center for Health Statistics for United States; Disease cades of work relating to these questions. Surveillance Points, Ministry of Public Health, for China. Chronic Obstructive Pulmonary Disease 599 Death rates from COPD for women are increasing more rapidly 1987). Very limited data from the annual health survey of than for men in the United States as the effect of more Pakistan show nonspecific respiratory disease accounting for widespread cigarette smoking among women is becoming evi- about 5 percent of total morbidity. Jamaica reported death dent. This trend is substantiated by changes in female mortal- rates from bronchitis, asthma, and emphysema of only 8 to 9 ity from lung cancer, now the leading cause of cancer death per 100,000 in the early 1980s (USAID 1987). In Bangladesh in among U.S. women. Total deaths in the United States from 1975-78, these disease categories accounted for 2 percent of COPD have more than doubled from 30,000 in 1970 to more total mortality. In Indonesia, COPD was the third ranked cause than 71,000 in 1986, with increases for each age group (figure of death in the age group forty-five through fifty-four, and the 24-2). The corresponding male-to-female ratio declined from main killer in the age group over fifty-five (Indonesia 1984). 4.3:1 to 1.8:1. Indirectly, COPD contributes to death from a wide Review of epidemiological information from a number of Af- variety of other conditions, including most particularly cardio- rican countries also points to relatively high prevalence of vascular diseases and infectious pulmonary diseases. In many COPD and to the primary need for prevalence surveys with places death reports do not specify contributory causes reliably, consistent criteria and methodology to establish better base- and it is very difficult to estimate the true overall mortality line data (Chaulet 1989). contribution of COPD. In the United States, copiu is estimated Detailed data from China on cause of death provide the best to be a contributing cause of death about 1.7 times as often as perspective of coprn epidemiology from a developing country. 3 it is a primary cause (U.S. Surgeon General 1984). Because of These data indicate that progressively higher age-specific rates the silent, progressive nature of the disease, much underlying prevail just as they do in the industrial nations (figure 24-3) morbidity is neverdiscovered or reported. In the United States, and at rates which are much higher than even in the United COPD morbidity is estimated at about 10 times direct COPD States (figure 24-2). Other data reported by China in 1989 to mortality, and many patients suffer illness and disability for the World Health Organization (WHO) on death among a many years before death (U.S. Surgeon General 1984). This population of 100 million corroborate this and add evidence hidden morbidity burden may be even larger in developing that COPD may be a much more serious problem for the devel- countries in which health care contacts are few. oping world than generally recognized. In China many large- Availability ofgood prevalence and incidence data for COPD scale surveys of respiratory disease document three common in most countries is seriously hampered by a lack ofconsistency manifestations of COPD (cor pulmonale, chronic bronchitis, in reporting. The disease has been redefined and reclassified and emphysema). Cigarette smoking is an identified primary frequently under the international classification systems. In- cause in most studies, but nearly equal sex ratios, a poverty- consistency and weaknesses in recording of underlying causes linked gradient, and dust and smoke exposure show that other of death in most countries further mask COPD's contributory effect and the disability burden which the disease poses. In the Figure 24-3. Rates and Deaths from COPD in China, developing world COPD has been little studied and does not Age 24-3. ResadeDes f9i n rank high on the public health agenda. Yet both relatively and absolutely, it is certainly a more important cause of death and Deaths (thousands) Rate (per 100,000) illness in the developing countries than in industrial nations. 120 1,000 Current Levels and Trends in the Developing World 100 / 800 Data on CoPD for the developing world is scarce. Cor pulmo- 80 -/ 6 nale accounted for 20 percent of hospital admissions, evenly /600 distributed between the sexes, for heart disease in Delhi, India 60 - (Pahmavati 1958). Studies in India have shown chronic bron- / i , 400 chitis prevalence to be between 1.5 and 12 percent, and similar 40 - figures have been reported for other parts of the world. In 20 200 Nepal, prevalence rates for chronic bronchitis (about 18 per- I cent male; 19 percent female) are close to parity, and a hospital o 0 survey found prevalence of emphysema and cor pulmonale to 35-44 45-54 55-64 65-74 be 3 percent and 1.5 percent, respectively (Pandey 1984a and Age (years) 1984b). Nepalese data also indicate significant morbidity bur- dens. Among 39,000 inpatient admissions, corn-type diseases accounted for 5.4 percent in 1984. Among 127,000 health post 0 Urban male deaths Urban female deaths visits by patients (fifteen posts, Kaski district), 3.4 percent of -- Rural male aths Rural female rates the illnesses were bronchitis, emphysema, and asthma. In the --------- Urban male rates Urban female rates 210,000 population served by these health posts, these dis- eases, at a rate of 60 per 100,000, were the third leading cause of death in 1985 (Nepal: His Majesty's Government/WHO Source: Disease Surveillance Points, Ministry of Public Health, China. 600 J. Richard Bumgarner and Frank E. Speizer factors are at work (Yan 1989). Smoking of traditional forms traditional and modern risk factors and their relative roles, of tobacco may be part of the explanation. Although no other than cigarette smoking. Trends in developing countries detailed data on traditional tobacco use in China are available, for traditional risk factors are largely unknown. Prevalence of it is common in rural areas and among middle-aged women as cigarette smoking and per capita consumption are following well as men.4 More detailed examination of the cause of death upward trends, sometimes sharply, as is the initiation of smok- data from China also reveals that the high coPLe rates are not ing at younger average ages. simply misdiagnosed rheumatic heart disease or misclassified Costs and the attributable benefit of programs for COPD respiratory disease. prevention must remain educated guesses at best, given the Whatever their cause, the high rural rates in China may absence of much better data on risk factors, disease prevalence, signal a potentially serious development for China and perhaps and case treatment. Three factors will, however, inevitably for other developing countries. The prevalence of cigarette make COPD a much more important disease in the future: smoking in China is increasing, and the age of initiation is falling. The epidemiological effect of the spread of cigarette * COPD is a disease which mainly strikes in middle and old smoking in the last decade cannot yet be fully reflected in the age. As populations in the developing world age, the abso- data shown in the charts, and large increases in smoking- lute number of COPD cases and deaths will increase. With induced COPD must therefore inevitably lie ahead. Declines in regard to absolute demand for care, and in comparison with chronic nonspecific respiratory illness, which China and other infectious and childhood diseases, COPD will become more countries might be expected to achieve as a result of reduction important. of poverty, fewer respiratory infections in childhood, cleaner * Factors which are not well understood cause high rates indoor air, and improved living standards generally, could be of COPD in many developing countries. The longer the offset, even overwhelmed, with increased morbidity and mor- exposure that aging populations have to risk factors, the tality attributable to a group of smoking-related diseases, in- greater probability they have of developing disease. Con- cluding much higher incidence of COPD. versely, prospects for reductions in these risks are uncertain Much of the present burden must be related to indoor and at best. outdoor smoke exposure, poor living and nutritional condi- * Cigarette smoking continues to increase in all parts of tions for some children, and childhood respiratory infections, the developing world and probably will add to the existing even though today we cannot conclusively document the age-specific rates, perhaps rapidly and substantially. causal role and interaction of these factors. Indoor air pollution from soft coal, wood, and dung used for household heating and These three factors allow at least a very rough estimation of cooking has been demonstrated to cause high levels of indoor the future course and seriousness of COPD. In table 24-1 we smoke pollution in the absence of adequate ventilation and provide an estimate of COPD mortality by major region of the chimneys and can add substantially to the level of outdoor world for 1985. We also show the increases by region in the smoke. Clifford (1972), in his study in the Kenyan highlands, number of COPD deaths which can be expected if age-specific estimated the exposure to (mainly indoor) airborne total sus- rates remain as they are today. These increases are the result pended particulates at 25,000 milligrams per year. Similarly, of the demographic shift that is taking place as more people Smith and others (1983) showed that women in kitchens in live to middle and old age because they are no longer dying Gujarat, India, were inhaling levels of benzopyrene, a potent from infectious diseases at younger ages. This increase in COPD carcinogen present in cigarette smoke, equivalent to smoking twenty cigarettes a day. Although the carcinogenic effects of this smoke level may have no direct link to COPD, they are Table 24-1. Estimated and Projected COPD indicative of the levels of smoke which must be present in such Mortality by Region, Considering Prospective "micro" environments. Overall, these examples strongly indi- Demographic Changes Only cate that copr is a much heavier burden on the poor in (thousands) developing countries, and on poor women and girls in partic- Year ular, than the literature has thus far generally recognized. Region 1985 2000 2015 Industrial market economies 205 255 316 The Progression of COPD in Developing Countries Industrial nonmarket economies 109 150 175 Latin America and the The prevalence of COPD in most developing countries is today Caribbean 18 28 43 only roughly known, despite the evidence summarized above Sub-Saharan Africa 57 90 145 that coPD is a significant cause of morbidity and death in many Middle East and North Africa 21 33 51 parts of the developing world. Data on the prevalence of COPD Asia and the Pacific 510 788 1,155 and its risk factors and on COPD deaths are not adequate for World 926 1,328 1,954 most countries to permit even crude quantitative estimates of Source: 1985 data from paper prepared for World Bank Health Priorities the overall economic effect of COPD and the benefits of its Review 1989; future deaths derived by using 1985 age-specific rates applied to prevention. Too little is yet known about the mix of both population cohorts presented in World Bank Population Projections 1987-88. Chronic Obstructive Pulmonarv Disease 601 deaths (and morbidity) that is induced by age structure will, of age groups in the United States from the mid-1950s to the course, be more pronounced in those areas of the world where mid-1980s. It entails the simple assumption that similar etio- present-day societies are mainly young and the number of logical effects will be evident among large populations else- middle-aged and old persons will grow sharply in the next where who smoke. thirty 30 years. The industrial countries are least affected by In developing countries it is possible that age-specific rates changes in age structure as these have already largely taken will rise even more rapidly than they did in the United States place. Overall there may be more than a doubling of COPD for three reasons. First, during the latter half of the period used mortality in the developing world. to derive these data, there were sizable reductions in the The second factor which will influence future coPn inci- United States in the number of cigarettes smoked and in dence and mortality is exposure to risk factors. It is clear that smoking prevalence. Similar trends may not occur in develop- there are multiple risk factors for COPD in the developing ing countries. In fact, many poor countries are experiencing countries, which can be only poorly documented or ex- rapid growth in both per capita consumption and in overall plained. These include early childhood respiratory infec- smoking prevalence. Second, the effect of increased cigarette tion, exposure to indoor smoke, air pollution, occupational smoking in a population with lung function already impaired dust exposure, and others. It is at least a plausible hypothesis by other long-standing risk factors is likely to be more severe that some of these risks will decline in time with general than smoking among a relatively unimpaired population such economic improvement, better access to primary health as that of the United States. Third, to the extent that poten- care, improved housing and living conditions, better nutri- tially offsetting improvements in living standards, health care, tion, and other changes. Factors such as these may underlie and so on are achieved in the developing countries during the the substantial decline in COPD prevalence in Britain among next thirty years, the beneficial effect on COPD mortality will nonsmokers in the first three-quarters of this century. It is be limited. The irreversible nature of most COPD and the effects equally plausible, however, that persistent poverty, malnu- of impaired respiratory development and of earlier lifetime trition, inadequate health services, poor housing, and rigid exposure mean that for most of those between the ages of social systems in poor countries will preclude early develop- fifteen and fifty today, prospective risk of COPD mortality will ment of a declining trend in COPD prevalence. not diminish rapidly. The effect of reduced risk exposure to Increased cigarette smoking in developing countries will factors other than cigarettes would mainly be realized among substantially affectcoPDrates, particularly among the young in the young and only during later periods of time, after 2015. If Asia, who in many places are becoming early and heavy risk factors other than smoking do not decline in importance, smokers. In table 24-2 we show an estimate of COPD mortality or if the synergistic interaction of these risks and smoking is which may occur from the combined effects of an aging popu- greater than the sum of their individual effects, then possible lation and increasing age-specific death rates resulting from COPD mortality may be considerably higher than indicated in cigarette smoking. It is assumed in these estimates that risks table 24-2. associated with other (traditional) factors remain constant and The conclusion is that very early, aggressive, publicly funded do not multiply cigarette-associated risk. The estimated programs of smoking cessation provide the only significant growth in age-specific mortality rates used to derive these hope of reducing this burden for the developing world. Efforts numbers is similar to that experienced by smokers in the same to address the other possible risk factors may contribute to reducing future morbidity and mortality from COPD, but the strategic mix of actions and their effect is much less clear. Table 24-2. Estimated and Projected COPD Exactly the opposite is true for smoking-the steps and their Mortality by Region, Considering both Prospective probable effect on future COPD is quite clear. Demographic Changes and Epidemiologic Changes Data to estimate present or prospective costs of the COPD (thousands) illness and mortality burden for different parts of the world do Year not exist. In industrial countries, efforts to quantify the effect Region 1985 2000 2015 of the disease have shown its large cost. In Britain, COPD accounted for a rate of 3.7 percent of the working-age popula- Industrial market economies 205 417 529 tion being incapacitated for work, or about 25 percent of the Industrial nonmarket economies 109 231 293 total inception rate for work incapacity (Alderson 1967). A Carnbbean 18 57 72 total of 300 million working days were lost annually in that Sub-Saharan Africa 57 191 243 country as a result of incapacitating illness lasting four or more Middle East and North Africa 21 68 86 days; this directly cost the National Insurance Fund more than Asia and the Pacific 510 1,524 1,934 220 million (in 1965; approximately 2 billion in 1991 World 926 2,446 3,104 prices).5 Of the total, 40 million days lost (13 percent) were attributable to COPD ("Incapacity for Work" 1966). Note: Projections based on table 24-1 but with deaths in 2000 and 2015 In the United States in 1979, COPD was estimated to cost the adjusted to account for age-specific COPD rate increases as experienced in the United States from 1960-85. economy about $6.5 billion (USDHHS 1982). Of this total, about Source: See table 17-1; World Bank Population ProjectionrL 1987-88. a third each was accounted for by direct treatment costs, 602 1. Richard Bumgamer and Frank E. Speizer indirect morbidity costs (for example, lost wages), and indirect age (older than forty); their needs are discussed in the section mortality costs (premature loss of life). entitled Case Management, below. In hospital studies in China, patients with cor pulmonale For the first group the strategy must aim at preventing or had one of the highest average lengths of stay (thirty-six days reducing the exposure to known and suspected risk factors. in a middle-size hospital; forty-six days in a large hospital), and Elements of this strategy would appear to be four, in order of the disease ranked in the middle of all diseases in costliness to priority: broad and comprehensive tobacco smoking control treat per episode (535 Chinese yuan, or about us$150, in a programs; early and widespread health education programs for middle-size hospital; 795 yuan, or about us$210, in a large both the community and for primary-level health workers; a hospital), equivalent to 48 percent and 72 percent of gross variety of investments and programs to reduce severe indoor national product per capita, respectively, for each patient's air pollution, particularly in the home and among the poor; hospitalization episode (Chen Jie 1986). With regard to direct and limited, focused programs to reduce severe exposure to costs attributable to treatment, COPD seems sure to present a workplace or industrial air pollution. Good case management large burden on already underfinanced health systems in many of other diseases may provide good primary prevention of COPD developing countries. The indirect economic losses due to by the identification of children at increased risk. This can be incapacity for work and premature death are almost certain to determined by identifying children who suffer from frequent be higher. In China, COPD is estimated to account for about 2.5 respiratory illness and who are failing in some way to grow and million premature years of life lost annually (3.2 percent of the develop normally. Efforts need to be directed not only to caring total premature years of life lost). Undiscounted, this would be for their respiratory illnesses but also, at a social level, to roughly equivalent to losses of us$750 million annually, con- improving nutrition and reducing indoor smoke exposure-for servatively valuing each year of life lost at average per capita example, by providing means to vent cooking stoves in the gross national product (World Bank staff estimates). Much fur- child's household. As the children get older and are in school, ther work in health economics and accounting needs to be done an integrated program of health education that emphasizes not to quantify better the effect and costs of COPD, particularly losses smoking needs to be instituted. and costs resulting from morbidity, to give public health leaders the facts they need to devise and defend an effective strategy for Primary Prevention Strategy Elements COPD prevention and case management. It seems clear, however, that both the present and prospective costs ofCoPD for develop- Because of the nature of these strategy elements, multiple ing countries are much higher than commonly realized. agencies of government and of the community would need to be involved. In order of priority these would include the Lowering or Postponing Disease Incidence following: * Financial, planning, health, agriculture, industry, and com- Clinical studies have shown that, once lung damage has oc- curred, cessation of smoking can only arrest the rate of further merce authorites, who can provide the most important ele- decline. Changes in venrilatory function that have occurred ments of a prevention strategy for COPD by adopting and to that point are essentially irreversible. There are no known implementing a cohesive tobacco smoking control program, studies to show whether removal of risk from indoor (non- probably consisting of at least the elements indicated in the cigarette) smoke exposure or from outdoor smoke pollution next paragraph. has a similar, immediate, salutary effect, but it seems reason- * Community, religious, and othercitizengroups, which need able to assume so. Avoidance of exposure to smoke and other to stimulate and cooperate with public health authorities to respiratory insults by patients suffering from ventilatory de- target women and their children who are subject to frequent cine has been shown to reduce acute attacks, complications, respiratory infections. The range of their activities should and premature death. include school, matemal and child health, family planning, and nongovemmental programs to improve childhood nu- Elements of a Preventive Strategy trition. Such action will build resistance to infection, a likely risk factor for impaired lung growth. The long incubation period of COPD and its silent, progressive, * Urban and rural development and housing authorities, who and irreversible nature require that prevention strategy be are responsible for, or who can influence, the design of founded on very early, continuous primary prevention efforts. homes and apartments to reduce indoor smoke. Prevention strategy must have two broad population targets: * Health authoities, who must provide the epidemiological persons who do not yet have detectable signs of excess deteri- and professional inputs for targeting of risk groups, for con- oration of lung function and persons who show early and tinuing operational research into causes and effective inter- moderate clinical signs of disease. Most persons in the former ventions, and for the technical content of health education group will he children, young adults, and those adults who have efforts. not been long exposed to known risk. Persons in the second * Education authorities, who must develop means and pro- group will almost all be adults, most probably already in middle grams to ensure broad and early introduction of appropriate Chronic Obstructve Pulmonary Disease 603 health education content as the influences of "modem," Freeaccessofgirlstoprimaryeducationmayofferthebesthope especially "prosmoking," images grow. of strengthening their position in society and their capacity to bring about eventual change in social practices which put TOBACCO SMOKING. The main risk factor for development them at risk not only of respiratory illness but of other diseases of COPD, tobacco smoking, is common to many of the other and injuries as well. chronic noncommunicable diseases that are important causes of mortality and morbidity in the developing world. OUTDOOR AIR POLLUTANTS. The less-than-clear causal link of Neither the etiology of these diseases nor the basis for a outdoor air pollutants to chronic lung disease mediates against preventive strategy are well understood in many govern- aggressive, expensive strategies solely on health grounds. Con- ment circles. Moreover, cigarettes are often associated with siderations of other aspects of environmental degradation and governmental revenue, powerful individuals in trade and of other benefits from reductions in the levels of pollution will commerce, and popular images of development and sophis- provide additional justification and perhaps contribute to bet- tication and are backed by sophisticated methods of propa- ter overall strategy formulation than if only COPD prevention ganda to encourage the initiation of smoking. International aspects are considered. This expanded view would call for a experience with tobacco control shows the necessity for a coordinated strategy by economic, urban, housing, agricul- comprehensive, prioritized approach of tax (price) increases tural, forestry, educational, social, and health authorities to and legislation and regulation of access to tobacco and of its ensure that well-considered programs of better fuel utilization advertising. Good epidemiologic surveillance, analysis, and and atmospheric pollution reduction are developed. The pros- effective forms of health education and publicity need to be pects of these programs having salutary health effects are good, aimed at creating a social environment supportive of cessa- but the many other considerations call for a broad, multisecto- tion of smoking by individuals. The overall scope of these ral strategy. strategy elements goes far beyond the responsibility of any Government programs to affect potential indoor and out- single government agency in any country and requires the door air pollution as a risk factor will be difficult to implement informed attention and action of the highest political and and may be costly if"alternative fuels" and housing reconstruc- community leadership in each country. Tobacco control tion programs are given high priority. In conditions of poverty efforts also will provide one of the only country-effective and restricted opportunity, resources needed for such programs primary prevention strategies for lung, larynx, oral, esopha- may be more effectively used in other ways. Realistic strategies geal, bladder, pancreatic, and kidney cancers. It will sub- would seem to include concentrated antismoking efforts, cou- stantially reduce risks of coronary heart disease, stroke, fetal pled with well-financed health education programs to counsel mortality and spontaneous abortion, prematurity, subse- nutritional, social, and behavioral changes which reduce the quent respiratory distress syndrome death, low birth weight, risk to the poor without adding to their economic burden. and subsequent infant morbidity. The implications of to- Social and health retums from this strategy are likely to be bacco control mean that this difficult topic needs to be near higher. the top of the health policy agenda in most countries. These efforts call for strong social and political leadership on the importance of tobacco control in the developing world INDOOR SMOKE EXPOSURE. Indoor smoke exposure is related together with substantial, sustained, and targeted public fund- both to fuels and their price and to housing styles and familial ing for health and social education for the poor-a goal which customs and expectations. Poor persons will generally be those has proved virtually unattainable in even the richest of socie- most necessarily reliant on the cheapest and potentially most ties unless a national consensus is reached on the definition of toxic fuels (wood, soft coal briquettes, and animal dung). They problems and strong leadership effectively directs the focus of will also be those who are limited to the most simple housing intervention. Examples are provided by earlier success in designs and among whom social practices may dictate addi- China, Sri Lanka, Kerala State in India, and in parts of Latin tional risk-for example, lack of ventilation, attributable to America with control of some endemic and infectious diseases, beliefs that windows may allow evil spirits to enter or to efforts by the worldwide eradication of smallpox, and by the impend- to ward off insects, and customs requiring that women and ing eradication of polio. young children, especially girls, spend considerable time in- doors, often in the rooms in which cooking or heating fires are Case Management located. Substitution of cooking gas and kerosene for powdered charcoal briquettes, dung, and wood could help to reduce The management of patients with COPD is complicated by the indoor air pollution, but problems of expense, distribution, and difficulties in defining when a subject becomes a patient. Most adaptation of customs will surely impede widespread imple- patients seek medical attention when they have sufficient mentation in many countries. Simpler, and likely cheaper difficulty from shortness of breath that it interferes with their solutions, such as reduction of indoor smoke exposure through activities of daily living. Persons with sedentary jobs may better ventilation, seem likely to have a more significant effect perceive themselves to be less affected by the disease than on illness and premature death in the short and medium term. those who work at manual labor and may seek help later in the 604 J. Richard Bumgarner and Frank E. Speizer development of the disease. At present the only objective Secondary Prevention measure of disease is the degree of reduction of air flow and volume from the values expected by age, sex, and stature. Secondary prevention requires health care providers to begin routinely to question patients about respiratory symptoms in a Case Management Strategies standardized manner and carry out simple measures of pulmo- nary function to identify those subjects at greatest risk of Unfortunately, by the time lung function shows excessively developing disabling lung disease. Antismoking efforts and rapid declines, the disease generally has progressed to a point reduced occupational dust exposure must be emphasized, par- at which most of the loss of lung function is permanent and ticularly among those subjects identified as being at risk. will not be recovered with treatment. This is not to say, Medical treatment of exacerbations requires smoking cessa- however, that treatment is not warranted. Treatment may tion on the part of the patient, the forcing of fluids, and the shorten the duration of an exacerbation of symptoms and may use of a broad spectrum of antibiotics. Only if symptoms persist prevent mortality. If the patients survive, however, they are or worsen does the patient need to go beyond the primary retumed only to their pre-exacerbation state, and then con- provider to a facility staffed by physicians, where the same tinue on an unrelenting downward course unless the putative treatments with the added opportunity for bed rest and fluid risk factor (most often cigarette smoking) is removed. administration would be available. Because little can be offered to patients with established disease, the most cost-effective method of case management Rehabilitation: Management of Exacerbations requires the finding of and intervening in the natural history of the disease at a preclinical stage with simple and relatively Ideally, treatment of exacerbations of COPD depend both on inexpensive procedures that require neither highly trained severity of symptoms and clinical findings and level and sever- technical personnel nor equipment (table 24-3). ity of the obstructive components of the disease. For patients Table 24-3. Case Management, Primary and Secondary Care Prinmary care Secondary care Objective Diagnosis Management Diagnosis Management Secondary Positive findings of respiratory Stop smoking n.a. n.a. prevention symptom questionnaire Abnormally reduced lung Reduce occupational n.a. n.a. function on simple pulmonary dust exposure. function testing (such as Improve indoor spirometry) ventilation. Cure n.a. Cure not possible n.a. Cure not possible Rehabilitation Changes in symptoms Stop smoking Severe respiratory distress Stop smoking and treatment (increased mucus secretion, Administer fluids n.a. Administer fluids of exacerbations breathlessness) Use broad-spectrum antibiotics Maintenance Monitoring symptoms Stop smoking Response to Trial of corticosteroids care for chronic bronchodilators condition Improve indoor n.a. n.a. ventilation Rapid loss of pulmonary Breathing exercises n.a. n.a. function Reduced activities of daily living Palliation n.a. None possible Fluid retention Home oxygen Severe shortness Ventilatory support of breath in hospital only if patient has respiratory reserve Blood gases determination n.a. Not applicable. Source: Authors. Chronic Obstrucdtve Pulmonary Disease 605 with mild cases, removal of the inciting agent (cigarette smok- public programs as long as primary and secondary prevention ing), hydration, and being kept warm may be sufficient to efforts remain underfunded). It must be stressed that these reduce morbidity from a given exacerbation. Antibiotics often measures have no effect on the natural history of the disease are added to such a regime; however, the evidence is weak that but serve to keep patients comfortable and functioning for they contribute substantially to shortening the duration of the whatever time they have. exacerbation. In those patients with more severe disease, hos- These considerations, the costs of treatment, and the long pitalization and ventilatory support may be necessary for treat- period of disablement which often accompanies COPD before ment of acute episodes. At some time during the course of death strongly mediate in favor of effective primary prevention severe disease patients should be given a trial of corticosteroids programs with substantial public funding. In most countries it or at least a test of the degree of reversibility of their airways has been extremely difficult to achieve a consensus in this obstruction. regard. This seems to be at least partly attributable to the Because treatment has such little effect on the course of the complex history and characteristics and multifactorial causes disease the most important strategy in first dealing with pa- of the disease. The diffuse pattern of the disease burden in the tients with established disease is to remove existing risk factors. community, the socioeconomic classes most affected, and the This would most certainly include requiring that the patient slow progression of symptoms have added to the problem of stop smoking; reducing or providing ventilatory protection achieving a consensus of COPD as a disease priority. With better from occupational dust and fume exposure,where possible; and understanding in the last decade of the causes, diagnosis, reducing indoor smoke exposure. appropriate classification, and importance and effectiveness of prevention in relation to treatment altematives, this situation Maintenance Care may begin to change. Chronic obstructive pulmonary disease seems to be one of Maintenance care is required for those patients with increasing the chronic diseases for which broad, well-funded, targeted, development of disability. This can be monitored by the pri- primary prevention programs and health education should be mary provider through evidence of increasing symptoms- adopted by government agencies. The efforts that seem likely "rapid" loss of pulmonary function and decreased activity in to be most cost-effective are those provided in the context of daily living. Again the treatment is to have the patient stop regular primary health care and the frequent use of widespread, smoking and to teach the patient self-care strategies with nonhealth mass media (for example, newspapers, television, breathing exercises and graded exercise training. Because a and radio). modest percentage of these patients will have or may develop a reversible component to their disease, an evaluation for Priorities reversibility should be carried out. This can be a relatively simple test of response to a bronchodilator with simple mea- A structured approach to progress in reducing the future bur- sures of pulmonary function before and after the test drug. A den of COPD morbidity and mortality requires prioritization in more formal test of response to a therapeutic trial of broncho- several key dimensions. dilator or corticosteroid lasting several weeks may be appropri- ate in some settings. The latter requires that the health care Priorities for Resource Allocation worker see the patient on several occasions and measure pul- monary function repeatedly. The priorities of primary prevention programs, mainly smok- ing control but also low-technology, low-cost interventions to Palliation reduce exposure to suspected risks, especially among the poor, should be clear from the previous discussions. The limited Palliation requires specialized treatment which in general has effectiveness and high cost of attempts at secondary preven- not been shown to have a significant effect on the course of tion and treatment and cure mean that these strategies will in the disease. Patients with severe fluid retention (cor pulmo- most countries be options only for those who are relatively well nale), severe shortness of breath, hypoxemia, and hypercapnia off. Public financing to deal with COPD illness will be unafford- are at risk of immediate death. There are no cost-effective able for most countries and, if pursued, might only increase measures to deal with these conditions. Only life-saving treat- inequities of health service provision and detract from spend- ment can be rendered, and it is only temporarily effective. ing which should be allocated to primary prevention. Unless the prior physiologic state of lung function is known to A number of priority research activities are suggested by the have been at a level compatible with independent functioning, current state of knowledge about COPD and its effect in the ventilatory support (up to and including assisted ventilation) developing world. Basic research should include both epidemi- is not warranted. On rare occasions in which blood gas deter- ology and a number of clinical and biological questions. Col- minations have been carried out and when local geography (for lection of consistent age- and sex-specific morbidity and example, altitude) dictates, home use of oxygen may be re- mortality data on COPD and its course in the developing world quired to support specific patients and may be affordable to is needed to provide a better assessment of the magnitude and some individuals (it cannot cost-effectively be included in trend of the problem. In addition, given the uniqueness ofsome 606 ] . Richard Bumgarner arnd Frank E. Speizer of the risks of respiratory disease to which people in developing * Approximation of the attributable risk of the synergistic countries are exposed, there are some questions which should interaction of exposure to general smoke and to tobacco be explored to investigate respiratory infection or illness as smoke, for both the smoker and the nonsmoker predisposing factors: * Determination of effective modes and health education * How does nutritional status interact with the frequency, messages to convey the risks and causes of COPD to different duration, and severity of acute respiratory infection in the social groups in developing countries first two years of life? * Cross-national epidemiologic studies to confirm and * Does the immune status (both passive and active) affect quantify better the burden which COPD poses for the poor an infant's (less than two years old) or child's (age two to and for women in particular five) response to a viral or bacterial respiratory infection? In addition, because the exposure to risk factors seems to be * What is the relation between these early life infections so high in rural areas and the population at risk is so large, it and eventual development of co--)Pn? may be that the greatest marginal return to national expendi- Similar questions can be applied to other unique environ- ture on air pollution control will be with low-technology mental settings-for example, in the villages in Nepal, strategies in rural areas rather than, for example, through the where cold and high altitude interact with soft coal and purchase of high-technology emission controls forfossil-fueled biornass fuels to produce high levels of indoor air pollution, power plants. In the near term, the most cost-effective means or in Mexico City, where altitude, weather, and high tem- of achieving a reduction in human exposure may well be a perature interact with auto exhaust to produce high levels concentration on the traditional rather than the modern sec- of ozone. tors of the economy. Research to understand better the prior- ities and advantages of this strategy may yield useful insights. Sociobehavioral Research These conclusions must remain speculative until further work is done, however, and this points to the overall need for Sociobehavioral research on the determinants of effective much better operational research on the cost-effectiveness COPD health education programs (social marketing) that are and cost-benefits of COPD prevention and treatment. Today focused specifically on the young, on women, and on rural we should, but cannot yet, analyze quantitatively the cost- populations provide another area in which research may yield effectiveness of alternative interventions to prevent and treat important findings. Subject matter to be covered by health COPD. Neither can we quantify how expenditure on COPD education could include both disease-specific information and prevention and the outcomes of it compare with the costs and primary prevention messages. The importance of early second- effects of other health interventions. Data to permit detailed ary prevention efforts can be conveyed to family members to analyses of these kinds do not yet exist. We can only broadly encourage them to help those who are already at risk to take assess some of the main factors which would determine the steps to stop progression of the disease. Smoking control poli- results of such analyses. cies provide a second important topic of research in the socio- From an overall cost and benefit perspective, the distribu- behavioral field. tion of copD mortality among relatively older people will mean that many future potential years of life saved by COPD preven- Educational Research tion are retirement years or the years close to them, even in developing countries. The long periods of disability preceding Educational research in both pedagogy and effectiveness of death which characterize coPD would mean that effective curricula for in-service training of health workers, and medical prevention and case management efforts could preserve pro- education curricula for new doctors and health workers, could ductivity, minimize dependency, and postpone or avoid com- include specific emphasis on the epidemiology and importance pletely the expensive phases of treatment of COPD. of COPD to developing countries, and on the importance of From a public health perspective, efforts to treat COPD seem primary prevention as the basis for health care strategy. Pri- likely to be hopelessly cost-ineffective compared with primary mary school health education programs need to develop and prevention aimed at the main known risk factors. Rehabilita- test programs which stress healthy lifestyle practices that chil- tion, maintenance, and palliative care for individual patients dren can relate to in a positive way. have no public health benefits. Because morbidity will already be substantial in most patients when they first seek medical Priorities for Operational Research attention, loss of some productivity will have already occurred, and economic benefits will thus also be limited. In comparison, Priorities would appear to include at least the following early secondary prevention can arrest further morbidity, pre- topics: serve productivity, and reduce or at least postpone expenditure for treatment. If conducted in a primary health care setting, * Determination of the COPD risk attributable to both opportunistic screening through simple questioning of those indoor and outdoor air pollution in developing countries suspected of having COPD, followed by spirometry, should allow Chronic Obstructive Pulmonary Disease 607 rapid, low-cost finding of suitable candidates for intensive bronchitis was observed. Chronic airways obstruction also was found to be counseling, job change, house or work ventilation improve- frequent with significant reductions in FEVI values (Malik, Behera, and Jindal ment, or bronchodilator therapy (listed in descending order of 13 A bi)on is 1,000 million. their probable cost-effectiveness). It is less clear, however, that the cost-effectiveness of early secondary prevention would compare favorably with that of References primary prevention. It seems probable that priority should be given to primary prevention for three reasons. First, most Alderson, M. R. 1967. "Data on Sickness Absence in Some Recent Publica- primary prevention efforts for COPD will have an effect on other tions of the Ministry of Pensions and National Insurance." Brinsh Journal of chronic and communicable diseases. Second, the main risk for Preventive SocialMedicine 21:1-6. COPD in the future appears certain to be cigarette smoking. Anderson, D. O., B. G. Ferris, and R. Zickmantel. 1964. "Levels of Air School and public educational campaigns and workplace and Pollution and Respiratory Diseases in Berlin, New Hampshire." Amencan regulatory programs are known to be low cost and many have Review of Respiratory Diseases 90:877-87. proved effective. Third, among the most effective tools for Becklake, M. R., J. Bourbeau, R. Menzies, and P. Emst. 1988. "The Relation- smoking reduction are taxes and fees attached to the sale of ship between Acute and Chronic Responses to Occupational Exposures." In Current Pulmonology, Vol. 9, Chapter 2. London: Year Book Medical tobacco to help reduce demand for cigarettes, particularly Publishers, Inc. among vulnerable low-income groups and youth. It would not Camow, Bertram W., Robert M. Senior, Robert Karsh, Stanford Wessler, and be unreasonable to expect that all primary prevention efforts Louis V. Avioli. 1970. "The Role of Air Pollution in Chronic Obstructive for smoking-related diseases could be fully funded from tobacco Pulmonary Disease." JAMA 214(5):894-99. taxes and thus be highly cost-effective, even profitable, from a Chaulet, P. 1989. "Asthma and Chronic Bronchitis in Africa: Evidence from public health perspective. Epidemiologic Studies." Chest 96(Supplement 3):334s-39s. Operational research in these areas to quantify better the Chen Jie. 1986. Paper prepared for the World Bank bv Shanghai Medical characteristics and patterns of the costs of COPD morbidity and versity, Department of iealth Management and Economics. mortality, and the benefits of avoiding or minimizing these Clifford, Peter. 1972. "Carcinogen in the Nose and Throat: Nasopharyngeal Carcinoma in Kenya." Proceedings of the Royal Society of Medicine 65(8): costs, is clearly needed and would help to establish and under- 682-86. pin effective programs to deal with the disease. Collev,J. R. T., and D. D. Reid. 1970. "Urbanand Social OriginsofChildhood Bronchitis in England and Wales." British MedicalJournal 2:213-17. Commission of the European Communities. 1975. "Researchers on Chronic Notes Respiratory Disease." Medical Symposium 18. Luxembourg. Cooreman, J., S. Redon, M. Levallois, R. Liard, and S. Perdri.et. 1990. During his experience as a Hubert H. Humphrey North-South Fellow at "Respiratory History during Infancy and Childhood, and Respiratory Con' Johns Hopkins University, Sanjoy Ghose of the Umul Rural Health Research ditions in Adulthood." lnternanonalJournal of Epidemiology 19(3):621-27. and Development Trust (Bikaner, Rajasthan). participated in the early stages Deane, M., J. R. Goldsmith, and D. Quma. 1965. "Respiratory Conditions in of research and conceptualization of this chapter. His contributions, enthusi- Outside Workers." Archives of Environmental Health 10:232-331. asm, and dedication are gratefully acknowledged. Densen, P. M., E. W. Jones, H. E. Bass, and others. 1967. "A Survey of 1. The classification of these conditions has changed several times with Respiratory Disease among New York City Postal and Transit Workers." revisions of the Intenational Classification of Diseases. These changes reflect Environmental Research 1:265-86. better understanding of the disease and its forms buit make cross-national and longitudinal data analysis very difficult. Good analysis of COPD mortality and Dockery, D. W., C. S. Berkey, J. H. Ware, F. E. Speizer, and B. G. Ferris, Jr. morbidity has probably been adverselv affected by the classification changes 1983. "Distrbution of Forced Expiratory Volume in One Second in Chil- , v . , , . . s L ~~~~~~~~~~~dren 6 to I11 Years of Age." American Rev)iew of Expi*atorn Disease 128: and their subsequent, gradual, adoption in various national cause-of-death d reporting systems. 405-12. 2. The mechanism appears to be an inability to neutralize the activity of Eriksson, S. 1965. "Studies in Alpha I-Antitrysin Deficiency." Acta Medica protease normally occurring in the respiratory tract. Scandinavica 177 (Supplement 432):1-85. 3. The China data reported are for a 9 million population sample widely Ferris, Benjamin, Jr. 1973. "Chronic Bronchitis and Emphysema: Classifica- distributed across the country and are only for cor pulmonale. death from tion and Epidemiology." Medical Clinics ofNorth America 57(3):637-49. right-sided heart disease, which in the normal cause-of-death reporting for Gold, Diane, I. B. Tager, S. T. Weiss, T. Tosteson, and F. E. Speizer. 1989. China would include most patients with long-standing respiratory insuffi- "Acute Lower Respiratory Illness in Childhood as a Predictor of Lung ciency which had culminated in deterioration of the circulatory system. This Function and Chronic Respiratory Symptoms." American Review of Respira- category excludes tuberculosis, pneumonia, pneumoconiosis, byssinosis, and tory Disease 140:877-84. silicosis. It also excludes deaths diagnosed as respiratory disease for unspecified Holland, Walter W. 1983. "Evidence for the Implication of Environmental causes. Inclusion of these deaths would raise the reported rates by about 75 Factors in the Aetiology of Chronic Bronchitis." Zeitschrift Erkrank. Atm. percent in most age groups. _)rg. 161:130 37. 4. Traditional tobacco smokers elsewhere are known to be at risk. In a study O in rural India, "reverse chutta" smokers experienced high rates ofCOPD. Holland, Walter W., and D. D. Reid. 1965. "The Urban Factor in Chronic "Reverse chutta" smoking involves consuming home-grown tobacco rolled in Bronchitis." Lancet 2:445-48. a semidried leaf with the lighted end inside the mouth during inhalation. Holland, Walter W., D. D. Reid, R. Seltser, and others. 1965. "Respiratory Chronic bronchitis was diagnosed in 33 percent of the chutta smokers; a high, Disease in England and the United States." Archives of Environmental Health 24 percent, prevalence was found in those age thirty-one through forty years; 10:338-43. and of those older than forty-one, a remarkable 49 percent prevalence of "Incapacity for Work." 1966. British MedicalJournal 8:61-62. 608 J. Richard Bumgarner and Frank E. Spezzer Indonesia, Ministry of Health Statistics. 1984. Jakarta. of Chronic Bronchitis in British Coal Miners." British Journal of Industrial Lambert, P. M.,and D. D. Reid. 1970. "Smoking, Air Pollution, and Bronchitis Medicine 30:217-26. in Britain." Lancet. Saha, N. C., and S. K. Jain. 1970. "Chronic Obstructive Lung Disease in Delhi: Lawther, P. J., R. E. Waller, and M. Henderson. 1970. "Air Pollution and A Comparative Study." IndianJournal of Chest Diseases 12(1/2):40-51. Exacerbations of Bronchitis." Thorax 25:525-39. Samer, M., 1. B. Tager, and F. E. Speizer. 1983. "The Relationship Between Malik, S. K., D. Behera, and S. K. Jindal. 1983. "Reverse Smoking and Chrnnic Respiratory Illness in Childhood and Chronic Air-Flow Obstruction in Obstructive Lung Disease." BntishJournal of Diseases of the Chest 77:199-201. Adulthood." American Review of Respiratory Diseases 127:508-23. Nepal: His Majesty's Government/WHo (World Health Organization). 1987. Smith, Kirk R., A. L. Aggarwal, and R. M. Dave. 1983. "Air Pollution and Geneva: Management Group Report. Rural Biomass Fuels in Developing Countries: A Pilot Village Study in India Palmer, K.N.V. 1954. "The Role of Smoking in Bronchitis." British Medical and Implicarions of Research and Policy." Atmnospheric Environment 17(l 1): Journal, 1473-74. 2343462. Pandey, Mrigendra Raj. 1984a. "Domestic Smoke Pollution and Chronic Speizer, F. E., and 1. B. Tager. 1979. "Epidemiology of Chronic Mucous Bronchitis in a Rural Community of the Hill Region of Nepal." Thorax Hypersecretion and Obstructive Airways Disease." Epidemiologic Review 1: 39:337-39. 124-42. 1984b. "Prevalence of Chronic Bronchitis in a Rural Communitv of Tager, 1. B.. S. T. Weiss, A. Mueioz, B. Rosner, and F. E. Speizer. 1983. the Hill Region of Nepal." Thorax 39:33 1-36. . "Longitudinal Study of Matemal Smoking on Pulmonary Function in Children." New EnglandJournal of Medicine 309:8699-703. Peters, ]. M., and B. G. Ferris. 1967. "Smoking and Morbidity in a College-Age Group." Amencan Review of Respiratory Diseases 95:783-89. USAID (UlS Agency for Interational Development). 1987. Review of Jamaica Peto, Richard, F. E. Speizer, A. L. Cochrane, F. Moore, C. M. Fletcher, C. M. Tinker, 1. T. T. Higgins, R. G. Gray. S. NI. Richards, J. Gilliland, and B. uSDHHS(U.S.DepartmentofHealthandHumanSerxices). 1982.TenthReport Norman-Smith. 1983. "The Relevance in Adults of Air-Flow Obstruction, of the Director, Ten-Year Review and Five-Year Plan. Vol. 3, Lung Diseases. But Not of Mucus Hypersecretion, to Mortality from Chronic Lung Dis- NIH Report 84-2358. National Heart, Lung, and Blood Institute, Washing- ease." American Review of Respiratory Diseases 128:491-500. ton, D.C. Reid, D. D., D. 0. Anderson, B. G. Ferris, and others. 1964. "An Anglo-AmeT- U.S. Surgeon General. 1984. The Health Consequences of Smoking: Chronic ican Comparison of the Prevalence of Bronchitis." Bnitish Medical Journal Obstructive Lung Disease. U.S. Department of Health and Human Services, 2:1487-91. Washington, D.C. Reid, D. D., and A. S. Fairbaim. 1958. "Air Pollution and Other Local Factors Yan, Bi-ya. 1989. "Epidemiologic Studies of Chronic Respiratory Diseases in in Respiratory Disease." Bntish Journal of Preveentive Sucial Medicine 12:94-103. Some Regions of China." Chest 96 (Supplement 3 ):339s-43s. Rogan, J. NI., M. D. Attfield, M. Jacobsen, S. Rae, D. D. Walker, and W. H. Wynder, E. L., F. R. Lemon, and N. Mantel. 1965. "Epidemiology of Persistent Walton. 1973. "Role of Dust in the Working Environment in Development Cough." American Revieu of Respirator-y Diseases 92:679-700. 25 Injury Sally K. Stansfield, Gordon S. Smith, and William P. McGreevey Injuries are too often referred to as "accidents," suggesting that though often neglected, are the "adverse psychological and such events are random and have causes not within our con- social consequences" (WHO 1989a, p. 5) of such injurious trol. On the contrary, injuries occur with definable pattems events as rape or child abuse. which help to identify risk factors and thereby imply strategies It is important to distinguish the "pathological outcome" of for prevention. Injury control, including reduction of the an injurious event from its mechanism, or "external cause." frequency, severity, and consequences of injury, can reduce the Figure 25-1 is a schematic drawing of the chain of events of growing burden of injury in the developing world (Smith and injuries; the drawing also emphasizes that opportunities for Barss 1991). Already one of every four to nine persons suffers prevention of injury depend on (a) reduction of the probability a disabling injury each year in developing countries, and it is of an injurious event (through risk reduction) and (b) reduc- estimated that 2 percent of the world's population is currently tion of the severity of injury (through alteration of the nature disabled as a result of injury (WHO 1986). of the event). In contrast, case management depends on (c) In developing countries, injury is frequently viewed as an reduction of the consequences of the injury (through altering inevitable consequence of technological change and eco- the pathological outcome) once that injury has occurred. nomic development. In the view of both business and the Haddon (1980) referred to these three "phases" of injury community, short-term economic gains often outweigh the control as "pre-event," "event," and "post-event." cost of death and disability from injury. Because they are Injuries are categorized by both theirexternalcause (E-codes, socioeconomically and politically disadvantaged, people in which describe the injurious event) and by their pathological developing countries live daily with a risk of injury which outcome (N-codes, which describe the nature of the injury) in would be unacceptable in industrial nations. As emphasized in the WHO International Classification of Diseases (WHO 1977). In the recently adopted "Manifesto for Safe Communities," this table 25-1 we list the pathological outcomes most commonly "inequality in the safety status of an individual in developing associated with selected injurious events (extemal causes). and developed countries is of concern to all countries" (WHO 1989a, p. 7). Figure 25-1. Chain of Injury Events and The heterogeneity of the mechanisms and effects of injury Opportunities for Injury Control has interfered with awareness of its public health importance and thereby hampered the development of comprehensive Prevention Case programs to address this important health problem. Yet as injury emerges as the leading cause of death in more and more countries, there is a growing demand for the development of / national and intemational programs for injury control. "a.. Reduce Definitions - ~~~~~~~~~~~~~~~~~~~ ~ consequences of Definitions /0 \i,injury Physical injury is caused by an acute exposure to damaging energy (mechanical, electrical, thermal, or chemical) or by the sudden absence of essentials (such as the lack of oxygen in drowning, or heat in hypothermic injury). The study of injury Risk uevent Pathological focuses on the acute effects and long-term disability resulting factors L ,/ e)toutcome from the acute injury; therefore it does not include delayed or indirect effects of chronic exposure, such as those from carci- nogenic chemicals (Waller 1985; Robertson 1992). Important, Source: Authors' data. 609 610 Sally K. Stansfield, Gordon S. Smith, and Wldliam P. McGreevey Table 25-1. Pathological Outcomes Commonly injury and according to local patterns of transport and of Associated with Injury Events in the Developing domestic and occupational activities. World Demographic characteristics which are determinants of ex- Pathological outcome posure to risk of injury include gender, age, occupation, and Infectious socioeconomic status. The behavior patterns of males place Injury event Type of injury complications them at higher risk for most injuries (bums are one of the exceptions; see Taket 1986), and there is a trend of increasing Fire Bumn and thermal injury Yes difference between males and females with increasing age. Infanticidal deaths are, however, more common among female Electric shock Bum and thermal injury Yes children. Collision Crush and deceleration injury Yes Adolescents and younger adults, who already sustain more Abrasion and laceration injuries than others, further increase that risk with alcohol use. Dislocation and fracture For example, persons between the ages of sixteen and twenty- Fall Crush and deceleration injury Yes four drive approximately 20 percent of the total vehicle miles Abrasion and laceration traveled in the United States, yet they account for 42 percent Dislocation and fracture of the alcohol-related fatalities (NHTSA 1988). Although there Assault Abrasion and laceration Yes is less such information from developing countries, it appears Dislocation and fracture that the patterns are similar (Wyatt 1980; Sinha, Sengupta, Anoxic injury and Purohit 1981). Age also affects the case-fatality ratio, or Submersion Anoxic injury Yes the risk of mortality once an injury has been sustained, with (drowning) increased mortality observed among both the very young and Poisoning Toxic injury No the very old (Waller 1985; Baker and others 1992). The poor suffer disproportionately from homicide, assault, Source: Authors' data. pedestrian fatality, and burn injury fatality. The death rate from unintentional injury is also twice as high in low-income Different preventive strategies are also implied by the further areas as in high-income areas of the United States (NRC/1OM categorization of the external cause of injury as unintentional 1985). Similar patterns are evident in developing countries, or intentional (homicide, other assaults, and suicide). Reflect- especially in poor urban areas. The demands of economically ing the increasing recognition of the preventability of injury, and politically underprivileged groups for safe products, work- the term "unintentional injury" has been preferred to "acci- ing environments, and communities are less likely to meet with dents" (Langley 1988). success. Psychologic disorders, such as substance abuse, vio- lence (toward both self and others), isolation or withdrawal, Risk Factors and depression, are also more commonly found among popu- lations marginalized by poverty. These disorders, along with just as specific pathological outcomes are seen more commonly the risky behaviors associated with lack of safety education, all with certain injurious events, each injurious event is com- predispose people to injury. monly associated with specific risk factors. Some of the main In addition to affecting people's exposure to risk of injury, risk factors for the injurious events prevalent in developing socioeconomic status often alters the case-fatality ratio once countries are shown in table 25-2. The importance of each of the injury occurs (Baker, O'Neill, and Karpe 1984), in part these risk factors varies, however, for each external cause of because of variation in access to definitive surgical care. In Table 25-2. Selected Risk Factors Associated with Events in the Developing World Injurs event Submersion Risk factor Fire Electric shock Collision Fall Assault (drowning) Poisoning Male sex No Yes Yes Yes Yes Yes No Age Under 15 Yes No No Yes No Yes Yes 15-64 No Yes Yes No Yes No Yes 65 or older No No No Yes No No No Low socioeconomic status Yes No No No Yes No No Hazardous products Yes Yes Yes Yes Yes Yes Yes Alcohol use Yes No Yes Yes Yes Yes Yes Psychologic disorders Yes No No Yes Yes Yes Yes Poor safety education Yes Yes Yes Yes No Yes Yes Source: Authors' data. Injury 611 lower socioeconomic groups both the children (who are often Although there are few studies even in industrial countries, poorly supervised) and their parents (who must often take alcohol is also a risk factor for nonvehicular injuries greater occupational risks) have a higher frequency of both (Wechsler, Kasey, and Thum 1969; Dietz and Baker 1974; fatal and nonfatal injury. In their study of an urban slum in Rio Davis and Smith 1982; Mierley and Baker 1983; cDc 1984; de Janeiro, Reichanheim and Harpham (1989) found an asso- Pleuckhahn 1984; Smith and Kraus 1988). For example, alco- ciation between children's injury and maternal marital status, hol abuse underlies more than one-fifth of nonvehicular maternal stress or depression, and mothers' work outside the trauma deaths in urban areas of Papua New Guinea (Sinha, home. Sengupta, and Purohit 1981; Attah Johnson 1989). Alcohol Among the substance abuse disorders, alcohol consumption and other drug abuse is known to be a growing problem which represents the main avoidable risk for injury. In the United has undoubtedly been widely underestimated in developing States, alcohol-related mortality (ARM) accounts for 4.5 per- countries (Edwards 1979; Wyatt 1980; Weddell and centofdeathsfromall causes (cDc 1990). Unintentional injury McDougall 1981), and its importance needs to be further accounts for 28.7 percent of all ARM and more than half of all characterized by epidemiologic research. years of potential life lost ('PLL) before age sixty-five. Inten- tional injury accounts for 16.8 percent of all ARM and 29.1 Injurious Events percent of all YPLL before age sixty-five. Of fatally injured motor vehicle drivers in 1987, 38 percent were intoxicated (blood The severity of injuries and the case-fatality rate from them alcohol concentration greater than 0.10 percent), a decline depend largely on their cause. Specific external causes and from 44 percent in 1982 (NHTSA 1988). groups at risk for injury vary widely among countries by level Like the other lifestyle features which accompany eco- of industrialization and by occupational and cultural practices. nomic development, alcohol abuse and its adverse health Because of the lack of surveillance systems or population-based effects are beginning to be recognized for their importance studies, there is limited information available about injury in developing countries (Edwards 1979). Although the per frequency, risk groups, mechanism, and outcome. Much of the capita consumption of alcohol in industrial countries has information that has been developed regarding the epidemiol- decreased or at least remained stable, it is clearly on the ogy of injury has been descriptive, based on hospital data increase in most of the developing world. The total avail- rather than more reliable community-based studies. ability of alcoholic beverages in developing countries has increased by 146 percent since the early 1960s (Kortteinen Unintentional Injuries 1988). Per capita consumption and the fraction of annual income spent on alcohol have both increased each year Fatal injuries in the developing world most often result from since 1978 in Singapore (Curry 1989). motor vehicle collisions, burns, poisonings, drownings, and Even more dramatic are the data from the rural Kisii District falls (Manciaux and Romer 1986; Taket 1986; Smith and Barss of Kenya, where Bittah, Owola, and Oduor (1979) docu- 1991). As in the industrial countries (Barancik and others mented that up to 27 percent of randomly selected males and 1983), the leading causes of injury deaths are usually not the 24 percent of females met the WHO criteria (WHO 1952) for same as the most prevalent causes of nonfatal injuries. For alcoholism. Nearly half of male and one-quarter of female example, drowning is a frequent cause of death but an infre- heads of households in a Nairobi slum have been categorized quent cause of nonfatal injury due to its high case-fatality ratio. as alcoholics (Nielsen, Resnick, and Acuda 1989). In addition The main causes of nonfatal injuries include laceration by to predisposition to alcohol-related disease and mortality, in- cutting and piercing instruments and interaction with animals, cluding that from injury, these families are characterized by in addition to minor motor vehicle collisions and falls. Each worsening economic status, higher rates of separation and of the most important mechanisms of injury are discussed divorce, other psychiatric disorders, and premature mortality below; data are included from developing countries regarding among both the adults and their children. risk factors as they may pertain to the selection of intervention The role of alcohol and other drug use in predisposing strategies. people to injury in the developing world is also only beginning to be documented (Patel and Bhagwatt 1977; Jacobs and Sayer MOTOR VEHICLE COLLISIONS. Among injuries resulting in death, 1983). In addition to limited awareness among health and law those due to motor vehicle crashes are emerging as the most enforcement professionals of alcohol as a risk factor, the lack important for people between the ages of five and forty-five in of technology for blood or breath alcohol measurements is a manydevelopingcountries (Jacobs and Sayer 1983; Lourie and constraint to further definition of the problem (Ryan 1990). Sinha 1983; Mohan and Bawa 1985; Ezenwa 1986a, 1986b). In Papua New Guinea, where a few studies have been con- In some countries, particularly the oil-producing countries in ducted, one-third to more than half of fatally injured drivers which the number of vehicles and roads are rapidly expanding, were found to be legally intoxicated, and 69 to 90 percent of motor vehicle collisions rank first among causes of death for fatally injured pedestrians were found to have blood alcohol all ages (Bayoumi 1981). Mixed traffic, which may include levels above 80 milligrams per 100 milliliters (Wyatt 1980; trucks, buses, automobiles, motorcycles, mopeds, rickshaws, Sinha, Sengupta, and Purohit 1981). bicycles, and pedestrians, all moving at different speeds, clearly 612 Sally K. Stansfield, Gordon S. Smith, and William P. McGreetey predisposes people to collision and injury. Pedestrians and epidemiology and causes of poisoning. Traditional healing drivers of two-wheeled vehicles are at especially high risk in practices also account for the patterns of acute poisoning in developing countries, accounting for more than half of all some countries (Joubert and Mathibe 1989). The majority of fatalities (WHO 1987b). cases occur among children (Gaind, Mohan, and Ghosh 1977; Mortality rates per vehicle in Ethiopia and Nigeria in 1978, Banerjee and Bhattachariya 1978) and agricultural workers for example, were fifty times higher than in the United States (Hayes 1980;Jeyaratam,Senevirante,andCopplestone 1982); or United Kingdom (Jacobs and Sayer 1983). Where the use however, suicide accounts for many of the deaths among of motor vehicles is rapidly increasing, mortality related to adults. them is also increasing, as has been observed in Thailand, In Sri Lanka in 1978, for example, the more than 1,000 Papua New Guinea (a fourfold increase from 1969 to 1978) deathsfrompesticidepoisoningalonegreatlyexceededthe572 (Wyatt 1980), and Malaysia (a fourfold increase in the five deaths from polio, diphtheria, tetanus, and pertussis combined years preceding 1975) (Silva 1978). Similar trends have been (Jeyaratam, Senevirante, and Copplestone 1982). The United noted in Latin America and Africa. As development increases Nations Children's Fund estimates that up to 2 million pesti- further and safety improves, the mortality per vehicle or per cide poisonings and 10,000 deaths from such poisonings occur mile traveled typically decreases, although the rate per O00,000 annually (UNICEF 1989); however, recent reports from the population may continue to increase because of increasing Philippines suggest that these figures may represent a substan- exposure. tial underestimate (Loevinsohn 1987). Recent evidence also suggests that subacute poisoning and chronic disability may BURNS AND FIRES. Burns are most prevalent among women result from continuous low-level exposure. Eighty to 90 per- and children, with the great majority occuring in domestic cent of pesticide poisonings are caused by highly toxic prepa- environments (Sowemimo 1983; Bang and Saif 1989; Jamal rations which account for only 4 to 5 percent of pesticide use and others 1989). In Lagos, for example, more than half (56.2 (Xue 1987). That developing countries account for more than percent) of the bum injuries occurred among children less than half of all acute pesticide poisonings and 80 to 99 percent of fifteen years of age. Mortality ranges from 6.7 percent to 35 all deaths from such poisonings, despite their using only 20 percent among patients admitted to hospitals (Datey, Murthy, percent of the world's pesticide (WHO 1987a; Xue 1987), also and Taskar 1981; Sowemimo 1983; Bang and Saif 1989; Jamal indicates that this is an area for intervention. and others 1989). Bums caused from 15 to 45 percent of all Accidental ingestion of kerosene is a prominent cause of injury deaths seen in three hospital centers in India (Datey, poisoning among children (Ramesh, Srikanth, and Parvathy Murthy, and Taskar 1981), where fatality rates among hospi- 1987; Joubert and Mathibe 1989), most typically when it has talized patients range up to 35 percent. been stored in soft drink bottles or other inappropriate con- In many countries, burns are most commonly sustained by tainers. An important cause of death in a number of countries women who work over open stoves or cookfires (Saleh and is carbon monoxide poisoning, often from motor vehicle ex- others 1986; Gupta and Srivastava 1988) and are a significant haust or heating systems, such as those used in Korea, in which cause of death for women of childbearing age. Fatal bums in hot combustion gases circulated under floors may leak into Kanpur, India, occur primarily among young Hindu house- homes (Lee and others 1971). wives, whose unintentional burns are frequently attributed to Inappropriate use of medications is emerging as an impor- open cookfires or overturning stoves and their loose, highly tant cause of toxic ingestions, particularly in urban areas. For inflammable clothing. Approximately half of the women who example, in a recent report from Pakistan, Bhutta and Tahir suffer fatal burns, however, are intentionally burned or forced (1990) describe nineteen cases of loperamide hydrochloride to commit suicide, most often in association with marital (Imodium) poisoning, of which at least six deaths were a result disharmony (Gupta and Srivastava 1988). Untreated epilepsy of inappropriate marketing and use of this antimotility agent has also been shown to be a risk factor for bums in several for childhood diarrhea. studies (Subianto, Tumada, and Margono 1978; Barss and Wallace 1983). DROWNING. Ponds, irrigation ditches, and wells in develop- Cigarette smoking, a frequent cause of house fires and death ing countries represent the danger of drowning for young due to bum injuries in the United States (Mierley and Baker children. In Brazil, drowning occurs most often among chil- 1983; Technical Study Group 1987), needs to be investigated dren between ten and fourteen years of age and is second only for its role in developing countries as a risk for burn injury and to motor vehicle collisions as a cause of death among these death (Smith and Barss 1991). The prevalence of cigarette children (de Mello and Bernardes-Marques 1985). In many smoking, currently on the increase in the developing world, countries in Asia, drowning is the primary cause of injury death will likely continue to increase as cigarette manufacturers face (Ng, Chao, and How 1978; Meade 1980; Selya 1980; Gu and contracting markets in many industrial countries. Chen 1982; Kleevens 1982). Many of the deaths among young adults aged fifteen through twenty-four may represent suicide, POISONING. In many developing countries, poisoning has drowning being a preferred method in Asian countries (Ng, emerged asasignificantcause ofdeath(Smith and Barss 1991). Chao, and How 1978). Drowning is probably underestimated Local industry and agricultural practices often determine the as a cause of death in many countries because so few cases ever Injury 613 reach the hospital or are reported to police (Smith and Barss results in permanent disability in 25 percent of cases 1991). (Krishnarajah 1972). FALLS. Injury from falls is most prevalent in occupational Intentional Injury settings, among young boys, and, where larger such populations exist, among the elderly. Falls from roofs and trees, especially Distinction among injuries by motive has little import for case during the harvesting of fruit, are among the most important management; however, it has clear implications in the selec- causes of fatal and nonfatal injury (Barss, Dakulala, and tion of preventive strategies. Suicide, homicide, and genocide Doolan 1984). The burden to society of these injuries is (including war) are important causes of injury for which pre- substantial, particularly in view of the high incidence of para- ventive strategies must be identified. Terrorism and torture are plegia resulting from spinal cord injury. In Hong Kong, falls threats or acts designed to coerce individuals or groups, often accounted for 32 percent of all trauma patients who were resulting in long-term social and psychological injury. Rape discharged from the hospital, and hospitalized patients alone and child abuse also inflict psychological injury which can be accounted for a reported rate of 416 cases of fall injuries per considerably more disabling than any associated physical in- 100,000 per year prior to 1979, a figure even greater than that jury. Moredramatically than in any othercauseof injury, males for motor vehicle accidents (Kleevens 1982). are most commonly the actors in such interpersonal and inter- group violence. Women and children are frequently the vic- OTHER UNINTENTIONAL INJURIES. Animals, bicycles, and cut- tims (Chelala 1990). Locally important factors predisposing ting tools are additional important instruments of injury people to intentional injuries, such as poverty, racism, social in most developing as well as industrial countries (Smith isolation, and drug and alcohol abuse, should be investigated and Barss 1986, 1991), imposing a considerable burden on as risk factors (Rosenberg and others 1987). the health system. Gordon, Gulati, and Wyon (1962) found Criminal homicide represents a significant proportion of that 13 percent of injury deaths in rural India were linked injury deaths in many parts of the world. For example, to infectious complications of such minor injuries, suggest- homicide rates of 8.2 per 100,000 population are observed ing that many deaths could be prevented by simple in Latin America; comparable figures are 6.7 for the Carib- interventions such as proper wound care and tetanus im- bean, 4.7 in North Africa and the Middle East, and 2.3 per munization. Permanent disability resulting from ocular 100,000 in Asia. Such intentional interpersonal injury is trauma, which is responsible for 2.4 percent of all bilateral most prevalent in urban areas throughout the world. In the blindness in Nepal (Schwab 1990), represents another im- area surrounding Bangkok, for example, homicides are the portant preventable loss of productivity. leading cause of death due to injury, accounting for 27 Manmade disasters, such as the chemical leak in Bhopal and percent of injury mortality (WHO 1987a). The mortality rate the meltdown at the nuclear plant at Chemobyl, underline the associated with firearms was less than 1 per 100,000 in the importance of chemical agents and nuclear energy in human United States in 1980; it was nearly 30 per 100,000 in Sao injury (Bertazzi 1989). These and the natural disasters, such as Paulo, Brazil, in 1984. In neighboring Colombia, the homi- droughts, earthquakes, and floods, frequently crystallize local cide rate per 100,000 inhabitants rose from about 20 in the and international response more effectively than the less cat- early 1970s, before drug trafficking became such an import- astrophic but more common causes of injury death. Yet the ant problem, to more than 50 in 1987 (Losada Lora and more than 5,000 reported disasters in the last two decades have Velez Bustillo 1988). affected more than 2.3 billion lives and resulted in more than Suicide is also probably more important than is suggested by 4 million deaths (CRED 1991), most of which are due to injury. currently available statistics. In Sri Lanka, forexample, suicide These episodic calamities remind us periodically of the greater is the most common cause of injury death, with organophos- toll of injury in settings where capacity is limited to predict, phate pesticide poisoning a frequent method of choice (Sri prepare for, and respond to such events. Lanka Psychiatric Association 1982; Berger 1988). The importance of armed combat as a cause of injury mor- OCCUPATIONAL INJURY. Injuries sustained in the workplace, bidity and mortality cannot be ignored. Since 1980, forty-five primarily impact and overexertion injuries, are more frequent countries have been involved in forty wars with more than 4 and perhaps more severe in the developing world. The death million soldiers globally. More recent wars are tragically dis- rate for factory workers in India, for example, is 50 percent tinguished by the occurrence of the majority of the mortality higher than that in the United States (Mohan 1982). The (80 to 90 percent) among civilians, most of whom were women injury rate among coal miners in Nigeria is seven times that and children. More than 1 million people have perished in for the same occupational group in Britain (Asogwa 1980). In Uganda alone during the last twenty years of political unrest. Brazil during 1970, nearly 18 percent of industrial workers were For every death, three times as many people sustain a nonfatal injured (Pupo Nogueira 1987). More than a quarter of indus- injury (Werner 1987). The more indirect effect of such strife trial workers in Mexico experience a disabling injury each year on health status-diversion of national resources to defense (Cuellar 1980). The injury of workers in Sri Lanka occurs at a from health care-has been pointed out by Ogba (1989) in rate of 1,000 per 1,000 or one for every worker per year and Nigeria and Chelala (1990) in Central America. 614 Sally K. Stansfield, Gordon S. Smith, and William P. McGreevey The Public Health Significance of Injury The rates of nonfatal injury in developing countries are probably as high as or higher than those observed in the Worldwide, injury ranks fifth among the leading causes of industrial world. Data from the United States suggest that for death, accounting for 5.2 percent of the total mortality (Man- every fatal injury there are 16 hospitalizations and almost 400 ciaux and Romer 1986) and 10 to 30 percent of all hospital injuries serious enough to restrict activity or require medical admissions (WHO 1988). One review of global survey data has treatment (Rice and others 1989). It is likely that there are also suggested that one child in every five to ten sustains an injury several hundred nonfatal injuries for every fatal injury in each year (Manciaux 1984). A summary of global age-specific developing countries. Because of the lack of available survey pattems of mortality from injury and poisoning is presented as methods for identifying and quantifying disability, the real table 25-3. We developed these estimates and projections economic and social effect of these injuries is unknown. using the methodology outlined by Alan D. Lopez in chapter 2 of this collection. TRENDS IN THE PERIOD 1970 TO 1985. Observation of trends in the epidemiology of injury in developing countries raises the Current Levels and Trends in the Developing World question of the relationship between development and injury. Omran (1971) pointed out that developing countries move Injury morbidity in developing countries is more difficult to through an "epidemiologic transition," from a disease profile ascertain, because of the lack of adequate community-based dominated by infectious diseases to one characterized by the studies of injury (Smith and Barss 1991). Data collection "postransition" noncommunicable health problems, including regarding the incidence of milder injury is further hampered injury. This transition is brought about through development- by the absence of a consistent case definition for injury or associated evolution in three important determinants of the disability. In the United States, it is estimated that 1 in 4 pattern ofdisease: (a) changes in demographics due to changes people suffer injuries requiring medical attention each year, in fertility and mortality rates, (b) changes in the prevalence and I in 3 have a day of restricted activity or required medical of infectious disease resulting from improved control and re- attention (Collins 1985). The definition of the severity of duced incidence, and (c) changes in risk factors resulting from injuries as either requiring hospitalization or outpatient medi- technological and social change. cal attention is useful for defining the burden to the health Demographic changes (such as the shift in age structure system in industrial countries. Such case definitions, however, and urbanization) have had an effect on the epidemiology will clearly record many fewer cases of similar severity in of injury primarily through an increase in the incidence of countries in which medical care is less available. injuries which are more prevalent among the elderly (such as falls) and in urban environments (such as motor vehicle MORBIDITY AND MORTALITY LEVELS, ABOUT 1985. Nearly 3 collisions). Reduction in the prevalence of infectious dis- million deaths are reported from injury and poisoning annu- eases during the past fifteen years has resulted in a growth ally; two-thirds of these occur in the developing countries in the relative importance of injury because it has typically (WHO 1989c). In many industrial countries, injuries are now emerged as the most important cause of death for ages one the leading cause of death during the first half of the human to forty-four. life span (Baker, O'Neill, and Karpe 1984), and they are In many rapidly industrializing countries of the developing becoming one of the leading causes of death and disability in world, the absolute injury mortality rates have also grown developing countries (Wintemute and others 1985). Because rapidly. With development have come technological and so- of the greater toll taken by injuries among the work force and cial changes which alter the risk of injury. These changes have younger age groups, however, their importance to the public the potential to affect the incidence of injury either adversely health is best recognized when measured as years of potential (such as through increased hazards or increased risk-taking life lost, a reflection of premature mortality. behaviors) or beneficially (such as through safer products and Although prospective, population-based studies of inju- behaviors). In most developing countries to date, however, ries in developing countries are rare, Gordon, Gulati, and these changes in environment and lifestyle have exacerbated Wyon (1962) demonstrated a low incidence of disabling rather than ameliorated the problem of injury. (that is, causing disruption of normal activity) injury of 111 For example, Selya (1980) describes this trend in Taiwan per 1,000 people per year in eleven very rural Indian villages (China), where from 1960 to 1977 unintentional injuries rose in 1959. Gordon, Gulati, and Wyon's definition of "injury" from the seventh to the third leading cause of death, and the as that resulting in short- or long-term disability will be a absolute injury mortality rate increased from 38.9 to 57.2 per less sensitive measure of injury morbidity, although this is 100,000population.InShanghaiCounty,China,injurieshave probably the best definition where medical care is not uni- already emerged as the leading cause of death for people versally available. In a community-based survey of children between the ages of one and forty-five (Gu and Chen 1982). in an urban slum of Rio de Janeiro, Reichanheim and Harp- The increased motorization of transportation in develop- ham (1989) documented that 30 percent of the children had ing countries is perhaps the best-documented example of been injured within the last fifteen days; 85 percent of them the unintended negative consequences of technological were treated at home. change. The explosion in the number of roads and vehicles Injury 615 Table 25-3. Estimated Global Mortality from Injury and Poisoning (by Region, Age, Sex, and Year) (per 100,000) 1970 1985 2000 2015 Population Male Female Male Fenale Male Female Male Female World Under 1 96 84 95 78 78 62 68 50 1-4 99 85 67 57 47 35 43 30 5-14 31 14 28 13 26 11 23 10 15-44 100 22 92 22 86 20 79 19 45-64 124 32 115 31 111 28 107 27 65 and older 216 153 201 140 202 144 205 151 Industrial countries Under 1 78 30 63 24 44 19 38 16 1-4 363 312 112 109 28 26 19 20 5-14 24 3 17 2 13 2 11 1 15-44 63 11 53 11 47 8 39 6 45-64 88 28 74 25 84 21 81 20 65 and older 224 188 191 151 209 167 208 168 Nonmarket countries Under 1 81 45 86 45 62 36 58 28 1-4 64 49 52 43 45 22 44 25 5-14 32 11 27 10 23 6 22 4 15-44 105 20 88 18 84 14 74 11 45-64 137 33 116 31 116 29 114 27 65 and older 230 155 217 156 208 157 213 155 Latin America and the Caribbean Under 1 68 52 101 67 68 47 61 39 1-4 67 44 87 61 45 32 44 23 5-14 32 13 32 14 23 10 22 8 15-44 119 24 105 25 86 21 80 18 45-64 157 35 147 36 114 30 110 29 65 and older 258 133 263 152 206 144 204 151 Sub-Saharan Afinca Under 1 141 126 118 103 99 84 83 66 1-4 74 60 71 56 55 41 50 35 5-14 50 22 42 19 35 15 28 13 15-44 133 26 120 25 108 24 97 23 45-64 149 33 138 32 129 32 123 31 65 and older 210 118 204 120 200 121 199 121 Middle East and North Africa Under I 110 105 103 93 88 74 76 59 1-4 54 47 60 52 48 37 43 31 5-14 36 19 30 14 27 13 23 10 15-44 113 30 103 24 96 23 85 21 45-64 135 34 133 32 124 32 114 30 65 and older 197 110 208 130 202 129 195 127 Asia Under 1 94 91 91 82 75 63 66 50 1-4 65 59 58 49 46 36 43 31 5-14 29 15 27 13 25 12 23 11 15-44 104 24 96 24 90 23 81 20 45-64 131 32 124 32 117 30 110 29 Source: Lopez. chapter 2, this collection.. 616 Sally K. Stansfield, Gordon S. Smith, and William P. McGreeveY in Saudi Arabia has been associated with an increase in both (Chesnais 1985). Risingthendeclininginjuryratesfrommotor morbidity and mortality (nearly 600 percent) due to motor vehicles, such as those seen in Sao Paulo, Brazil, during the vehicle crashes (Ergun 1987; Ofosu 1988). In Thailand, where quarter century beginning in the 1960s (Haight 1980), will the motor vehicle mortality rate has increased almost 30 likely be observed. Similar trends may be expected in other percent each year (Punyahotra 1982), injuries have risen from developing countries as the demand for safety increases. sixth to first among all causes of death since 1947 (Choovo- Growing prominence of alcohol and drug abuse often ravech 1980). Motor vehicles alone were responsible for more accompanies economic development. So few developing na- years of potential life lost than tuberculosis and malaria tions have begun to recognize or address this problem that it combined. will likely worsen before abuse control measures are instituted. The social changes which accompany development have The relative lack of mechanisms to implement legislation, also generally led to an increase in the frequency of injury in taxation, or treatment and rehabilitation for the control of developing countries. With the rapid introduction and diffu- alcohol and drug abuse will also probably hamper efforts to sion of new technologies, they are frequently used without control the associated injuries. concern for their safety. For example, in India the grain mills The increasing availability of firearms is likely to increase were mechanized without appropriate protective shields over intentional injuries throughout the world. Global expenditure drive belts, resulting in an increase in the incidence of severe for arms continues to rise, already exceeding $1 trillion dollars injury (Gupta, Bhasin, and Khanka 1982). The economic annually. Arms transfers to developing countries exceeded benefits of mechanization of industry or transportation are $52.7 billion from the United States alone in 1987. There will seen as greater than the cost of injury or death that may result be, however, tremendous resistance to any international ac- from the inappropriate use of these technologies. The large tion to halt this lucrative trade in an effort to contain inter- underclasses found in developing countries, who are virtually group and international violence. denied access to the wealth of the dominant group, have little The currently increasing use of pesticides in agriculture will to lose by high-risk behavior. Hopelessness in the face of be associated with an increased risk of poisonings unless the poverty, racism, social isolation, and drug and alcohol abuse higher exposure can be offset by stringent safety controls and does not encourage the investments necessary to improve replacement of the more hazardous agents with safer alterna- safety or health. tives. Successful implementation of other injury control mea- sures would be expected to interrupt the worsening trends in Possible Morbidity and Mortality Pattems: 2000 and 2015 injury incidence on the road, in the workplace, and in the home. In those countries which have begun to address the Still, development may be accompanied by technological and potential hazards associated with technological and social social changes that raise income and improve the equity of its development, such as in Europe and North America, the distribution. Studies have confirmed that safety is considered absolute rates of injury mortality have already begun to fall a normal good, the demand for which rises with income (Baker and others 1992). (Peltzman 1975). The development of a complex institutional structure (including legislation, enforcement, insurance and Economic Costs of Injury litigation services, and complex capital markets) helps to reduce the incidence of injury by forcing implementation of The annual medical and social costs of injury are estimated to safety measures. Such social organization also applies disincen- exceed $500 billion worldwide (WHO 1989a). Injuries are re- tives to the creation or maintenance ofhazards by forcing those sponsible for up to one-third of all hospital admissions (WHO who do so to compensate the victims of resulting injuries. 1989a). In addition to costs for emergency services, tremen- Individuals and industry may thereby be coerced to reduce dous costs are incurred in continuing care, rehabilitation, and injury risks to the larger community if there is public support lost productivity due to both death and disability. It is esti- for such social change. mated that the cost of injury treatment in the United States Mechanization associated with development may reduce in 1985 was approximately $317,000 for each fatality, $34,000 the incidence of injury if it reduces the interaction between for each hospitalization, and $500 for each injury not requiring people and machines or replaces more hazardous methods. In hospitalization (Rice and others 1989). It has been estimated Nigeria, for example, mechanization of a coal mine was asso- that the cost of injuries from motor vehicle collisions alone ciated with a 60 percent reduction in mining injuries as well amounts to nearly I percent of the gross national product of as a reduced severity of those injuries (Asogwa 1988). In Sao many developing countries. Thailand, however, estimates that Paulo, Brazil, in 1970, nearly 18 percent of industrial workers the cost of these injuries is more nearly 2 percent of the gross suffered a work-related accident; with increasing mechaniza- national product, not including the costs of the long-term tion the incidence was reduced to 3.8 percent by 1984 (Pupo disabilities (WHO 1987b). Nogueira 1987). Improved traffic safety and occupational in- Because of its high toll among the younger age groups, injury jury control measures in the United States resulted in reduc- is the main cause of years of potential life lost in industrial tions in injury mortality on the road and in the workplace after countries. In the United States, for example, injury accounts industrialization had initially brought about increases for 40.8 percent ofYPLL, at an estimated cost of$158 billion per Injury 617 year for both fatal and nonfatal injury (Rice and others 1989). Figure 25-2. Direct and Indirect Costs of Injuries The relative economic importance of injury in developing and Respiratory Diseases countries such as Egypt is even greater, where it accounts for 78 percent of YPLL and 10 to 30 percent of all hospital admis- 100 sions (wi-io 1988). Disability, both temporary and permanent, resulting from nonfatal injury is perhaps the most important, yet often 80 overlooked, cost of injury. There are few studies which quantify such disability in developing countries. In the United States, disability from injury results in a loss of 60 normal activity for an estimated 3 days per person per year (Smith and Kraus 1988). In a study in Sri Lanka, Krishna- . rajah (1972) showed that the disability which resulted from industrial injuries accounted for annual losses of 1.6 million . working days. The World Health Organization (WHO 1986) 20 estimates that 13 percent of the world's population is dis- abled and that at least 15 percent of these disabilities result from injury. These data suggest that of 78 million persons 0. (2 percent of the world's population) disabled because of I Respiratory injury, most live in developing countries, where disabilities njures diseases frequently become handicaps because of the lack of appro- priate rehabilitation services. m Although little such cost data are available from the Indirect cost developing world, in a study in northeast Brazil, DeCodes, L Morbidity Baker, and Schumann (1988) assessed the direct and indi- rect costs of various categories of illness. Although injuries accounted for only 11.8 percent of all direct costs of disease Source: DeCodes, Baker, and Schumann 1988. or injury, they accounted for 27.5 percent of indirect and 25.5 percent of total costs. Most (68.7 percent) of the total Elements of Preventive Strategy injury costs accrue from disability-related losses in produc- tivity, whereas approximately a quarter (25.3 percent) of Of the risk factors for injury incidence and severity included the costs are from loss of life. In figure 25-2 we compare in table 25-2, those most amenable to change include alcohol these findings by DeCodes, Baker, and Schumann to the use, unsafe behaviors due to poor safety education, and poor analogous figures for respiratory diseases. product design. For each of the interventions designed to The costs of lost productivity and medical care for injury reduce these risks, we summarize in table 25-4 calculations of are often exceeded by costs of property damage, insurance, expected disability-adjusted life-years (DALYS) gained by the and other nonmedical items. Researchers who conducted a prevention of injuries from motor vehicles, falls, bums, and survey of five developing countries (Turkey, Thailand, and toxic substances. The derivations of the estimates and calcu- three African countries) between 1961 and 1971 found that lations in table 25-4 are described in appendix 25A. damage to vehicles and property accounted for 60 to 87 Candidate preventive interventions should be assessed for percent of the costs resulting from vehicle crashes (Jacobs their expected cost and effectiveness prior to implementation. and Sayer 1983; Baudouy 1989). In no case did medical costs Proposed interventions may then be assigned a priority by a exceed 9 percent of the cost of crashes. Medical attention multidisciplinary body of community members and experts in and subsequent disability payments account for only one- injury control, which must include persons familiar with local fifteenth of an estimated $1.5 billion in annual traffic- sociocultural constraints. For example, Calonge (1987) has accident costs in Brazil (Airton Fischmann, personal identifiedsixfactorsforconsiderationinselectinganinterven- communication, 1988). tion strategy for injury control: (a) demonstrated efficacy in reducing injuries; (b) demonstrated effectiveness when im- Lowering the Incidence and Severity of Injury plemented; (c) public acceptance; (d) ease of implementation considering political, economic, and logistic barriers; (e) level As emphasized in figure 25- 1, the opportunities for prevention of personal commitment required; and (f) cost-effectiveness. of injury lie in alteration of the risk factors (to reduce the A sample semiquantitative framework for assigning priorities probability of injury) and of the injurious events (to reduce the to interventions for reduction of injury from motor vehicle severity of injury). The strategies designed to reduce the con- collisions is provided in table 25-5. sequences of injury are discussed later, in the section on case The strategies for implementing preventive interventions management. include: engineering; legislation, regulation, and litigation; 618 Sall-v K. Stansfield, Gordon S. Smith, and William P. McGreevey Table 25-4. Effectiveness of Interventions for Injury Control in Developing Countries Demographic and intervention parameters Trans portation injury Fall injury Burn/fire Poisoning Demographics Incidence (per 100,000) 665 2,000 600 100 Case-fatality ratio (percent) 1.7 0.2 1.0 6.0 Average age at onset 30 30 10 10 Morbidity (life-years lost) per injury 0.22 0.03 0.86 0.01 DALYS lost 3,098 1.278 13,438 180 Preventive interventions Alcohol taxation Reduction in incidence (percent) 30 26 34 8 DALDS gained 929 332 4,569 14 Product/environmenwal improvements Reduction in incidence (percent) 70 50 70 80 DALYS gained 2,169 639 9,407 144 Behavioral change Reduction in incidence (percent) 40 40 50 40 LDALYS gained 1,239 511 6,719 72 Case Management interventions Acute care improvement Reduction in case-fatality ratio (percent) 50 50 60 60 ReduLction in disability (percent) 50 50 86 60 DALYS gained 1,549 639 11,557 108 Rehabilitation care Reduction in disability (percent) 70 70 70 70 DALYs gained 2,169 909 9,407 126 Source: Authors' data. taxation and other economic incentives; and education. The LEGISLATION, REGULATION, LITIGATION. Because of the failure effectiveness of an injury control program will depend not only of corporations and industry to regulate themselves in matters on the effectiveness of the intervention (seat-belt use does of product safety, it becomes incumbent upon governments to reduce injuries), but also on the effectiveness of the mecha- provide requirements or incentives to protect their citizens. nisms used to promote or implement that intervention (health Powerful industry lobbies are often formed to resist such efforts. education in the absence of supportive legislation and enforce- For example, U.S. legislators concemed with safety have been ment, for example, has not been effective in changing pattems unable to ban the use of additives in cigarettes that enhance of seat-belt use; see Robertson and others 1974). their buming, although such a ban would help to prevent the house fires which claim more than 2,300 lives annually in the ENGINEERING. Perhaps the most effective injury control strat- United States (Smith and Barss 1986; Technical Study Group egies are those that alter the design of environmental features 1987). Some legal and regulatory strategies have been more (such as roads) or equipment to reduce or eliminate the risk of effective in altering injury frequency and severity. Legislation injury. Such passive interventions, which do not rely on or regulation in the mid-1970s to improve automobile safety changes in volitional behavior, are generally more likely to be (for example, through design standards for brakes, door locks, effective than those that require the active participation of the restraints, fuel systems) is credited with saving 9,000 lives individual to reduce injury risk. Air bags, for example, which annually (Robertson 1981). Most of these interventions were inflate automatically in a motor vehicle crash will provide actually developed by industry itself but were not implemented passive protection even of occupants who fail to wear seat initially because of short-term cost considerations. belts. In industrial countries, litigation or the fear of it has led to Manufacturers seeking to limit costs of improving product increasing corporate responsibility for providing safer work- safety frequently argue that customers should be given the places and products (Teret and Jacobs 1989). In the United freedom to choose less expensive products without added States, for example, some states have strict product liability safety features. This issue is raised in both industrial and laws that impose penalties if injuries occur that could have developing countries. The difficulty, however, of ensuring been prevented through use of state-of-the-art safety designs individual "informed consent" to the risks of using unsafe (Robertson 1983). Although litigation may be an important products and the costs to the society that are imposed by means of injury control in the United States, its applicability injuries must be considered by govemments in establishing in developing countries is likely to be limited for the near policy. future. Injury 619 Table 25-5. Sample Programs for Control of Injury from Motor Vehicle Collisions Expected Accep- Feasibility/ Low Phase _ _ Intervention impact tability Enforceability cost Priority Pre-event Adopt the 1975 UN guidelines for issue and validity of + + + + + + + + 3 driving permits, with periodic visual screening of drivers Initiate vehicle registration requirements, with periodic + + + + .++ + + 3 inspection for safety features Limit dangerous vehicles (such as motorcycles over 250 +++ +++ +.++ +.++ . c.c.) through taxation or import restrictions Require imported vehicles to have padded dashboards, ++++ . +++ +.++ ++++ I anti-lacerative windshields Event Establish and enforce speed limits . ++ . +++ + + 2 Identify and improve "black spots" or hazards; divide ++++ .+++ ++++ + + I highways Create pedestrian and bicycle-segregated traffic areas +.++ + + + + .+. 3 Modify roadways through towns to ensure slowing of .+.+ + + +++ + + 3 traffic Improve roadside lighting + + .+.+ + + 4 Mandate and enforce use of searbelts and child restraint ++++ +++ + + +++ 2 systems in passenger vehicles Mandate and enforce use of crash helmets and daytime +...+ +.-+- ++.+ 2 headlights for motorcycles Provide basic emergency care training for police, public .+.. ... .... .+. 2 transport drivers, and others likely to be first at scene Post-event Train primary health care workers in injury diagnosis and ++++ .+.. +.++ +++ I primary management, including use of local materials for collars, splints, and stretchers Coordinate local communications and transport .+++ +++ .+. 2 resources to provide emergency transport to trauma centers Regionalize and upgrade trauma care in urban centers .++ .+++ .++ + 2 Improve or develop community-based rehabilitation .... ++++ +++ +++ 2 services, including training and referral resources at regional trauma centers + Low. +-+ moderate, +++ high, + ++++ very high. Source: Authors' data. TAXATION AND SUBSIDY. Taxes and other economic incen- oping countries may be even more responsive to price changes tives have been used creatively to reduce injury frequency and (Wamer 1990), and excise taxes have already been successful severity. Reductions in insurance rates for vehicles equipped in reducing cigarette consumption in Papua New Guinea with airbags, for example, have been used to promote their (Chapman and Richardson 1990). selection by consumers and thereby reduce the severity of Because most of the alcohol consumed in developing coun- injury in motor vehicle crashes. Taxes on the use of private tries is produced indigenously, it is frequently argued that state vehicles combined with subsidies of safer modes of transporta- monopoly systems should be instrumental in preventing alco- tion (such as trains and buses) can help to shift transportation hol abuse. Although the feasibility of control over traditional, preferences and reduce injury because all forms of mass trans- noncommercial alcohol production and consumption must be portation experience much lower death rates than do private considered, the potential profit (as well as public health ben- vehicles for the same number of miles traveled (Baker and efits) of state taxation and control of alcohol should provide others 1992). incentives for adopting national preventive policies. Success- Taxation has been particularly effective in modulating be- ful state control cannot be implemented, however, without haviors such as alcohol use. Alcohol sales and consumption public support and stable political conditions. If national alco- have been shown to be elastic such that price increases through hol policy is not viewed as a reflection of the society's attitudes, taxation effectively reduce consumption (Cook 1981). A re- black market trade in alcohol quickly emerges, undermining cent study estimated that a tax of approximately 35 percent on national control and revenues. the retail price of beer in the United States would eliminate Successes have been achieved in the Gambia, for example, half the alcohol-related fatalities, and a 50 percent tax would where 90 percent of alcohol is consumed as palm wine (made eliminate approximately 75 percent of these deaths (Phelps from the sap of the trees); fees are collected for the license to 1988). There is evidence to suggest that consumers in devel- tap the palm trees and for distribution to the local markets 620 Sally K. Stansfeld, Gordon S. Smith, and William P. McGreevev (Kortteinen 1989).Theeconomicsignificanceoftradeinpalm An additional obstacle to the development of comprehens- wine in rural areas is immense and represents an opportunity ive preventive programs has been the inherent need for inter- for governments to provide disincentives for alcohol abuse disciplinary and multisectoral action for injury control. through taxation and to finance other costs of alcohol abuse Although the health sector might best take the lead, health prevention or treatment with the revenue generated. ministries rarely have had the inclination or power to coordi- The availability of highly dangerous products may also be nate the multiple disparate groups necessary to plan and im- shaped by economic incentives through the use of import plement the necessary environmental, policy, and behavioral duties. Although duties on the import of every potentially changes. Agencies responsible for public health; curative and dangerous product may raise the cost of doing business and rehabilitative health care; legislative affairs; public policy; retard economic growth, a more targeted strategy could limit criminal justice; sociologic, psychiatric, and anthropologic injury frequency and severity without hampering develop- investigation; occupational hazards (including agricultural ment. For example, commercial interest in the importation of pesticides); regulation of alcohol and drug use; education; and more dangerous technologies might be altered in favor of safer transportation safety must all be coordinated to ensure a com- products by imposing a tax on imports which is proportional prehensive preventive strategy. to the risk of injury. The structure of duties could be adjusted Despite these obstacles, many developing countries have to encourage import, for example, of less toxic pesticides or made significant progress in strengthening injury prevention. safer cookstoves. Control or even elimination of the importa- Eleven of thirty-two developing countries in a recent survey tion of many products (such as handguns or other firearms), (WHO 1989c) have established a coordinating body for injury could probably be more effective in developing countries than prevention or traffic safety. Many have also established strat- it is in industrial countries because local production is more egies for planning and financing research and injury control limited. activities. The World Health Organization's Injury Prevention Programme provides assistance to member countries in the EDUCATION. Education is frequently advocated to effect planning and implementation of national injury control pro- changes in environmental and behavioral factors which alter grams. risk of injury, yet there is little evidence to support the effec- For example, countries such as Nigeria have begun to rec- tiveness of such interventions, even in the industrial world. ognize the need for national policy to address the growing The most promising results have been achieved with educa- problem of alcohol-related injury. Odejide, Ohaeri, and tional interventions which are intended to prompt a single Ikuesan (1989, p. 235) have called for policy reform, including behavior (such as installation of smoke detectors) rather than "legislation on age limit for purchasing or drinking alcohol in sustained behavior change (such as seat-belt use or reduction public places; legislation on drunk driving behavior; control of in alcohol use; see Robertson and others 1974; Miller and alcohol advertising and period of sale; provision of breathalizer others 1982). Training workers in safe work habits has been equipment in hospitals; provisional law enforcement agents; shown to reduce the risk of injury, although the behavior and massive education on the issue." changes were not sustained for long periods after that training Although many preventive technologies have been devel- (Margolis and Kroes 1975). It may be pointed out that the oped in industrial countries, there is no reason to believe that generally lower level of knowledge about safety in many devel- most would not be applicable in developing countries. Many oping countries may leave more room for gains to be made proven preventive measures are not costly, yet they have not through education. Educational interventions have been most been adequately exploited to reduce injury frequency and effective, however, when used in support of legislative, taxa- severity in the developing world. tion, or engineering interventions, to increase their accept- For example, the 27 percent reduction in passenger car ability (National Committee for Injury Prevention and deaths observed in the United States from 1965 to 1985 Control 1989). (despite a near doubling of mileage exposure) would probably be reproducible in many developing countries (Ezenwa Good Practice and Actual Practice: Are There Gaps? 1986b). This success was attributed to improved roads (Na- tional Safety Council 1986) and progress in making passenger Injury has not been recognized as a preventable public health vehicles more crash-worthy (Campbell 1987). The decline in problem worthy of allocation of resources proportionate to its motor vehicle mortality observed in Sao Paulo, Brazil, since importance. It is demonstrated in figure 25-3, for example, that 1980 has also been ascribed to multiple preventive interven- research funding for cancer in the United States is more than tions, including better control of driving speeds, greater police twenty-one times that for injury when compared on the basis surveillance, improved traffic engineering, and placement of of preretirement years of life lost (NRC/ION4 1985). Moreover, more footbridges over principal thoroughfares (de Mello and the research information vacuum created by the historical lack Bernardes-Marques 1985). of resources has been used to explain continuing neglect of the Seat belts are infrequently used in most developing coun- injury problem. In recent years, however, in both industrial tries, although they might be expected to reduce the risk of and developingcountries, injury control has been given higher fatality by 43 percent (Evans 1986) and serious injury by 40 to priority than it had previously. 70 percent in motor vehicle crashes (SAE 1984) on the basis of Injury 2) 2 Figure 25-3. Preretirement Years of Life Lost Annually and Federal Research Expenditures for Major Causes of Death, United States US$998 mIllion research expendinure US$624 million research y expenditure 4 1 mmlloon years of life liost Injury Cancer Heart disease/stroke Source. NRC!IOM 1985. the experience in thiel United States. In SingZapore, ; seat-belt is freqtent iv assoieated Wtth bumrs and drowiwnings, suggestinsg law accomipanieul by edLIcaltion and Iaw enforcemient eftforts that better treatment couldl reduice the incidence of inJuries in have Aready suICCess-tu_Illx redcficcd miotor vehicle Injuiries persolns wvith pot rlv controlled seizLire disrtiders (Buchanall (Chat, Kbh .o, and I o)n 1984). 197'; Pearn, Rart, and Yaiai itka 1978; Subianto, Tumada, alnd Othter Icgisl:ttive inters entions whiich have beeii efftectsve Miargono 1978; Sonnen I9)i0); Barss and WallIce 1983). in Indu.strial ctounrries and Wi tf uld I eIt ifbenefit in develiplnmg There is an urgent need ti O apply avOlready existiIg prevenItive co)untries inclUide lass reLfuiring minottrcvclhsts to) Luse helmtsilt strategies, sLich as reducting a;vailility ofd angerousu drUgs, Use and ti h;tve thter headlights tin at allt times (Miler 1982; Ziditr of chtldprt of caps, and proper torage of toixic substances, ttn 1981). The repeal (f a m and'atort hteInmet law in Texa;s wa redutce poisonings in developing coiuntries (Oliver anId Hetzel follotwed byv a 73 percent increalse in tata hties and 20 percenit 1972; Baker, O'Neill, and Karpc 1984). Information fromn incre;se In injuries asoc iatedl with 11tt ircycle accident>. The developing countries suggests that regulat ion and edLucititin to hiiglhitnctdenceotcrash injuir\ and tIn1)rt ,itt iiit loaongiotorcyc I retiuce lnaniprt prpit-e use ut ildic;itions and otlher poisi inings riders in desIe-loping ctintrites alst sggets that such lawxvs l better packaging andl laeling, addliition if emetics tir concerning helmet nmd headlight i Se ma be hefict se n stei, heints, and restricttion oAt the av:ii aiility ot highly toxic in Indiai, wIhere Cultural traditions rest iri.t hIlmeit use, stltudies prepiraitklnti 1 wouldf ;tl1 be expectel tto reduce orbidity anld sh1OWed that turbans offer partial pr; ittion Frkiii headi iti)tiry niirtalits (MoyMibreN 1986; ('?h(ufdry and i thers 1987). (S1ioid 1988). In Ziilmabvw, for examiple, 34 percent ofifagricuIRlral work- Evidlencesuggests that redulcing the lammabailityoftClothinig ers ui rking with dangeiroiis pesticides repoirted they had re- (suLchI .11 1as b Lsinlg a oraix rin,se) and Its looseness may be ceived nut salfetN instri on01k TA, and i8 percent believed tlhat expected ti reinUC morbidit; and mortality broin hurn nj Lrtt pesticide iontainers ciuldl be reusled for Other purposes, such [I)urrani 197 4; Durrani and Raza 1975; McLooughlln and oth- as storing foo tio r drinking water (Bwtititi anid others 1987). A ers 1977). Improved dIesign and safer use of stoves andl lampns progranm to redtice pesticile poisinings In China throuighi w1ould als1 be expecXted tJ i redIuce bur injrury (nl, uchChintoss and reculat IXMn, education, and other preventive measures resulted Grave I 976; Lee 1 982; Winte(muLte and others 1 985). Epilepsy In a redci Ct Ion in incidence ot niore thaln 98 percent. despite a 622 Sally K. Stansfield, Gordon S. Smith, and William P. M-Greevey nearly thirteenfold increase in the use of pesticides during the come which generally guides case management decisions. De- same period (Shih and others 1985). cisions in the transport or care of the injured must also consider Studies from industrial countries show that legislation man- the mechanism or external cause of injury, however, because dating adequate fencing, together with self-closing, self-locking this clinical history may be the sole clue to serious injury which gates, around swimming and other pools of water can prevent is occult or late in manifestation. most toddler drowning deaths (Milliner, Pearn, and Guard 1980). Improved supervision and enclosure of small ponds, Elements of the Case Management Strategy wells, and drainage canals in residential areas in developing countries would also be expected to reduce drowning deaths Deaths due to trauma in the industrial world have been de- among children (Gordon, Gulati, and Wyon 1962; Nixon, scribed by Trunkey (1983) as occurring in three periods after Pearn, and Dugdale 1979; Thapa 1984). the injury. "Immediate" death, occurring within an hourofthe Other examples of proven technologies which are under- injury, is generally the result of massive injury. "Early" deaths used include safety guards to prevent hair or clothes from being are defined as those occurring one to three hours after the caught in drive belts (Gupta, Bhasin, and Khanka 1982). The injury, are generally from intemal or other bleeding, and are important problem of injury due to falls must be addressed regarded as being preventable through early, good quality, using locally appropriate strategies. For example, use of the medical and surgical care. The "late" deaths occur days to new, shorter hybrid varieties of tropical trees such as mango weeks after the injury and are generally due to infectious and oil or coconut palms would undoubtedly prevent many complications or multiple organ failure. Trunkey estimates serious injuries in areas where these products are prominent. that more than half of the trauma deaths in the United States Falls from rooftops, windows, or into wells can clearly be are immediate, and approximately 30 percent are early. In the prevented by appropriate design modifications or barriers. developing world, however, it appears that many more deaths Regulations for safe workplace conditions and regular in- occur in the early and late periods, suggesting that considerable spections can clearly reduce injuries. In Brazil, a 79 percent death and disability might be prevented through improvement reduction in the proportion of industrial workers suffering in the quality of care and transport. accidents between 1970 and 1984 has been ascribed to legis- Five major factors are linked to the outcome of care for lation regarding the protection of workers (Pupo Nogueira the injured: the severity of the injury, the age of the injured 1987). individual, the preexisting health status of the injured indi- For criminal assault, homicide, and suicide, limiting access vidual, the time elapsed from injury to definitive care, and to the means, such as firearms and medications, is likely to be the quality of care. Therefore, once an injury is sustained, the most effective means of prevention. The evidence from the the strategies available to reduce morbidity and mortality United States and Canada suggests that the reduction of the through improved case management are improvement in availability of weapons which inflict fatal injury would be the quality of care, improvenent in the emergency transport effective in reducing the incidence of homicide and suicide system, or both. (Sloan and others 1988). Elimination of carbon monoxide from the coal gas in Birmingham, England, resulted in a 50 IMPROVED QUALITY OF CARE. The overall excess case-fatality percent reduction in suicide deaths (Hassall and Trethowan ratio of injuries sustained in the developing world also suggests 1972). Selection of effective strategies to prevent homicide, room for intervention through improved case management suicide, and genocide (including war) must also clearly take (WHO 1988). There is evidence that modern treatment tech- cultural factors into account. niques can reduce mortality rates of burn victims (Demling The prevention of arms sales to countries at war is one 1985) and that early and effective rehabilitation of such vic- strategy that would be effective if international cooperation tims can reduce disability (Sinha 1984). Many needless deaths could be achieved. Yet the barriers to rational public health are caused by preventable complications of minor injuries, policy for the prevention of war are great. Werner (1989) poses such as bleeding and infection. Gordon, Gulati, and Wyon the key question: Has the huge arms industry been created in (1962) found, for example, that three of twenty-three (13 the service of national security, or has the national security percent )deathsfrom injury werefrom infectionsfollowing the paranoia been created in the service of the arms industry? injury. There is little information on the cost-effectiveness of Industry's pressures upon governments and governments' pres- improving the quality of care. Because these interventions sures upon intemational agencies may explain the otherwise would depend largely on training existing health professionals, incredible neglect of this important public health problem. however, the expense would likely be minimal. Even at the community level, appropriate first aid may Case Management reduce the consequences of injury. Although there are few prospective studies of the effectiveness of such basic care, not Although preventive strategies to control injury primarily seek many people would doubt the beneficial effect of controlling to alter the extemal causes of injury, case management inter- hemorrhage, cleansing wounds, or stabilizing fractures. One ventions act to limit or reverse the pathological outcome of study in India demonstrated the effectiveness of cooling bums injuries. It is the extent or character of the pathological out- with cold water (Mohan and Varghese 1990). Yet in some Injury 623 developing countries few community members are familiar so, a difference of 70 percent. These observations are supported with such first-aid principles. by the estimates of those concerned with rehabilitation in At the primary health care level, workers may provide the developing countries that one-half to three-quarters of those care above in addition to managing minor trauma, including rehabilitated can subsequently return to income-generating simple fractures and minor open wounds and burns. Appropri- activities (David Wnemer, personal communication, 1990). ate immediate stabilization of suspected fractures prior to transport of the injured person, especially in neck injuries, IMPROVED TRIAGE AND TRANSPORT SYSTEMS. A significant often prevents subsequent paralysis or other disability. Simple challenge of emergency care of the injured is the process of removal of toxins and prompt use of emetics at this level may "triage," which ensures that injury is managed with the significantly limit the extent of injury due to poisoning. Rou- appropriate priority and at the appropriate level in the tine rehabilitation care, such as that after bone fracture, am- health care system. In multiple injury and disaster settings, putation, or bums, is also best conducted at the primary health it is of paramount importance that case management activ- care level to optimize the function of injured persons within ities be prioritized and tasks allocated to optimize use of their own community. health care resources. Each community must take responsi- More sophisticated care is required at the secondary and bility for disaster preparedness by ensuring that local re- tertiary levels for appropriate surgical treatment, such as for sources (for communications, transport, and health care) thoracoabdominal trauma, and for inpatient care of injury or are assessed, priorities established, and tasks assigned in poisoning. Data from Trinidad and Tobago (Ali and Narayn- advance or as promptly as possible. singh 1987) suggest that once the injury victim arrives at the The model from industrial countries for the case manage- hospital, the twofold higher death-to-injury ratio observed ment of injuries includes an elaborate system of emergency there than in North American hospitals might be reduced by medical services (EMS), including communications, transport, improved in-hospital care. In another study, in Natal, Bullock and prehospital and hospital care. Although this strategy of and others (1988) estimated that one-third of deaths due to case management of the injured is of interest to many devel- head injury could have been prevented with adequate medical oping countries, the cost of such a system is often prohibitive. treatment at referral sites. Adala (1983) found that, in dev- Even in the United States, EMS systems are generally available eloping countries, treatment of eye injuries in tertiary care only in large urban areas(West, Williams, andTrunkey 1988). centers is often delayed, resulting in extensive infection and The benefits of improved communications and transport sys- loss of sight. tems, however, would be observed in improved case manage- Rehabilitation, although, ideally, community-based, also ment of other medical and surgical problems, most notably requires the backup and referral services of secondary and obstetrical emergencies. Among patients transported by EMS tertiary care centers for specialized problems (Smith and Barss systems in the United States, for example, only about one in 1991). To limit disability due to injury effectively, rehabilita- six has sustained an injury (Meador, Cook, and Larkin 1989). tion requires long-term, multidisciplinary, and comprehensive At least four essential elements should be included in EMS service delivery. The goal of rehabilitation is the optimal systems: (a) detection and assessment of emergencies; (b) return of function-physical, psychological, social, and voca- notification and coordination of transport and definitive care tional-to the injured individual. Rehabilitation is accom- services; (c) organization, training, and performance evalua- plished both through changes in the functional capacity of the nion of key participants in the EMS system; and (d) stabilizarion disabled person (for example, development of compensatory and provision of definitive emergency care. If these four ele- muscle strength or the use ofprosthetics) and through changes ments are developed, integrated, and strengthened appropri- in the physical and social environment. Simple rehabilitative ately, the EMS system wvill also be capable of coordinating the practices, such as range-of-motion exercises to reduce contrac- medical response to natural or other disasters. tures after burns, may have tremendous effect in improving Ali and Naraynsingh (1987), in their study ofTrinidad and function. Strategies which have been outlined for community- Tobago, document that excess numbers of injury victims who based rehabilitation of the disabled with the limited resources are alive after the injury event are dying before they reach the available in developing countries (Miles 1985; Werner 1987) hospital, suggesting that mortality could be reduced by basic can considerably reduce the economic and sociopsychological improvements to transport and prehospital care. Bhatnagar costs of injury-related disabilities. and Smith (1989) point out the increased fatalities associated In an evaluation of a rehabilitation program for patients with lack of transport among soldiers injured in the Afghan with spinal cord injuries in Taiwan, Wong, Chen, and Lien war. Compared with their counterparts in wars in which there (1981) suggested that significant reduction of disability could was prompt access to good medical care, few patients with be achieved with rehabilitation interventions. After rehabili- thoracoabdominal injuries survived long enough to make it to tation 75 percent of the patients were able to walk unassisted, the only hospitals available, which were on the Pakistan and 68 percent were able to resume activities of daily living. border. Of the rehabilitated group, 17.6 percent were able to resume Although studies before and after development of improved working at their previous occupations, whereas only 5.2 per- prehospital care and transport have shown large differences in cent of those who received routine treatment were able to do survival rates (West, Williams, and Trunkey 1988), such sys- 624 Sally K. Stansfield, Gordon S. Smith, and William P. McGreeve', tems should be developed only where good quality definitive Training of currently available personnel at secondary and care is reliably available. Where such secondary and tertiary tertiary care centers in basic trauma care also represents an care is adequately developed, simple emergency communi- inexpensive intervention to reduce morbidity and mortality cations and transport systems may effectively reduce injury which would probably be effective. For example, existing com- morbidity and mortality. In their study of one such system munications and transport resources (such as police, fire, or in Papua New Guinea, based on radio linkages and intermit- military systems) might be used to improve triage and prehospi- tent charter use of local aircraft, Barss and Blackford (1983) tal care, avoiding duplication and limiting costs. In industrial documented an annual cost per capita of $0.20 for all med- countries, however, creation of separate EMS systems and ical, surgical, and obstetric conditions, with a cost per life development of regional trauma centers may represent the saved of $450.00. most cost-effective next step to reduce injury mortality and A telephone-based EMS for prehospital care and transport disability. has been established in five cities and surrounding areas on the three most populated islands in Indonesia (Pusponegoro Priorities for Injury Control 1989). This system has been successful in providing prompt triage, treatment, and transport during several disasters as well Enough is already known to establish the importance of injury as more than 5,000 routine ambulance responses in Jakarta as a public health problem in developing countries. Proven alone. Similar systems in the Dominican Republic (Eaton and interventions exist to begin to address key injury problems. Perez Mera 1989), Egypt (Sarn 1989), and Nigeria (Owosina Although more work is needed to define locally specific injury 1989) have been planned or implemented with creative meth- problems further, the need for more information to refine ods of community participation, financing, and use of currently strategies continuously should be addressed in the design available resources. of injury control programs but should not delay prompt The estimated effectiveness of simple improvements in the action. acute care and rehabilitation of the injured for four common The First World Conference on Accident and Injury Pre- sources of injury is presented in table 25-4. Effectiveness esti- vention was held in "response to the urgent need for promoting mates are stated as DALYs gained as a result of appropriate acute accident and injury prevention and to mitigate their conse- care and rehabilitation. The derivations of the estimates and quences on the health of people" (WHO 1989a). Policymakers calculations in the table are described in appendix 25A. from fifty countries developed a "Manifesto for Safe Commu- nities," elaborating recommendations for action to (a) formulate Good Practice and Actual Practice: Are There Gaps? public policy for safety, (b) create supportive environments, (c) strengthen community action, and (d) broaden public services. Efforts to design strategies to reduce injury morbidity and They emphasized the need to encourage politicians and deci- mortality through improved case management are hampered sionmakers to recognize the importance of injury and to iden- by the lack of data on the effectiveness of such interventions tify injury prevention as a priority goal. in the developing world. In the absence of such information, Recognizing and addressing the importance of injury will many countries' first efforts are focused on emulating the require integrated efforts at the international, national, and elaborate EMS systems for emergency prehospital care and community levels. The tasks of highest priority at each of these transport which are such a dramatic element of case manage- levels are outlined below. ment strategies in industrial countries. Many moderately industrial cotntries might profit from improved triage, region- Priorities for Action by International and Donor Agencies alization of medical services, communications, and transport systems; however, there is no benefit to saving minutes or even The focus of the "Manifesto" on the need for injury control hours of transport time if referral centers are not adequately action at the national and community levels in developing staffed and equipped to provide definitive trauma care. countries is appropriate. The governments of some of the more Another important obstacle to the development of appro- economically powerful nations, however, currently defend the priate primary care for the injured has been the emphasis in profits of their national and multinational industries at the developing countries on vertical programs addressing the tra- expense of the health and safety of the people of developing ditional infectious disease causes of mortality. Few programs countries. Because injurious products, including alcohol, to- emphasizing such a selective strategy incorporate the basic bacco, pesticides, pharmaceuticals, and arms, are often pressed principles of management of minor injuries into the curricu- upon developing countries, little progress can be made in lum for training community health workers. Although the injury control until such marketing tactics have been additional cost of training workers in basic first-aid practice proscribed. would be minimal, this opportunity to reduce injury mortality Yet these industries, which together yield $619 billion per has generally been neglected. Incorporation of first aid into the year (Werner 1989), will not easily be encouraged to be more curriculum in primary schools in developing countries, for responsible in their exploitation of the markets of developing example, is an inexpensive intervention which would be effec- countries. They have already successfully blocked legislation tive in reducing the consequences of injury. designed to restrict or control the export or import of such Injury 625 hazardous products (Hill 1988; Werner 1989). The first prior- Priorities for National Action in Developing Countries ity for injury control, therefore, is immediate legislative reform in the industrial nations and prompt policy reform both in the Enough is presently known about the importance of injury in industrial countries and in international agencies. developing countries to justify action on a national level in Donor agencies concerned with bilateral development as- every country. The "Manifesto for Safe Communities" (WHO sistance should take responsibility for educating lawmakers 1989a, p. 8) states, "As part of its national health plan, each and politicians in their own countries to encourage such policy govemment should formulate a national policy and plan of reform. As a first step to restriction of "dumping" of hazardous action to create and sustain safe communities." products, donor agency regulations must ensure that products Formulation of a national policy and plan of action would imported with donor assistance should be made to the same or best be preceded by a review of the available national data similar safety standards as those in the donor country and regarding locally important injury problems (Smith and Barss country of origin. 1991). The first step would be to conduct an assessment of the International and donor agencies should assist governments injury situation, using existing data sources to characterize in recognizing that the minimal short-term cost savings in injury epidemiology, cost and effectiveness of interventions, buying such hazardous products is likely outweighed by the and needs for additional information. Frequently the data that social and medical costs of the resulting injuries. The cost are already available have not been analyzed or used to develop savings resulting from the prevention of injury are societal, strategies of injury control. Information available from police whereas the costs of specific interventions are often borne by or hospital records, for example, should be exploited before individuals or special interest groups. In order to protect their additional data needs are identified and new information sys- profit margins, industry can be expected to provide organized teins developed. resistance to injury prevention. International and donor agen- A national injury control program must establish a mecha- cies should support injury control efforts, recognizing that nism for intersectoral and multidisciplinary collaboration for long-term national economic growth will be hampered if safety policy planning and coordination of program implementation. is sacrificed to spare the short-term profits of these special A national injury control program cannot, in view of the need interest groups. for multisectoral action, be designed as a vertical program to Donors should complete an assessment of the injury effects be fully contained within a ministry of health. Central coordi- prior to funding nonhealth sectors to ensure that hazards, nation might best be achieved, therefore, through establish- such as those created by the construction of roads or the ment of a task force or coordinating committee which reports development of industry or trade, are recognized and mini- directly to senior-level government officials (WHO 1987a). mized. Funding should also be made contingent upon com- Such an administrative arrangement would also reflect the pliance with international standards for worker, roadway, necessary national political commitment to ensure the coop- and product safety. Countries may be encouraged to address eration of sectors responsible for education, transportation, the public health importance of injury responsibly by ear- industry and trade, housing, legislation, and enforcement. marking a percentage of the funding for each project to be The central coordinating body should then act to provide allocated to injury prevention activities. Donor agencies the necessary political and technical support for the injury should also allocate development assistance resources for control activities at intermediate (for example, regional or core funding to support the development of injury control district) and community levels. It is at these levels that specific programs and research. hazards will most often be recognized and dealt with to prevent International agencies, such as WHO, should continue to injury. The political and technical support will include provide information and guidance regarding policy and re- strengthening education, communication, research, and train- search priorities of intemational significance. Other United ing for injury control. In establishing priorities and providing Nations agencies, including the United Nations Environment support, governments must be mindful of their special respon- Programme and the United Nations Disaster Relief Office, sibility for addressing injury problems among politically and represent global resources which assist nations in achieving socially disadvantaged groups, such as children, women, and disaster preparedness or provide assistance to them during a minorities. disastrous event. Ongoing monitoring or periodic evaluation should be con- Intemational organizations concemed with health, human ducted to document the economic effect of injury problems rights, and security must take courageous action, sometimes and the cost and effectiveness of interventions. These infor- against the will of powerful member states, to halt inter- mation systems, once established, will be instrumental in re- govemmental and intergroup violence. Some responsibility fining strategies for implementation of any successful injury also rests with these agencies to coerce the governments of control program. Pilot projects may be established and evalu- industrial nations to halt the dumping of their unsafe prod- ated prior to the widespread implementation of unproven ucts in developing countries. Many major donor countries, interventions for injury control. however, which often control the international agencies, Governmentsofdevelopingcountries should supervise trad- will resist such pressure because of the huge financial profits ing partners and collaborate with donors to ascertain that their to be made in the sale of hazardous products such as arms. countries are not dumping grounds for hazardous products 626 SaUy K. Stansfield, Gordon S. Smirh, and WillUam P. McGGree vey which have been banned or become less marketable in their Table 25-6. Initial Injury Research Priorities country of origin (Navarro 1984). Import policies and duties Action Examples and national taxes or subsidies should be designed to increase product and environmental safety. Other national legislation Develop and test injury Identification and quantifi- should also be reviewed to ensure that it addresses the envi- surveillance and survey cation of disability; sensitivity techniques and specificity of case ronmental and behavioral hazards associated with locally im- tdefinitions portant injury problems. Elucidate local role of specific Alcohol and drugs Although no single list of interventions can be appropriate factors as risk for injury to every developing country, several injury control measures may be identified which address problems common to most Investigate cost, effectiveness, Preventive measures (espe- countries. The following low-cost interventions are likely to and impact of specific inter- cially those known to be have the greatest cost-effectiveness in most countries and ventions for injury control effective in industrial coun- tries); case management at should be considered as first priority actions: primary health care level; EMS Import policy and product and environmental im- systenms; regionalization of care (including trauma and poison provements to address burn, fall, and poisoning injuries control centers); rehabilitation * Legislation, regulation, and enforcenment to improve technologies and programs occupational and transportation safety Source: Authors' data. * Alcohol abuse control, especially through taxation * The strengthening of education for first aid and acute munities are best able to recognize local hazards and pressure care of the injured at the community and primary health local and national governments to improve the prevention and care levels treatment of injury. Effective communications to communities * Maintenanceoftrainedpersonnel,basictreatmentfacil- regarding injury problems, through both the press and the ities, and essential drugs for secondary-level management government, will help ensure community support for injury and prevention of complications due to injury control efforts. * Coordination of existing transport and communications Collaboration with the private sector may be an important resources to speed access to emergency care at secondary- contribution to the success of national injury control programs. and tertiary-level health centers Education and communications efforts might best be targeted for private sector decisionmakers, managers, and educators * The strengthening of rehabilitation services at the com-' munity level and referral resources at secondary and tertiary who are in positions to affect preventive practices. Industries levels may already have well-developed concerns regarding injury prevention and provide a ready source of support for program design and implementation. Many intemational nongovem- Priorities for Action at the Community Level mental organizations have stated an interest in injury preven- tion (WHO 1986), and their local chapters may also help to The impetus for injury control programs comes most appropri- organize critical support. ately and effectively from the community level. As develop- ment permits individuals and communities to take control of Priorities for Injury Control Research the economic and political forces which affect their lives, industry and goverments can be pressed to address the public The continuing need for evidence of the effectiveness of health problem of injury. Grassroots advocacy groups have specific interventions to aid in program design must be accomplished much where governments have failed in the addressed by incorporating strong evaluation or operational industrial world-for example, to change environmental fac- research components into every injury control program. tors (such as product design for safety) and behavioral risk Development of the capacity for conducting such opera- factors (such as alcohol use). tional research must be a high priority for resource alloca- Unfortunately, however, such spontaneous community or- tion. There are many examples of interventions which ganization to demand social change is less likely to occur in have had the opposite effect intended, emphasizing the the politically disadvantaged communities in many develop- need to assess the effectiveness of every intervention for ing countries. Communities are unlikely to demand increased injurv control. personal security when more basic needs are unmet. A larger TheWWorldHealthOrganization(1989c)hasidentifiednine share of the responsibility for ensuring adequate protection of areas of injury research: epidemiological and vital statistical, its citizens will therefore fall to governments until injuries are behavioral and psychological, mechanical and biomechanical, seen as unacceptable by an organized community. therapeutic, rehabilitative, environmental design standards, Meanwhile, however, communities should begin to work economic and legislative policy, toxicological and pharmaco- with their governments to increase public knowledge of injury logical, and health systems research. Several research topics problems and demand for safer communities. Organized com- which should be of the highest priority because of their impor- Injury 627 tance for program design are listed in table 25-6. Initiatives in ries in Taiwan (China) is detailed earlier in this chapter as an each of these research areas will be most cost-effective if illustration of data contained in table 25-4. undertaken as operational research, in conjunction with the design, implementation, and evaluation of national, regional, Transportation Injury or community programs. The incidence of motor vehicle injury in the United States is 2,266 per 100,000 people; of those injured, 9.7 percent are Appendix 25A. Sources of Data for Effectiveness hospitalized and 0.86 percent (19.4/100,000) are fatally in- Calculations jured (Rice and others 1989). The incidence of all injuries due to motor vehicle collision in developing countries is likely to Even for industrial countries, the effectiveness of specific in- be at least 665 per 100,000. This estimate is based on the jury control strategies has been poorly documented, except for average reported mortality rates for twenty-one developing certain motor vehicle injury prevention efforts (Rice and countries of 11.3 per 100,000 (PAHO 1986) and the injury- others 1989). The data from developing countries are even to-fatality ratio, or case-fatality ratio (cmr), of 1.7 percent, more limited, but estimates of the effectiveness of injury con- double that of industrial nations (Ali and Naraynsingh 1987). trol strategies may be made by using the available information The cFR for motor vehicle injury of 1.7 percent is consistent and generalizing from the industrial world if these figures are with the observed fatalities of 18 to 21 per 100,000 in Argen- poor or not available. Although the effectiveness of injury tina, El Salvador, CostaRica, andThailand (although ahigher control interventions would likely vary considerably from one rate [34 per 100,000] is found in Mexico, for example, and a country to another, figures have been selected which might lower rate [6 per 100,000] is found in the Philippines). best represent global averages. The lack of information on the The average age of thirty at the time of injury reflects the cost of implementing these injury control interventions in fact that persons injured in motor vehicle collisions in devel- developing countries currently precludes calculation of any oping countries are older than their counterparts in the indus- cost-effectiveness estimates. trial world (PAHO 1986; Ali and Naraynsingh 1987; Salgado The effectiveness estimations presented in table 25-4 are and Clombage 1988). For those not fatally injured, the mor- calculated for four of the most frequent causes of injury mor- bidity (in life-years lost) per injury is assumed to be the same tality and disability in developing countries (Manciaux and as the 0.22 calculated by Rice and others (1989) for these Romer 1986; Taket 1986). It must be realized, therefore, that injuries. because these injuries represent perhaps only 35 to 40 percent The likely effectiveness of alcohol taxation as a preventive of injury disability and mortality, the opportunity for effect on measure for motor vehicle injury is based on an AF of approx- health is much greater than suggested by these four model imately 40 percent, as is observed in both the United States intervention programs. Estimates are calculated as disability- (CDC 1990) and Papua New Guinea (Wyatt 1980; Sinha, adjusted life-years gained. Sengupta, and Purohit 1981). With the expected 75 percent Estimates of percentage reductions in incidence or disability effectiveness in reduction of alcohol-related fatalities through to be achieved with multiple interventions may exceed 100 taxation and other such economic incentives (Phelps 1988), percent, because there may be considerable overlap in the an estimated reduction of injury incidence of 30 percent is effect of specific interventions. For example, the same motor calculated. vehicle collision injury of an intoxicated driver might be Environmental and vehicle improvements might be ex- prevented (or reduced in severity) through prevention of alco- pected to result in at least a 70 percent reduction in incidence hol abuse, improved design of the vehicle, or by changing the or severity of injury due to motor vehicle collisions. Support driver's behavior to incorporate seat-belt use. for this estimate is provided by the 40 percent reduction in For each injury problem, the expected effectiveness of alco- mortality observed in the United States resulting only from hol taxation in the prevention of injury is calculated by mul- vehicular improvements (Robertson 1984). It has also been tiplying the expected reduction in alcohol-related injury times estimated (Smith and Falk 1987; Rice and others 1989) that the proportion of that injury problem which has been attrib- 75 percent of motor vehicle fatalities could be prevented uted to alcohol (cDc 1990), or the "alcohol attributable frac- through the use of currently available vehicular and environ- tion" (AAF). On the basis of the work of Phelps (1988) and mental safety standards (including vehicle modifications, pro- recent suggestions that price elasticity may be even higher in vision of airbags, and reduction of roadside hazards). Behavior developing countries than that observed in the United States change through education and appropriate enforcement of (Chapman and Richardson 1990; Warner 1990), a 75 percent seat-belt use might reduce injury incidence and severity of expected effectiveness in reducing alcohol-related injury these injuries by 40 percent (Evans 1986; Rice and others through taxation has been assumed. 1989). Because data from developing countries regarding the effec- Data from Trinidad and Tobago (Ali and Naraynsingh tiveness of rehabilitation in reducing disability are so limited, 1987) and similar observations in the United States indicate the same estimate is used for all four of the injuries. An analysis that improvement of trauma care for victims of motor vehi- of a rehabilitation program for patients with spinal cord inju- cle injury would result in an estimated 50 percent reduction 628 Sally K. Stansfield, Gordon S. Smith, and William P. McGreeveey in the CFR. A similar reduction in the disability of those Arabia (3.7 percent in United States), because the observed not fatally injured is assumed to be achievable with these hospitalizationrate is 16per 100,000 (jamalandothers 1989). interventions. The incidence of burn injury in the United States, for compar- ison, is 617 per 100,000 (Rice and others 1989). Falls The average age of ten at the time of injury reflects the higher incidence among children. For example, Sowemimo The incidence of injury due to falls in the United States is (1983) and Haberal and others (1987) reported that well over 5,184 per 100,000 people; of those injured, 0.1 percent are half of burn victims admitted in Lagos (56.2 percent) and fatally injured and 6.4 percent require hospitalization (Rice Turkey (69.7 percent) were less than fifteen years of age. In and others 1989). The incidence of injury from falls in devel- each case, most of these were less than six years of age. Similar oping countries is probably at least 2,000 per 100,000, in view age distributions for bum injuries have been observed in India of the observed death rate of 4 per 100,000 and an assumed (Gupta and Srivastava 1988). injury-to-fatality ratio of 0.2 percent (double that observed in Data from India (Gupta and Srivastava 1988) and Saudi the industrial world, as is the case for other traumatic injuries; Arabia (Jamal and others 1989) indicate that the cFR for see Ali and Naraynsingh 1987). The average age of thirty at moderate to severe bum injuries is about 1 percent, although the time of injury reflects the higher incidence among the mortality rates for hospitalized patients are generally much elderly and the occupational nature of many of these injuries. higher. For comparison, the case-fatality ratio among burn For those not fatally injured, the morbidity (in life-years lost) injuries in the United States is 0.4 percent. Forthose notfatally per injury is assumed to be the same as the 0.03 calculated by burned, the morbidity (in life-years lost) per injury is assumed Rice and others (1989) for these injuries. to be the same as the 0.86 calculated by Rice and others (1989) The probable effectiveness of alcohol taxation as a preven- for these injuries. tive measure for injury due to falls is based on an AAF of 35 The likely effectiveness of alcohol taxation as a preventive percent , as is observed in the United States (cDc 1990), and measure for injury due to fires and burns is based on an AAF of an expected 75 percent effectiveness in reduction of alcohol- 45 percent, as is observed in the United States (cDc 1990), and related fatalitiesfrom falls (Phelps 1988).These figures suggest an expected 75 percent effectiveness in reduction in alcohol- that a 26 percent reduction in injury incidence and severity related fatalities (Phelps 1988). These figures suggest that a 34 may be achieved with high alcohol taxes. percent reduction in the incidence and severity of these inju- Environmental improvements might be expected to re- ries might be achieved at a tax rate of 50 percent. sult in approximately a 50 percent reduction in incidence Product improvements such as safer stoves and less flamma- or severity of injury due to falls. Asogwa (1988) observed, ble clothing might be expected to result in a substantial reduc- for example, a 60 percent reduction in mining injuries tion in incidence and severity of bum injury. Sixty-three (primarily falls) when environmental improvements were percent of burn deaths in Minuflya, Egypt, are of women and made in the workplace in Nigeria. Education designed to are ascribed to overturned portable stoves (Saleh and others improve safety behavior has been observed to reduce the 1986); this epidemiology of burns is typical of that in many incidence or severity of fall injuries among children by 40 developing countries. In addition, more than 30 percent of percent (Kravitz 1973). That these estimates are realistic (or burns are related to clothing ignition (Durrani 1974; Barss and even low) is suggested by the 92 percent reduction in injury Wallace 1983). In the United States the introduction of im- from falls among children which was reported in New York proved flammability standards for children's sleepwear reduced City following initiation of a program including both envi- these burn deaths among children by more than 98 percent ronmental regulation and education (Bergner, Mayer, and between 1968 and 1980 (Baker, O'Neill, and Karpe 1984). It Harris 1971; Bergner 1982). is therefore estimated that improved stoves and less flammable Data on excess deaths among trauma victims in Trinidad fabrics in developing countries would achieve approximately and Tobago (Ali and Naraynsingh 1987) indicates that im- a 70 percent reduction in injury incidence. provement of trauma care for victims of injuries from falls Educational interventions might also be used to reduce the would likely result in a 50 percent reduction in the cFR. frequency of overtumed stoves. It has been estimated that 90 Although it is likely to be conservative, a similar estimate can percent of burns to children in Central Africa might be pre- be made of the reduction in the disability of those not fatally vented with simple barriers around open fires and cookstoves injured. (Auchincloss and Grave 1976). Education designed to im- prove safety behavior is rarely fully effective in altering the Fires and Burns target behaviors, however, so the expected reduction in the incidence or severity of these injuries is estimated to be 50 The incidence of moderate to severe bum injury is probably at percent (Barss and Wallace 1983; Schelp 1987). least 600 per 100,000 people in developing countries. In sup- Improvement of trauma care for bum victims would result port of this estimate, a similar incidence is implied if 2 to 3 in an estimated 60 percent reduction in the cFR, if the cFR is percent of such burn victims require hospitalization in Saudi reduced to that observed in industrial nations. The observed Injury 629 reduction in the proportion of burn victims requiring skin to result in an 80 percent reduction in the incidence and grafts when cold water was immediately applied to a burn severityofpoisoning.A95percentreductioninpoisoningswas suggests that this intervention could reduce the disability of observed, for example, with the elimination of carbon monox- those not fatally injured by an estimated 86 percent (Mathews ide from coal gas in Birmingham, England. Because organo- and Radakrishnan 1987). Improved rehabilitation of burn phosphates account for nearly half of poisonings, and 80 to 90 victims would likely contribute an estimated 70 percent reduc- percent of pesticide poisonings are caused by highly toxic tion in disability, because simple range-of-motion exercises preparations which account for only 4 to 5 percent of pesticide greatly reduce the formation of disabling contractures after use (Xue 1987), one could expect nearly a 50 percent reduction burn injury. in overall poisoning incidence with use of less toxic organo- phosphate preparations. Educational interventions and safer Toxic Injury use of toxic substances might also be expected to contribute a 40 percent reduction in the incidence of poisoning (Shih and The incidence of organophosphate poisoning alone is esti- others 1985). mated to be 100 per 100,000 people (Xue 1987) in China's Improvementofemergencycareforpoisoningvictimsmight largely agricultural society. In Sri Lanka, however, the inci- be expected to result in an estimated 60 percent reduction in dence is probably much greater, because pesticide poisonings the cFR, if fatality rates are reduced to near those observed in severeenoughtohospitalizethevictimoccurin90per 100,000 industrial countries. Although the disability rate is small, people. Organophosphates account for 20 to 50 percent of improved rehabilitation for poisoning victims might result in poisoning injury, so it can be estimated that the overall inci- an estimated 70 percent reduction in the severity or duration dence of poisoning in developing countries is approximately of disability. 300 per 100,000 people. The incidence of injury from poison- ing in the United States, for comparison, is 718 per 100,000 (Rice and others 1989). The higher incidence in the United Notes States might be expected in view of the higher prevalence of toxic substance use and the higher likelihood that mild toxic- The authors gratefully acknowledge the support and comments of Susan ity would be detected. Baker, Peter Barss, Lawrence Berger, Carlos F. C. Dora, Philip Graitcer, Dean The average age of ten years at the time of toxic injury lamison, Claude Romer, M_ C. Thuriaux, and David Wemer. reflects the higher incidence among children (Joubert and Mathibe 1989), although many poisonings occur in the work- place. On the basis of the ninefold excess fatality rate among References cases of organophosphate poisoning in China (Xue 1987), we estimate the c FR for poisonings to be approximately nine times Adala, H. S. 1983. "Ocular Injuries in Africa " Social Science and Medicine the 0.7 percent case-fatality ratio in the United States (Rice 17:1729-53. and others 1989), or 6 percent. This percentage is consistent Ali, lameel, and Vijay Naraynsingh. 1987. "Porential Impact of Advanced with average cFRs observed for otlher toxic ingestions (Shih and Trauma Life Supporr (ATLS) Program in a Third World Country." Interna- others 1985; Joubert and Mathibe 1989; Bhutta and Tahir tional Surgery 72(3):179-84. 1990) Forthos notfataly ijured themorbdity(in ife-ears Armstrong, K.. R. Sfeir, J. Rice, and M. Kerstein. 1988. "Popliteal Vascular 1990) . For those not fata lly inj ured, the morb idity ( in life -years lrrrInjuries and War: Are Beirut and New Orleans Smilar '?" Journal of Trauma lost) per injury is assumed to be the same as the 0.01 calculated 28(6):836-39. by Rice and others (1989) for these injuries. Although most Asogwa, S. E. 1980. "A Review of Coal-Mining Accidents in Nigeria over a poisoning victims who survive will recover without disability, 10-Year Period."Journal of Social and Occupationial Medicine 30:(2)69-73. the high disability ratio observed, for example, in Bhopal - 1988. "The Healrh Benefits of Mechanization at the Nigerian Coal (where 3 to 5 persons were permanently disabled for each of Corporation." Accidenti Analysis and Prevention 20:103-8. the estimated 10,000 fatalities) suggests that this is probably a Attah Johnson, F. Y. 1989. "Preventionand ManagementofProblemsRelared conservative estimate. to Alcohol Abuse in Papua New Guinea throuigh Primary Health Care." Alcohol abuse is associated with poisonings from other Medicine and Law 8:175-89. toxins, especially in suicides. The likely effectiveness of alco- Auchincloss, J. MN., and G. F. Grave. 1976. "The Problem of Bums in Central hol taxation as a preventive measure for poisoning is based on Africa." Tropical Doctor 6:114-17. an AAF of 10 percent, as is observed in toxic ingestions in Sri Baker, S. P., B.O'Neill, M.1. Ginshury,andG. Lt. 1992. The lnjurx FactBook. Lanka (Hettiarachchi and Kodituwakku 1989), and an ex- 2d ed. New York: Oxford Universit)' Press. pected 75 percent effectiveness in reduction of alcohol-related Baker. S. P., B. O'Neill, and R. S. Karpe. 1984. The lnjiuryFacrBook. ILexington, fatalities (Phelps 1988). These figures suggest that alcohol Mass.: Lexington Books. taxation (at a rate of 50 percent ) might be expected to result Banerjee, P., and S. Bhattachariya. 1978. "Changing Pattern of Poisoning in Children in a Developing Country." Tropical Pedaamcs and Encironmental in an 8 percent reduction in toxic ingestions. Child Health 24:1 36-39. Product improvements (that is, use of less toxic preparations Bang, R. L., and J. K. Saif. 1989. "Mortality from Bums in Kuwait." Burns of pesticides, child-proof caps, and so on) might be expected 15:315-21. 630 Sally K. Stansfield, Gordon S. Smith, and WiUliam P. McGreeve-y Barancik, J. I., B. F. Chatterjee, Y. C. Greene, E. M. Michenzi, and D. Fife. InstitutNationald'EtudesD8mographiques(iNED) and Intemational Union 1983. "Northeastem Ohio Trauma Study: 1. Magnitude of the Problem." for the Scientific Study of Population (iussp), Ordina Editions. Amenican Journal of Public Health 73:746- 51. Choovoravech, P. 1980. "Motor Vehicle Accident in Childhood." Journal of Barss, P. G., and C. Blackford. 1983. "Medical Emergency Flights in Remote the Medical Association of Thailand 63:304-9. Areas: Experience in Milne Bay Province, Papua New Guinea." Papua New Choudhry, V. P., A. J. Jalali, G. Haider, and M. A. Qureshi. 1987. "Spectrum Guinea MedicaLJoumal 26:198-202. of Accidental Poisonings among Children in Afghanistan." Annals of Barss, P. G., P. Dakulala, M. Doolan. 1984. "Falls from Trees and Tree Tropical Pediatrics 7:278-81. Associated Injuries in Rural Melanesianis." British Medical Journal (Clincal Collins, J. G. 1985. Persons Injured and Disability Days Due To Injuries, United Research) 289:1717-20. States, 1980-81. Vital and Health Statistics Series 10, 149. DHHS Publica- Barss, P. G., and K. Wallace. 1983. "Grass Skirt Burns in Papua New Guinea." tion (PHS) 85-1577. Public Health Service, National Center for Health Lancet 1:733-34. Statistics, Washington, D.C. Baudouy, J. 1989. "Road Accidents: An Emerginig Epidemic in Developing Cook, P. 1981. "The Effect of LiquorTaxes on Drinking, Cirrhosis, and Auto Countries." Harvard School of Public Health. Boston, Mass. Accidents." In M. H. Moore and D. R. Gersteion, eds., Alcohol and Public Bayoumi, A. 1981. "The Epidemiology of Fatal Motor Vehicle Accidents in Policy: BeyondtheShadowofPro/uhitnon. Washington, D.C.: NationalAcad- Kuwait." Accident Analysis and Prevennon 13:339-48. emy Press. Berger, Lawrence R. 1988. "Suicides and Pesticides in Sri Lanka." American CRED (Center for Research on the Epidemiology of Disasters). 1991. Disasters Journal of Publc Health 78:826-28. in the World: Statistcal Update from CRED Disasters Events Database. Univer- Bergner, L. 1982. "Environmental Factors in Injury Control: Preventing Falls sity of Louvain School of Public Health, Brussells. from Heights." In A. Bergman. ed., Preventing Childhood Injuries. Report of Cuellar, A. 1980. "Occupational Health and Safety in the Smelting and the Twelfth Ross Roundtale on Cntcal Approaches to Common Pediatrc Foundry Industries in Mexico." American Journal of Industnal Medicine 1: Problems. Ross Laboratories, Columbus, Oh. 261-63. Bergner, L., S. Mayer, and D. Harris. 197 1. "Falls from Heights: A Childhood Curry, Robert L. 1989. "Beverage Alcohol Spending in Singapore: A Potential Epidemic in an Urban Area." AnsericanJouial of l'ublic Health 61:90-96. Development Constraint?" lnterrational Journal of the Addictions 24(8): Bertazzi, P. A. 1989. "Industrial Disasters and Epidemiology: A Review of 821-28. Recent Experiences."Scandn7iavianJouTnalofWorkand EnvironmentalHealth Datev, S., N. S. Murthy, and A. D. Taskar. 1981. "A Study of Burn Injury 15:85-100. Cases from Three Hospitals." Indian Journal of Public Health 15(3): 117-24. Bhatinager, M. K., and G. S. Smiith. 1989. "Trauma in the Afghan Guerilla Davis, S., and L. S. Smith. 1982. "Alcohol and Drowning in Cape Town: A War: Effect of Lack of Access to Care." Surgery 105:699-705. Preliminary Report." South African MedicalJournal 62:931-33. Bhurra, Tariq Iqbal, and Khalid lqbal Tahir. 1990. "Loperamide Poisoning in DeCodes, J., T. D. Baker, and D. Schumann. 1988. "The Hidden Costs of Children." Lancet 335:363. Illness in Developing Countries." Research in Human Capital Development 5: Bittah. 0., J. A. Owola, and P. Oduor. 1979. "A Study of Alcoholism in a 127-45. Rural Setting in Kenya." East African MedicalJournal 56:665-70. de Mello, M. H. Jorge, and M. Bernardes-Marques. 1985. "Violent Childhood Buchanan, R. C. 1972. "The Causes and Prevention of Bums in Malawi." Deaths in Brazil." Bulletn of the Pan-American Health Organizanon 19(3): Central African.Journal of Medicine 18:55-56. 288-99. Bullock, M. R., M. D. du Trevou, J. R. van Delleis. 1. P. NeJ, and C. P. Demling, R. H. 1985. "Bums (Medical Progress)." New England Journal of McKeoiwn. 1988. "Prevention of Deathi from Head Injury in Natal." South Medicine 313:1389-98. Afncan MedicalJournal 73:523-27. De Wind, Christina M. 1987. "War Injuries Treated under Primitive Circum- Bwititi, T., 0. Chikuni, R. Loewenson, W. Murambiwa, C. Nhachi, and N. stances: Experiences in an Ugandan Mission Hospital." Annals of the Royal Nyazema. 1987. "Health Hazards in Organophosphate Use among Farm College ofSurgeons of England69:193-95. Workers in the Large-Scale Farming Sector." Central African Jounail of Dietz, P. E., and S. P. Baker. 1974. "Drowning: Epidemiology and Prevention." Medicine 33:120-26. Amencan Joumnal of Public Health 64:303-12. Calonge, N. 1987. "Objectives for Injuiry Control I ntervention-The Depart- Durrani, K. Ni. 1974. The Epidemiology ofBum Injuries. Bums Research Project, insent of Health and Human Services Model." Public Health Reports 102: Civil Hospital. Dow Medical College, Karachi, Pakistan. 602-5. Durrani, K. M., and S. K. Raza. 1975. "Studies on Flammability of Clothing Campbell. B. J. 1987. "Research Trends In Injun' Prevention." Public Health of Bum Victims, Changes Therein, and Their Wearability after a Borax Reports 102:592-93. Rinse."Journal of the Pakistan Medical Associanon 25(5):99-102. cLux (Centers for Disease Control). 1984. "Alcohol and Violent Death: Erie Eaton, D., and A. Perez Mera. 1989. "The Rise and Fall of EMS Innovation: A County, New York, 1973-1983." Morbidity and Mortality Weekly Report Cautionary Tale from ihe Dominican Repubic." Paper presented at 3 3:2 26-27. Cuinr aefo h oiia eulc"PPTpeetda the Intemational Conference on Emergency Health Care Development, . 1990. "Alcohol-Related Mortality anid Years of Potential Life Lost- Crystal City, Va. United States, 1987." Morbidity and Mortality Weekly Report 39:173-79. Edwards, G. 1979. "Drinking Problems: Putting the Third World on the Map." Chao, T. C.. J. H. Khoo, and W. N. Poon. 1984. "Road Traffic Accident Lancet 2:402-4. Casualities in Singapore (with Special Reference to Drivers and Front Seat Passengers)." Annals ofthe Academcy ofMedicine ofSitigapore 13(1):96-101. Ergun, G. 1987. "Condition of Vehicles in Saudi Arabia." Accident Analysis Chapman, S., and J. Richardson. 1990. "Tobacco Excise and Declining andPreventon 19:343-58 Tobacco Consumption:TheCaseofPapuaNewGuinea."AinericanJournal Evans, L. 1986. "The Effectiveness of Safety Belts in Preventing Fatalities." of Pubizc Health 1990; 80:537-40. AccideyntAnalysisandPretention 18:229-41. Chelala. Cesar A. 1990. "Central America: The Cost of War." Lancet 335: Ezenwa, A. D. 1986a. "Prevention and Control of Road Traffic Accidents in 153-54. Nigeria." Joumal of the Royal Society of Health 106(1) :25-26. Chesnais, J. C. 1985. "The Prevention of Deaths from Violence." In J. Vallin . 1986b "Trends and Characteristics of Road Traffic Accidents in and A. D. Lopez, eds., Health Policy, Social Policy, and Mortality Prospects. Nigeria." Journal of the Royal Society of Health 106(1 ):27-29. Injury 631 Fox, D. K., B. L. Hopkins, and W. K. Anger. 1987. "The Long Term Effects of Lee, K. N., Y. 0. Choi, C. H. Kim, and D. R. Yun. 1971. "An Epidemiological a Token Economy on Safety Performance in Open-Pit Mining." Journal of Study on the Incidence of Carbon Monoxide Poisoning in Korea." Jounial Applied Behavior Analysis 20(3):215-24. of the Korea Prevenave Medicine Society 4:95-106. Gaind, B. N., M. Mohan, and S. Ghosh. 1977. "Changing Pattern of Poisoning Lee, S. T. 1982. "Two Decades of Specialized Bums Care in Singapore." Annals in Children." Indian Pediatrics I 4(4):295-301. of the Academy of Medicine of Singapore II (3):358-65. Gordon,J.E.,P.V.Gulati,andJ. Wyon. 1962."Traumatic Accidents in Rural Loevinoshn, M. E. 1987. "Insecticide Use and Increased Mortality in Rural Tropical Regions: An Epidemiological Field Study in Punjab, India." Amer- Central Luzon, Philipinnes." Lancet 1:359-62. icanJournal of Medical Science 243(3): 158-78. Losada Lora, R., and E. Velez Busrillo. 1988. "Muerras violentas en Colombia, Gu, X. Y., and M. L. Chen. 1982. "Vital Statistics (of Shanghai County)." 1979-1986." Instituto SER de Investigaci6n, Bogota, Colombia. AmencanJournal of Publc Health 72(supplement): 1 9-23. Lourie, J., and S. Sinha. 1983. "Port Moresby Road Traffic Accident Survey." Gupta, R. C., S. K. Bhasin, and B. S. Khanka. 1982. "Drive-Belt or Patta Papua New Guinea Medicaljournal 26:186-89. Injuries." Jnjuryv 13(6):495-99. McLoughlin, E., N. Clarke, K. Stahl, and J. D. Crawford. 1977. "One Pediatric Gupta,R. K., and A.K.Srivastava. 1988."StudyofFatal BurnCasesinKanpur Bum Unit's Experience with Sleepwear-Related Injuries." Pediatrics 60: (India)." Forensic Science Internanonal 37(2):81-89. 405-9. Haberal, M., Z. Oner, U. Bayraktar, and N. Bilgin. 1987. "Epidemiology of Manciaux. M. 1984. "Accidental Injuries in the Young: From Epidemiology Adults' and Children's Bums in a Turkish Bum Center." Burns Including to Prevention." Effective Health Care 2(l):21-28. Therrmal Injuries 13(2):136-40. Manciaux, M., and C. J. Romer. 1986. "Accidents in Children, Adolescents, Haddon, W., Jr. 1970. "On the Escape of Tigers: An Ecologic Note." Amencan and Young Adults: A Major Public Health Problem." World Health Statistical Journal of Public Health 60:2229-34. Quarterly 39:227-31. - 1980. "Options for Prevention of Motor Vehicle Injury-" IsraelJournal Margolis. B., and W. Kroes. 1975. The Human Side of Accident Preventon. of Medical Science 16:45-65. Springfield, IL.: Charles C. Thomas. Haight, F. A. 1980. "Traffic Safety in Developing Countries."Journalof Safety Mathews, R. N., and T. Radakrishnan. 1987. "First Aid for Burns." Lancet Research 12:50-58. 1:1371. Hassall, C., and W. H. Trethowan. 1972. "Suicide in Birmingham." Bntish Meade, M. S. 1980. "Potential Years of Life Lost in Countries of Southeast MedicalJournal 1:717-18. Asia." Social Science and Medicine 14D:277-81. Hayes, W.J. 1980. "Factors Limiting Injury from Pesticides."Journul of Environ- Meador, S., R. T. Cook, and G. L. Larkin. 1989. "Advanced Life Support mental Science and Heakh BI 5(6):1005-21. Medical Care Distribution in a Rural/Urban Population." Paper presented Hettiarachchi, J ., and G. C. S. Kodituwakku. 1989. "Self Poisoning in Sri at the lntemational Conference on Emergency Health Care Development, Lanka: Motivational Aspects." InternationalJournal of Social Psychiatry 35(2): Crystal City, Va. 204-8. Mierley. M. C., and S. P. Baker. 1983. "Fatal Housefires in an Urban Popula- Hill, R. 1988. "Problems and Policy for Pesticide Exports to Less Developed tion." JAMA 249:1466-68. Countries." NaturalResourcesJournal28(4):699-720. Miles, M. 1984. Where There is No Rehab Plan. Mental Health Centre, Jacobs, G. B., and 1. Sayer. 1983. "Road Accidents in Developing Countries." Peshawar, India. Accident Analysis and Prevennon 15:337-53. Miller, R. E., K. S. Reisinger, M. M. Blatter, F. Wucher. 1982. "Pediatric Counseling and Subsequent Use of Smoke Detectors." Amenican Journal of Jamal, Y. S., M. S. M. Ardawi, A. A. Ashy, H. Merdad, and S. A. Shaik. 1989. Public Health 72:392-93. "Bum Injuries in the Jeddah Area of Saudi Arabia: A Study of 319 Cases." Milliner, N, J. Peam, and R. Guard. 1980. "Will Fenced Pools Save Lives? A 10-Year Study from Mulgrave Shire, Queensland." Medical Journal of Aus- Jeyaratani, J., R. S. de Alwis Senevirante, and]. F. Copplestone. 1982. "Survey tralia 2:510-11. of Pesticide Poisonings in Sri Lanka." Bulletin of the World HealthOrganiza- Mohan, D. 1982. "Accidental Death and Disability in India-A Case of tion 0(4):615-19. Criminal Neglect." Indusmal Safety Chronicle 24-43. Joubert, P. H., and L. Mathibe. 1989. "Acute Poisoning in Developing Coun- Mohan, D., and P. S. Bawa. 1985. "An Analysis of Road Traffic Fatalities in tries." Adverse Drug Reactions and Acute Poisoning Retieuws 8(3): 165-78. Delhi, India." Accident Anal'ysis and Prevennon 17:33-45. Kleevens, J. W. 1982. "Accidents in Hong Kong." Public Health, London Mohan, D., and M. Varghese. 1990. "Fireworks Cast a Shadow on India's 96(5):297-304. Festival of Lights." World Health Forum 11:323-26. Kortteinen, Timo 1988. "Intemational Trade and Availabilitv of Alcoholic Mowbrey, D. L. 1986. "Pesticide Poisoning in Papua New Guinea and the Beverages in Developing Countries." BntishJournal of Addiction 83:669-76. South Pacific." Papua New Guinea MedicalJournal 29:1 31-41. - 1989. "State Monopoly Systems and Alcoholism Prevention in Muller,A. 1982."AnEvaluationoftheEffectivenessofMotorcycleHeadlight Developing Countries: Report on a Collaborative Intemational Study." Use Laws."Ainericailjournalof PublicHeakh 72:1136-41. BntishJournal of Addicton 84:413-25. National Committee for Injury Prevention and Control. 1989. "Injury Pre- Kravitz, H. 1973. "Prevention of Falls in Infancy by Counseling Mothers." vention: Meeting the Challenge." American Journal of Preventive Medicine Illinois MedicalJournal 144:570-73. 5(supplement): 1-303. Krishnaraiah, V. 1972. "Industrial Accidents-A Survey." Ceylon Medical NHISA (National Highway Traffic Safety Administration). 1988. Fatal Acci- Journal 18-27. dent Reporting System, 1987. U.S. Department of Transportation, Washing- Landemann Szarcwald, C., and E. Ayres de Castilho. 1986. "Mortalidade por ton, D.C. causas extemas no estado de Rio de Janeiro no periodo de 1976 a 1980." NRC/tOM (National Research Council/institute of Medicine), Committee on Cadernos de Salude Publica, R.J. 1:19-41. Trauma Research, Commission on Life Sciences. 1985. Injury in Amenrica: Langley, J. D. 1988. "The Need to Discontinue the Use of the Term 'Accident' A Contnuing Public Health Problem. Washington, D.C.: National Academy when Referring to Unintentional Injury Events." Accident Analysis and Press. Prevention 20:1-8. National Safety Council. 1986. Accident Facts. Chicago, Ill. 632 Salls K. Stansfield, Gordon S. Smith, and WiUiam P. McGreevee Navarro, V. 1984. "Policies on Exportation of Hazardous Substances in West- 1990. "Car Design and the Risk of Pedestrian Dearhs." American em Developed Countries." New EnglandJournal of Medicine 311:546-48. Journal of Publc Health 80:609-10. Ng, S. C., T. C. Chao, and J. How. 1978. "Deaths by Accidental Drowning in . 1992. Injury Epidemiology. New York: Oxford University Press. Singapore, 1973-76." Singapore MedicalJournal 19:14-19. Robertson. L. S., A. B. Kelley, B. O'Neill, C. W. Wixom, R. S. Eiswirth, W. Nielsen. M. F. J., C. A. Resnick, and S. W. Acuda. 1989. "Alcoholism among Haddon, Jr. 1974. "A Controlled Study of the Effect of Television Messages Outpatients of a Rural District General Hospital in Kenya." BntishJourinal on Safety Belt Use." AmnericanJournal of Public Health 64:1071-80. of Addiction 84:1343-51. Rosenberg, M. L., R. J. Gelles, P. C. Hollinger, M. A. Zahn, and others. 1987. Nixon. J. W.. J. H. Peam, and A. E. Dugdale. 1979. "Swimming Ability of "Violence: Homicide, Assault, and Suicide." AmericanJournal of Preventive Children: A Survey of 4000 Queensland Children in a High Drowning Medicine 3(5 supplement): 164-78. Region." MedicalJournal of Australia 6:271-72. Ryan. G. Anthony. 1990. Assignment report, for Prevention and Control of Odejide, A. 0., J. U. Ohaeri. and B. A. Ikuesan. 1989. "Alcohol Use among Road Traffic Accidents Project. (WP)MNNHI1CP/APROOI-E. World Health Nigerian Youths: The Need for Drug Education and Alcohol Policy." Drug Organization, Geneva. and Alcohol Dependence 23:231-35. SAE (Society of Automotive Engineers). 1984. Advances in Belt Restraint Otosu, J. B., A. M. Abouammoh, and A. Bener. 1988. "A Study of Road Traffic Systems: Design, Pesfortnance, and Usage. Special Publication P- 141. War- Accidents in Saudi Arabia." Accident Analysis anid Preventon 20:95-101. randale, PA. Ogba, Leo Oko. 1989. "Violence and Health in Nigeria." Health Policy and Saleh, S., S. Gadalla, J. A. Fortney, S. M. Rogers, and D. M. Potts. 1986. Planning4(1 ):82-84. "Accidental Bum Deaths to Egyptian Women of Reproductive Age." Burns Okonkwo, C. A. 1988. "Spinal Cord Injuries in Enugu, Nigeria-Preventable 12:241-45. Accidents." Paraplegia26(1):12-18. Salgado. M. S., and S. M. Clombage. 1988. "Analysis of Fatalities in Road Oliver, R. G.. and B. S. Hetzel. 1972. "Rise and Fall of Suicide Rates in Accidents."ForensicSciencelnrerational36(1/2):91-96. Australia: Relationship to Sedative Availabilitv." Medical Journal of Aus- Sam, J. 1989. "Intemational Donot Aid to EMS Development." Paper pre- tralia 2:919-23. senred at the International Conference on Emergency Health Care Devel- Omran, A. R. 1971. "The Epidemiologic Transition: A Theory of the Epidemi- opmenr, Crystal City, Va. ology of Population Change." Milbank Memorial Fund Quarterly 49:509-38. Schelp, Lothar. 1987. "Community Intervention and Changes in Accident Owosina, F. A. 0. 1989. "National Policy on Injury Control and Emergency Pattem in a Rural Swedish Municipality." Health Promotion 2(2):109-25. Medical Services." Paper presented at the International Conference on Schwab, Larry. 1990. "Blindness from Trauma in Developing Nations." Inter- Emergency Health Care Development, Crystal City. Va. national Ophthalmology Cknics 30(1):28-29. PAHO (Pan-American Health Organization). 1986. Health Conditions in the Selya.R.M. 1980. "Deaths DueToAccidents inTaiwan: A Possible Indicator Americas. 1981-1984. Vol. 1. Scientific Publication 500. Washington, D.C. of Development." Social Science and Medicine 14D:36 1-67. Patel, N. S., and G. P. Bhagwatt. 1977. "Road Traffic Accidents in Lusaka and Shanmugasundaram, T. K. 1988. "The Care of SCI Patients in the Developing Blood Alcohol." MedicalJournal of Zambia 11(2):46-49. Nations-Can We Stem the Rot?" Paraplegia 26(1):10-11. Pearn, J. H., R. Bart, and R. Yamaoka. 1978. "Drowning Risks to Epileptic Shepherd, A. 1980. "Road Traffic Accidents: A View from the Highlands." Children: A Study from Hawaii." British MedicalJournal 2:1284-85. Papua New Guinea MedicalJournal 23:57-58. Peltzman, S. 1975. "The Effects of Automobile Safety Regulation."Journal of Shih, J. H., Z. Q. Wu, Y. L. Wang, Y. X. Zhang. S. Z. Xue, and X. Q. Gu. 1985. Pobtical Economy 83(4):677-725. "Prevention of Acute Parathion and Demeton Poisoning in Farmers around Phelps, C. E. 1988. "Death and Taxes: An Opportunity for Substitution." Shanghai." Scandinavian Journal of Workers' and EnVironmental Heakh 11 Journal of Health Economics 7:1-24. (Supplement 4):49-54. Pleuckhahn, V. D. 1984. "Alcohol and Accidental Drowning: A 25-Year Silva, J. F. 1978. "A Comparative Study of Road Traffic Accidents in West Study." Medical Journal of Australia 141:22-25. Malaysia." Annals of the Royal College of Surgeryy of England 60:457-63. Punyahotra, V. 1982. Epidemiology of Road Trafic Accident in Thailand. Na- Sinha, R. N. 1984. "Burns in Tropical Countries." Clincs in Plasuc Surgery rional Safety Council, National Accident Research Center, Bangkok, Thai- I(l) :1 -27. land. Sinha, S. N., S. K. Sengupta, and R. C. Purohit. 1981. "A Five Year Review Pupo Nogueira, C. 1987. "Prevention of Accidents and Injuries in Brazil." of Deaths Following Trauma." Papua New Guinea Medical Journal 24: Ergonomics 30(2):387-93. 222-28. Pusponegoro, A. D. 1989. "Pre-hospital Emergency Care in Indonesia: Con- Sloan, J. H., A. L. Kellerman, D. T. Reay, J. A. Ferris, T. Koepsell, F. P. Rivara, cepts and Problems." Paper presented at the Intemartional Conference on C. Rice, L. Gray, and J. LoGerfo. 1988. "Handgun Regulations, Crime, Emergency Health Care Development, Crystal City, Va. Assaults, and Homicide: A Tale of Two Cities." Alew England Journal of Ramesh, S., S. Srikanth, and V. R. Parvathy. 1987. "Poisoning in Children." Medicine 319:1256-62. IndianJournal of Pediatrics 54:769-73. Smith, G. S., and P. G. Barss. 1986. "Beyond the Motor Vehicle: The Reichanheim, M. E., and T. Harpham. 1989. "Child Accidents and Associated Importance of Other Unintentional Injuries as a Preventable Cause of Ill Risk Factors in a Brazilian Squatter Settlement." Health Policy and Planning Health in Developing Countries." Paper presented at Risks Old and New: 4(2):162-67. A Global Consultation on Health, April 27-May 1, Emory University, Rice, D. P., E. J. MacKenzie, and others. 1989. Cost of Injury in the United States: Atlanta, Ga. A Repor to Congress. Institute for Health and Aging, Universiry of Califor- . 1991. "Unintentional Injuries in Developing Countries: The Epide- nia and Injury Prevention Center (San Francisco) and Johns Hopkins miology of a Neglected Problem." Epidemiologic Reviews 13:228-66. University. Smith, G. S., and H. Falk. 1987. "Unintentional Injuries." Amnerican Journal Robertson, Leon S. 1981. "Automobile Safety Regulations and Death Reduc- of Preventive Medicine 5(supplement):143-63. tions in the States. " Amencan Journal of Public Health 71:818-22. Smith, G. S., and J. F. Kraus. 1988. "Alcohol and Residential, Recreational, .1983. Injuries: Causes, Control Strategies, and Public Policy Lexington, and Occupational Injuries: A Review of the Epidemiologic Evidence." Mass.: Lexington Books. Annual Review of Public Health 9:99-12 1. . 1984. "Automobile Safety Regulation: Rebuttal and New Data." Smith,R.S. 1974. "TheFeasibilityofanInjuryTaxApproachtoSafety."Law American Joumal of Public Health 74:1390-94. and Contemporary Problems 38(4):730-44. Injury 633 Sonnen, A. E. H. 1980. "Epilepsy and Swimming." In Epilepsy: A Climcal and Annual International Health Conference, National Council on Interna- Expenriental Research. Monographs in Neural Science 42. Basel: Karger. tional Health, June, Arlington, Va. Sood, S. 1988. "Survey of Factors Influencing Injury among Riders Involved West, J. G., M. J. Williams, and D. D. Trunkey. 1988. "Trauma Systems: in Motori-ed Two-Wheeler Accidents in India: A Prospective Study of 302 Current Status, Future Challenges."JAMA 259:3597-600. Cases."Joumnal of Trauma 28(4):530-34. WHO (World Health Organization). 1952. Technical Report 48. Geneva. Sowemimo, G. 0. 1983. "Bum Injuries in Lagos." Burns Including Thermal . 1977. lnternational Classificanon of Diseases. Manual of the International Injunes 9(4):280-83. Statistcal Classificanon of Diseases, Injuries, and Causes of Death. 9th rev. Sri Lanka Psychiatric Association. 1982. National Plan on Mental Health for Sn Geneva. Lanka. - . 1986. Report of the Second Global liaison Meeting on Accident and Injury Subianto, D. B., L. R. Tumada, and S. S. Margono. 1978. "Burns and Epileptic Prevention. IRP/APR 218 m2IA. Geneva. Fits Associated with Cysticercosis in Mountain People of Irian Jaya." . 1987a. Accident and Injury Prevention at the Primary Health Care Level. Tropical Geography and Medicme 30(3):275-78. IPRlAPR 218 H. Inter-Regional Consultation on Research Development for Taket, A. 1986. "Accident Mortality in Children, Adolescents, and Young Injury Ptevention, Pattaya, Thailand. Adults." World Health Statistical Quarterly 39:2 32-56. . 1987b. Report of the Asian Seminar on Road Safety. [PR/APR 218 G. WHO Technical Study Group. 1987. Toward a Less Fire-Prone Cigarette: Final Report Regional Office for Europe, Geneva. of the Technical Study Group on Cigarette and Little Cigar Fire Safety. U.S. . 1988. Global Medium-Term Programme: Accident Prevention. APR/MTP/ Consumer Product Safety Commission, Washington, D.C. 88. 1. Geneva. Teret, S. P., and M. Jacobs. 1989. "Prevention and Torts: The Role of . 1989a. "ManifestoforSafeCommunities: Safety-A Universal Con- Litigation in Injury Control." Law, Medicine, and Health Care 17:17-22. cern and Responsibility for All." Resolution adopted at the First World Conference on Accident and Injury Prevention, September 20, Stockholm. Thapa, N. B. 1984. "Injury Prevention in Nepal." Souvenir Napais Journal ___18bNwprahsomroRaSft.Tcncleot71 3 (1):136-39. - 1 989b. New Approaches to Improve Road Safety . Techn ica l Report 78 1. Trunkey, D. D. 1983. "Trauma." Sciendific American 249:28-36 Geneva. Tmnke'~ D 0. 983."Traua." cienificAmerian 29:2836. ____. I1989c. Research Development for Accident and Injury Prevention. UNICEF (United Nations Children's Fund). 1989. State of the World's Children. [PR/APR 216 m3 sR, 892 3DE Geneva. New York: Oxford University Press. Wintemute, G. J., S. P. Baker, D. Mohan, S. P. Teret, and C. J. Romer, eds. Waller, J. A. 1985. Injury Control: A Guide to the Causes and Prevention of 1985. Pnnciplesfor Injury PrevennoninDevelopingCountries. IPRJADR 217-40. Traurna. Lexington, Mass.: D. C. Heath. World Health Organization, Geneva. Wamer, K. E. 1990. "TobaccoTaxationas Health Policy in theThird World." Wong, M. K., C. F. Chen, and I N. Lien. 1981. "Evaluation of the Results of AmencanJournal of Public Health 80:529-31. a Rehabilitation for Spinal Cord Injury over a Recent Ten Year Period." Wechsler, H., E. H. Kasey, D. Thum, H. W. Demone, Jr. 1969. "Alcohol Level Journal of the Formosan Medical Association 80:433-41. and Home Accidents." Public Health Reports 84:1043-50. Wyatt, G. B. 1980. "The Epidemiology of Road Accidents in Papua New Weddell, J. M., and A. McDougall. 1981. "Road Traffic Injuries in Sharjah." Guinea." Papua New Guinea MedicalJournal 23:60-65. IntewrationalJournal of Epidemiology 10:155-59. Xue, Shou-Zhen. 1987. "Health Effects of Pesticides: A Review of Epidemio- Wernet, David. 1987. Disabled Village Children. Palo Alto, Calif.: Hesperian logic Research from the Perspective of Developing Nations." American Foundation. Journal of Industrial Medicine 12:269-79. .1989. "Health forNoOne by the Year 2000: The High Cost of Placing Zador, P. 1983. "How Effective are Daytime Motorcycle Headlight Use Laws?" 'National Security' before Global Justice." Paper presented to the 16th American Journal of Public Health 73:808-10. 26 Cataract Jonathan C. Javitt "Cataract" refers to an opacity in the natural, crystalline lens ual, the family, and the community that far outweigh the of the eye. Although cataract surgery has been performed for investment in surgery. A study in rural India found that indi- more than 2,000 years, cataract remains the most common viduals undergoing cataract surgery demonstrated increased cause of blindness in the world today (Dawson and Schwab productivity amounting to a 1,500 percent annual return on 1981; Whitfield and others 1983; Al Salem and Ismail 1987; the cost of the surgery (Javitt, Venkataswamy, and Sommer Schwab 1987). The World Health Organization (WHO) esti- 1983). mates suggest that 17 million people are currently blinded by cataract (Wilson 1980) worldwide. New data show that in Public Health Significance of Cataract China alone there are 5.4 million cases of cataract blindness, suggesting that the world total may be higher than previously The current backlog of curable blindness from cataract is the thought (cDc 1983). As of 1983, incident cases of cataract- result of an interplay between disease incidence, surgical rates, related blindness exceeded 1.25 million annually (cDc 1983). and mortality. As illustrated in figure 26- 1, most individuals in Because of increasing life span and an expanding elderly pop- the developing world who are blind from cataract are likely to ulation in the developing world, the prevalence of blinding be so for the rest of their lives. Only the minority are likely to cataract is expected to double by the year 2010. have sight restored. Decreasing the backlog of cataract blind- The crystalline lens of the eye is normally transparent and, ness can be achieved both by increasing the volume of curative together with the cornea, focuses light on the retina. Although surgery and decreasing risk factors for cataract wherever possi- a small degree of opacity may interfere minimally with vision, ble. These strategies must include not only increased provision cataract can and frequently does cause severe vision loss. of care but also operations research to determine methods for Cataract blindness, according to the definition of the World increasing individual participation in care. Health Organization, results when the degree of opacity re- duces vision to less than 3/60 (i.e., inability to recognize the Current Levels and Trends in the Developing World largest letter) on the standard eye chart. Cataract progression is characterized by painless, progressive Enormous variation exists in the prevalence of cataract blind- loss of vision. In general, vision loss induced by cataract is ness worldwide. As can be seen from figure 26-1, rates range entirely reversible upon removal of the opacified lens. Only from 14 per 100,000 people (0.014 percent) in Scandinavia to occasionally can cataract-induced lens swelling or leakage 1,525 per 100,000 (1.525 percent) in Asia (WHO Programme cause permanent damage to the eye. for the Prevention of Blindness 1987). The prevalence of a Like many degenerative and disabling conditions associated disease in the population is a function of its incidence in that with age, blindness from cataract is associated with complete population, the duration of disease in those affected, and the disability, an increased need for support from family members, likelihood that the disease will resolve on its own, be cured, or loss of social status and authority within the family and com- result in an increased risk of mortality. Variations in any one munity, and early demise. Unlike many degenerative and of these factors may lead to variation in observed prevalence. disabling conditions, however, cataract blindness is entirely There are several possible explanations when the prevalence curable. The current cost of restoring sight in a mass-surgery of cataract blindness is higher in one region than another: program ranges from $15 for cataract extraction in an Indian eye camp (temporary mass surgical facility) to $22 in an * An increased incidence of cataract by age group African mobile surgical facility to $33 in an urban Latin * Anincreasedlongevityofthosewithcataractorofolder American public hospital (HKI 1986). The limited studies persons in general conducted to date suggest that restoration of sight by cataract * A decreased likelihood that individuals with cataract surgery produces economic and social benefits to the individ- will have sight-restoring surgery 635 636 Jonathan C. Jat.tt Figure 26 1. Prevalence of Cataract Blindness, served difference could have been observed if chance alone was by Region operating is less than I percent (p < .01). Cataract blindness per 100,000 population Cataract Surgery in Developing Countries 1,600 Approximately 1 million cataract extractions were performed 1,400 in 1988 in both India (Venkataswamy, 1987) and the United States (HCFA 1986), although the population of India is more 1,200 than triple that of the United States. Even if the prevalence of cataract were equal in the two countries, the lower rate of 1,000 surgery might account for a good portion of the excess in cataract blindness. Fewer than 10 percent of blinding cata- 800 _ racts are extracted annually in developing countries (HKI 1986). Although cataract has recently been shown to be the 600 leading cause of blindness in an American urban population, the overall prevalence of cataract blindness is an order of 400 i-, F1 magnitude lower than in the developing world (Sommer and 200 r others 1991 ). [Jl l One way to estimate the effect of variations in rates of 0 . . L _ cataract surgery on overall prevalence of cataract blindness is Europe United Former Peru Australiaa Africa Asia to examine the prevalence of cataractous change by age. If States U.S.S.R. individuals with cataract and those who have had cataract extraction are combined, such a survey should yield a reliable a. Aboriginal population. indication of the overall prevalence of cataract by age in the Source: Wilson 1980. population. Few surveys of this type have been conducted because the * Errors and biases in data collection that may affect data survey methodology to detect any clinically significant degree validity of cataract is far more complex than that required to detect blinding cataract. In the United States, researchers for the In considering the epidemiology of cataract blindness, one Framingham Eye Study examined a representative sample of must study variations in the incidence of cataractous change, residents of Framingham, Massachusetts, a primarily white, in the rate of progression to cataract blindness, in the likeli- middle-class community (Sperduto and Hiller 1984). Re- hood that an individual will receive cataract surgery, and in searchers for a second American study, the National Health the longevity of individuals who are blind from cataract. and Nutrition Examination Survey, drew a random sample of The prevalence of cataract blindness varies substantially, all Americans, based upon census data (Ederer, Hiller, and not only from continent to continent but within smaller Taylor 1981;Hiller,Sperduto,andEderer 1983). Infigure 26-2 regions as well. The prevalence of blindness reported from the U.S. data are compared with those obtained in the Nepal survey data published by the World Health Organization is Blindness Survey, a random cluster sample of the entire king- shown in table 26-1. The data presented are all drawn from dom of Nepal (Brilliant and others 1988). This is the only government-sponsored blindness surveys that specifically ad- epidemiologic survey in developing countries in which the dressed the rate of cataract-related blindness in relation to entire population of a country has been sampled and studied other causes of blindness. A critical consideration of the sam- for eye disease. pling techniques and survey methodology employed in each As is readily apparent from figure 26-2, even when catarac- country is beyond the scope of this chapter. The reader must tous and aphakic individuals are combined, the prevalence of evaluate the reliability of the underlying data, however, before cataract is substantially higher in Nepal than in the United drawing conclusions based upon small variations in cataract States in each age group. Thus, the difference in surgical rate rates from one region to another. alone does not account for the higher rate of cataract blindness Even within a country, rates of cataract blindness obtained observed in at least one developing nation. Because some from survey data may vary considerably. In table 26-2 the rates degree of lens change inevitably accompanies aging, these data from nine provinces of China are shown (WHO 1987). In may also be interpreted to suggest that individuals in Nepal comparing Hunan and Sichuan, it is important to note that, develop cataract at a younger age than their counterparts in although the overall prevalence of blindness is the same, the the United States. prevalence of cataract blindness in Sichuan is double that of If the prevalence of cataract itself is truly greater in the Hunan. Because of the large number of people examined, the developing world, there are only two possible explanations. reliability of these data is unusually high and, assuming that Either the incidence of cataract by age is greater in developing unbiased samples were obtained, the likelihood that the ob- countries, or persons with cataract in such countries live longer Cataract 637 Table 26-1. Prevalence of Cataract Blindness, by Country Prevalence Percent of blindness Prevalence of cataract Population Population with cataract CountrY of blindness from cataract blindness (millions) blindness (millions) Afoca Botswana 1.4 0.45 0.63 1.01 0.01 Chad 2.3 0.48 1.104 4.79 0.50 Egypt 3.3 0.32 1.056 44.50 0.47 Ethiopia 1.3 0.46 0.598 33.68 0.20 The Gambia 0.7 0.55 0.385 0.80 0.00 Kenya 1.1 0.67 0.737 18.78 0.14 Liberia 2.1 0.45 0.945 2.06 0.02 Malawi 1.3 0.40 0.52 6.43 0.03 Mali 1.3 0.32 0.416 7.53 0.03 Nigeria 1.5 0.41 0.615 89.02 0.55 Sudan 6.4 0.30 1.92 20.36 0.39 Togo 1.3 0.45 0.585 2.76 0.02 Tunisia 3.9 0.52 2.028 6.89 0.14 Zimbabwe 1.2 0.40 0.48 7.14 0.03 Amencas Brazil 0.3 0.10 0.03 129.70 0.04 Peru 1.0 0.34 0.34 18.70 0.06 United States 0.2 0.13 0.026 233.70 0.06 Asia Afghanistan 2.0 0.31 0.62 17.22 0.11 Bangladesh 0.9 0.33 0.30 94.65 0.28 Chinaa 0.875 0.22 0.14 1040.00 1.41 Hong Kong 0.2 0.34 0.07 5.31 0.01 India 0.5 0.55 0.27 732.00 2.01 Indonesia 1.2 0.67 0.80 159.00 1.28 Japan 0.3 0.23 0.07 119.00 0.08 Korea 0.1 0.361 0.04 40.00 0.01 Nepal 0.8 0.67 0.54 15.74 0.08 Pakistan 2.3 0.60 1.38 89.00 1.23 Saudi Arabia 1.5 0.55 0.82 10.40 0.09 Sri Lanka 2.0 0.46 0.92 15.00 0.14 Syrian Arab Rep. 0.3 0.35 0.10 9.60 0.01 Thailand 1.1 0.57 0.62 49.00 0.31 Viet Nam 0.8 0.39 0.31 57.00 0.18 Yemen, Rep. of 3.6 0.34 1.22 2.16 0.03 Europe Germany 0.1 0.04 0.004 61.42 0.002 Norway 0.2 0.07 0.014 4.13 0.001 Sweden 0.3 0.05 0.015 8.00 0.001 U.S.S.R.8 0.27 0.16 0.0432 272.50 0.12 a. Average. b. Entire country (European and Asian parts). Source: WHO Programme for the Prevention of Blindness 1987. than those in the industrial world. Because the rate of mortality sampled, recruited for study, and examined, incidence surveys from all causes of those between the ages of forty-five and require that those same individuals be located and reexamined sixty-five in Asia is 1.65 times that of the industrial world and after a defined time interval. The logistics and expense of the mortality rate of those over sixty-five 1.25 times greater, conducting a longitudinal study of this nature are orders of increased longevity is an unlikely explanation for the increased magnitude greater than for a prevalence study. prevalence of cataract. Increased incidence of cataract in developing countries must play a significant role. Possible Explanations for Increased Cataract Incidence Although the evidence strongly suggests an increased inci- dence of cataract in the developing world, few actual studies One cannot rule out genetic differences as an explanation for of cataract incidence have ever been conducted. Whereas increased cataract incidence in developing countries, but these prevalence surveys require that a large population be randomly factors are least amenable to modification. During the past 638 Jonathan C. Javitt Figure 26-2. Prevalence of Cataract by Age, United clear albumin of an egg white congeals, opacifies, and turns States and Nepal white with heat, the natural crystalline proteins of the lens are Prevalence known to coalesce and discolor in response to certain stimuli. Cross-linking and denaturation of lens proteins have been 50 produced in laboratory settings by ultraviolet light and chem- ical oxidants. Thus, "oxidative-stress" from a variety of causes may be a common mechanism in the formation of cataract. 40 - The human lens is constantly exposed to oxidant stress, both from environmental light and from naturally occurring free radicals that are ubiquitous in the human body. Highly effi- 30 - cient enzyme systems are present within the eye to prevent damage from these agents. Inquiry into the etiology of cataract has therefore focused upon the environmental and nutritional 20 - exposures of individuals with cataract as well as any deficien- cies that may exist in their antioxidant enzyme systems. 10 - ASSOCIATION WITH DIABETES. Diabetes has been associated with an increased risk of cataract in Americans under fifty-five years of age (Hiller, Sperduto, and Ederer 1983) as well as with 0 an increased rate of cataract surgery (Hiller and Kahn 1976). 50-64 65-74 75-85 However, diabetes-related cataracts account for only 6 percent Age (years) of all cases in the United States and an even lower proportion of cases worldwide. Despite this, cataract formation in diabet- Framingham National Health _ Nepal ics is of substantial research interest, because of evidence that Eye Study (U.S.) Nutrition Blindness the underlying biochemical mechanism may involve the ab- Survey (U.S.) normal formation of sugar alcohols (sorbitol) in the lens, resulting in subsequent lens swelling and opacification. Note: No data for population age 78-85 in the National Health Nutrition Human studies are now under way to determine if inhibition Examination Survey. of t . . . . . . . . Source: Framingham Eye Study (Sperduto and Hiller 1984); National Health of this biochemical pathway IS feasible and if such inhibition Nutrition Examination Survey (Hiller, Sperduto, and Ederer 1983); Nepal reduces the rate of cataractogenesis in humans. The longevity Blindness Survey (Briltiant 1988). wrdi ogvt of diabetics in the developing world is substantially lower than decade, epidemiologists and lens biochemists have worked to elsewhere, and thus diabetes is unlikely to account for the gain a better understanding of the process of cataract formation excess cataract rate. and the role played by nutritional and environmental factors. Although "age-related" or "senile" cataract is the most prev- ASSOCIATION WITH SUNLIGHT. The association of cataract alent form, cataracts may be metabolic, traumatic, nutritional, and sunlight has long been suspected on the basis of case- or toxic in etiology. In general, cataract results from a denatur- control studies. Hiller and colleagues found a higher cataract- ation of natural proteins within the lens of the eye. Just as the to-control ratio for persons age sixty-five or older in areas with longer duration of sunlight (Hiller, Giacometti, and Yuen 1977; Hiller, Sperduto, and Ederer 1983). Taylor (1980a) Table 26-2. Prevalence of Cataract Blindness reported an association of cataract with increased ultraviolet in Nine Provinces of China light, latitude, and average hours of sunlight. Other investiga- Percent Prevalence tors have noted higher prevalence of cataract among Tibetans Prevalence of blindness of cataract Population living at altitudes of 4,000 meters than those living at altitudes Province of blindness from cataract blindness examined of 2,000 meters (Wen-shi 1979). Although a survey of 29,683 Fujian 0.2 0.35 0.07 50,620 residents of the Punjab revealed a lower prevalence of cataract Sichuan 0.3 0.4 0.12 21 869 among those who lived in mountain regions than among those Hunan 0.3 0.2 0.06 94,222 on the plains (Chatterjee, Milton, and Thyle 1982), this study Anhui 0.7 0.36 0.25 13,852 did not control for the effect of cloud cover on the actual hours Tianjin City 0.3 0.39 0.12 74,348 of sunlight exposure in each region. Researchers in the Nepal Guangdong 0.4 0.571 0.23 10,180 Blindness Survey calculated mean sunlight exposure for each Guangxi 0.5 0.37 0.19 26,210 village sampled on the basis of altitude, skyline obstructions, Heilonjiang 0.2 0.39 0.08 40,097 and cloud cover. As can be seen from table 26-3, there was a Huairon Custrong association between average daily sunlight hours and Note: Based on government-sponsored surveys, cataract prevalence (Brilliant and others 1983, 1988). Even so, Source: WHO Programme for the Prevention of Blindness 1987. this study controlled only for the sunlight exposure of the Cataract 639 Table 26-3. Cataract Prevalence by Average Daily detected an association with lower levels of ascorbic acid. In a Hours of Sunlight case-control study of Americans with and without senile cat- Population Cataract Prevalence Odds aract reported by Jacques and co-workers, the risk of cataract Sunlight examined cases (per 100) ratio was reduced for individuals with higher blood levels of carot- enoids, vitamin D, and vitamin E, whereas the risk was in- Low (7-9 hours) 7,236 133 1.84 1.0 creased for those with lower levels of vitamin C (Jacques, High (12 hours) 10,286 476 4.63 2.6 Hartz, Chylack, McGandy, and Sadowski 1988). Carotenoids, along with vitamins C and E, are potent anti- a. Based on sunlight exposure for lifelong residents of ninetv-seven rural oxidants and thus quench free radicals. Therefore, if the theory villages. ofoxidative stress is valid, it should not be surprising that their Source: Brilliant and others 1988. o xdtv tesi ai,i hudntb upiigta hi levels are associated with cataract risk. Jacques, Chylack, McGandy, and Hartz (1988) reported that higher levels of an village and did not take into account the exposure of individ- "antioxidant index," composed of vitamins C and E and carot- uals based upon their occupations and use of such protective enoids, along with antioxidant enzymes found in red blood clothing as hats. cells were associated with lower risk of cataract. Corroborating Taylor and co-workers (1988) addressed the issue of individ- data regarding this antioxidant index were reported from the ual exposure to sunlight and ultraviolet radiation in their Indian and U.S. case-control study (Mohan and others 1989). survey of 838 Maryland watermen who earn their living by fishing on the Chesapeake Bay. Using dosimeters, the re- ASSOCIATION WITH SEVERE DIARRFIEA AND DEHYDRATION. searchers correlated the watermen's actual ocular exposure to Minassian,Mehra,andjones (1984) have advanced the theory ultraviolet radiation (UVR) with working hours, sheltered vs. that severe diarrhea and subsequent dehydration might lead to unsheltered work sites, and protective devices worn. They an elevated level of blood urea nitrogen and, thus, to alteration found a clear association between increased exposure to UVR of lens proteins and cataract. This theory, attractive from the and the presence of cortical cataract. As reported in a related biochemical point of view, has not been corroborated in epi- paper by Bochow and others (1989), the outdoor worker demiologic studies. Khan, Khan, and Sheikh (1987) found no wearing sunglasses and a hat has only twice the exposure to correlation between cataract risk and cholera-related diarrhea UVR of the indoor worker. When no eye protection is wom, in a case-control study performed in Bangladesh. Similarly, however, the exposure of the outdoorworker to UVR is eighteen Bhatnagar and colleagues (1988) found no association be- times that of the indoor worker. The authors of this related tween cataract and remembered episodes of severe diarrhea study compared patients suffering from posterior subcapsular from a study in South India. cataract who were exposed to UVR during sunlight hours with normal controls who were similarly exposed. A strong associ- Lowering or Postponing Disease Incidence ation (p < .001) was detected between ocular sunlight expo- sure and cataract. Currently, there are no proven interventions that prevent cataract or delay its onset. Although cataract surgery is likely ASSOCIATION WITH NUTRITIONAL AND METABOLIC FACTORS. to remain the most cost-effective means of treating an existing Although nutritional factors might be intuitively associated cataract, strategies that decrease or delay the onset of cataract with cataract, their significance is quite difficult to prove. are of vital importance. Because of the strong association Blood levels of vitamins reflect only current nutritional status between cataract and aging, a ten-year increase in the average and cannot detect previous periods of hypovitaminosis. An life span in developing countries is likely to double the preva- association between cataract and diet was observed in the lence of cataract. Similarly, a ten-year delay in the onset of Nepal Blindness Survey, in which vegetarians who never ate cataract would halve its prevalence in the population. Al- meat or fish were found to have twice the cataract prevalence though this may seem an impossible task, one must remember of those who ate fish or meat, even occasionally (Brilliant and that eighty-year-old residents of Framingham, Massachusetts, others 1988). A caveat in interpreting these data is that have the same prevalence of cataract as seventy-year-old resi- vegetarianism was most common in the regions with the dents of India and Nepal. Although nutritional and metabolic highest sunlight exposure and it was impossible to separate deficiencies may serve as a risk factor for cataractogenesis, these two possibly interacting variables, interventions in this area go far beyond the scope of eye In their study of the Punjab, Chatterjee, Milton, and Thyle disease. Even if improved nutrition does delay the onset of (1982) reported a relative increase in prevalence of cataract cataract, the interventions required are identical to those among individuals with lower protein consumption. The au- required to combat all other malnutrition-associated condi- thors of the Indian and U.S. case-control study of age-related tions in developing countries. cataracts (Mohan, Sperduto, Angra, and others 1989) similarly If, in fact, exposure to ultraviolet radiation is a risk factor for detected an increased risk of posterior subcapsular and nuclear cataractogenesis, public health interventions that decrease cataract in those individuals who had a history of a diet ocular UVR exposure may have a useful role. They are intriguing deficient in protein. Biochemical analysis from the same study from the international development point of view in that they 640 Jonathan C. Javitt may be quite inexpensive. Rosenthal and co-workers (1988) the lens capsule, expressing the lens nucleus and aspirating have shown that a brimmed hat reduces ocular UVR expo- remaining lens cortex, leaving intact, if all goes well, the lens sure by approximately 50 percent, and the addition of UVR- capsule and its zonular attachments. The ECCE method enables absorbing sunglasses further lowers transmission to 1 percent the surgeon to insert an intraocular lens into the remaining of ambient UVR. Whereas UVR-absorbing sunglasses are expen- lens capsule and is thought to preserve better the anatomy of sive to manufacture and obtain by the standards of developing the eye. countries, locally manufactured hats of straw or other ubiqui- Christy and Lall (1973) reported an infection rate of 0.46 tous materials are practically free. Furthermore, although prop- percent for 54,000 icCEs performed at a mass-surgery program erly manufactured uvR-absorbing sunglasses may potentially in Pakistan. Although this is higher than the 0.17 percent block 86 percent to 99 percent of ambient UVR, a 0.6 centime- infection rate following ICCE in the United States (Javitt and ter displacement of sunglasses away from the forehead results others 1991a), it is certainly acceptable by local standards. in a substantial increase in ocular UVR exposure (Rosenthal, In this setting, functional vision is restored to between 85 Bakalian, and Taylor 1986). Additional studies need to be percent and 92 percent of patients who undergo surgery (Javitt, performed in which individual exposure to UVR is measured Venkataswamy, and Sommer 1983; Al Salem and Ismail 1987; during outdoor activities and inexpensive interventions are Whitfield 1987; Brilliant and others 1988). Suboptimal out- tested. comes are a function of preexisting retinal disorders, as well as complications of surgery. The standard intracapsular cataract Reducing the Burden of Cataract Blindness extraction commonly performed in developing countries is a mature surgical procedure that requires little in the way of The National Eye Institute and its collaborating institutions technical improvement. have embarked on a long-term research effort to develop drugs Because an aphakic eye (one that has undergone cataract that may delay the onset of cataract. Although this research extraction) is left with an extreme refractive error, corrective has yielded invaluable insight into the biochemistry of the lens spectacles, contact lenses, or intraocular lens implants (loLs) and possible metabolic pathways of cataractogenesis, clinical are required for visual rehabilitation of the patient. In most trials are only in the earliest planning stages. A chemothera- developing nations, locally manufactured spectacles in a stan- peutic strategy that could delay the onset of cataract by ten dard aphakic power are available for $5 to $12. At present, loLs years in the United States would have the potential to save are considerably more expensive, and contact lenses with their $500 million or more annually in surgical costs. Because of the need for frequent replacement and sterile solutions are totally low cost of cataract extraction in developing countries and the impractical. Davies and colleagues (1986) have shown that likely high cost of chronic use of any new drug, it will probably ioLs are more cost-effective than contact lenses in the National be many years before this can be a cost-effective strategy for Health Service of the United Kingdom and may offer consid- the developing world. erable advantages in patient comfort and reduction of subse- quent complications. Cataract Surgery in Developing Countries Although conventional wisdom has long held that in- tracapsular cataract extraction with provision of aphakic spec- Once cataract has developed, the only known treatment is tacles is the appropriate technology for the developing world, surgical removal. Helen Keller International has reported that this approach needs re-evaluation. Aphakic spectacles are the cost of cataract extraction ranges from $15 in a mass- notably thick and uncomfortable to wear. Although straight- surgery setting on the Indian subcontinent to $22 in an Afri- ahead vision can be corrected to 20/20, they cause considerable can mobile surgical facility, to $33 in an urban Latin American distortion of peripheral vision and complete obscuration of public hospital (HKI 1986). Fortunately, cataract extraction is objects between 30' and 45 in the periphery. Moreover, highly successful, even with the limited resources, lower stan- magnification induced by aphakic spectacles makes objects dards of sterility, and older instruments found in developing such as steps and curbs appear closer than they are. Ellwein and countries. Substantial improvements in outcome of surgery others (1991) have noticed that as many as half of those who were achievedduringthe 1960s and 1970s, when microsurgical receive aphakic spectacles in cataract surgery programs in the techniques and watertight closure of cataract wounds with fine developing world do not wear them and hence, suffer ex- silk or nylon sutures were universally adopted. tremely limited postoperative vision. There remain two main methods of cataract extraction In recent years, there has been increased interest in convert- today: extracapsular cataract extraction (ECCE), which has ing to extracapsular cataract extraction with intraocular lens been adopted by nearly all the industrial world, and intracapsu- implantation, which is currently the dominant procedure in lar cataract extraction (ICCE), which is employed in less than industrial nations. Analysis of 330,000 cases of cataract extrac- 10 percent of cases in the United States and considerably more tion in the United States reveals that patients who undergo frequently in the developing world. The latter type of extrac- ICCE have a 1.7-fold higher likelihood of infection and retinal tion involves removing the entire lens with disruption of the detachment than those who undergo ECCE (Javitt and others zonular fibers which form the attachment of the lens capsule 1991a and 1991b). Although there is likely to be an improve- to surrounding ocular structures. The former entails incising ment in outcome and reduced risk of complication following Cataract 641 ECCE, this procedure requires the use of an operating micro- this approach may be the only practical one in settings where scope, more delicate surgical techniques, and sterile irrigating long-term follow-up of patients is infeasible. solutions. Approximately three tosix monthsoffull-time training Current techniques achieve the purpose of restoring func- is required to teach the newer extracapsular technique to a skilled tional vision to most of those who are blinded by cataract. ophthalmologic surgeon familiar with older methods. Unless an individual is accustomed to reading or performing Extracapsular cataract extraction enables the surgeon to similarly demanding tasks, the visual outcome of ICCE with place an intraocular lens in the posterior lens capsule (see spectacle correction may be acceptable. As the price of an figure 26-3), where it is least likely to cause ocular discomfort intraocular lens continues to decline, however, it may even and long-term corneal complications. However, the lens cap- rival the cost of aphakic spectacles. With improved technology sule itself is likely to opacify over a period of years in 25 percent ECCE may become the preferred method of cataract extraction of those who undergo the procedure. In the industrial world, in the developing world, as well. this circumstance is routinely managed by using a solid state (Neodymium: YAG) laser to create an opening (capsulotomy) Economic Return on Cataract Surgery in the opacified lens capsule. An alternative is to incise the lens capsule with a needle-knife, which is easily performed in The economic cost associated with cataract reflects the near- the physician's office. This latter approach was routinely em- total disability associated with this condition. My colleagues ployed in the United States and Europe until introduction of and I performed a pilot study to determine the cost-benefit the solid state laser in the early 1980s. While the capsulotomy ratio of restoring sight via cataract surgery in developing coun- procedure is quite simple to perform, opening the lens capsule tries (Javitt, Venkataswamy, and Sommer 1983). The Aravind in this manner increases the risk of subsequent retinal detach- Eye Hospital in Madurai, India, is a private charitable institu- ment to approximately the same level as that following in- tion that currently performs more than 20,000 free cataract tracapsular cataract extraction. extractions annually on indigent patients who are functionally An alternative to postoperative capsulotomy in the 25 per- blind at the time of surgery. The cost of surgery is funded by cent of patients who are likely to develop capsular opacity is revenues from paying patients and private voluntary organiza- to perform a capsulotomy at time of surgery after placement of tions. One hundred patients were randomly selected from the intraocular lens. The disadvantage of this approach is that among those who visited the hospital for follow-up care be- all patients will be subject to higher risk of retinal detachment tween six months and two years after surgery. By means of an associatedwithdisruption oftheposteriorcapsule. Conversely, interview in their native language, the patients were queried as to the effect of losing their sight on their economic and social circumstances. In addition to earning ability before and Figure 26-3. Lense Implant with Subsequent after surgery, patients were asked about other members of the Capsulotomy family who were able to return to work after the patient's surgery or who were forced to leave work when the patient beaer initially became visually incapacitated. The interview data Cloudy lens were correlated with the outcome of surgery. capsule Retina Eighty-five percent of the patients surveyed achieved post- Lens operative visual acuity of 6/36 or better. Eight percent had implaAnt - vision of 6/60, and 7 percent had visual acuity of less than 6/60. /(\ \ > t i Patients were included in the study data regardless of the surgical outcome. Eighty-five percent of the males and 58 percent of the - t>V~<' i / . kltn 9\ r ,females who had lost their jobs as a result of blindness regained those jobs. A number of those who did not return to work did free other family members from household duties, thereby enabling them to return to work. Eighty-eight percent of male Holemade patients and 93 percent of female patients who reported having \\roross-s ction through lost authority within their family and their community stated through the eye capsule that they had regained their social standing. after cataract removal 1-1rato curt.sy At the time of the study the marginal cost of performing a and insertion of lens implant G.. g.t.o, U-,e-sty MeO,ca Ce.ter cataract extraction was $5 dollars (53 rupees). Economic data were compared with that investment cost. The results showed that the average individual regaining functional vision through cataract extraction in this setting generated 1,500 Note: Following extrapsular cataract extraction, the lense implant is placed percent of the cost of surgery in increased economic produc- into the remaining capsule of the natural lense (figure left), which may subsequently opacity. The Neodynium: YAG laser can then be used to make tivity during the first year following surgery. This benefit was an opening (capsulotomy) in the opacified lense capsule without harming the implant (figure nght). generated both by the patients and by their family members 642 Jonathan C. Javitt who were able to return to work. No data yet exist on the Table 26-4. Cost-Utility of Cataract Surgery annual return on investment for the long term. and Other Types of Health Interventions Cost per DALY saved Cost-Effectiveness, or Cost Utility, of Cataract Surgery Intervention (dollars) Elsewhere in this collection, a year of the life of a functionally Bilateral cataract surgery blind person has been equated to a loss of 0.5 disability- $26.80 per eye 21.50 adjusted life-year (DALY). Torrance (1982) reported empirical $42.50 per eye 31.80 data suggesting that individuals associate a utility value of 0.39 (with perfect health represented as 1.0) with a lifetime of being $18.00 per eye 15.00 "blind, deaf, or unable to speak." Drummond and co-workers $26.80 per eye 19.00 (1987, 1988) employed a utility scale calibrated in disability- $42.50 per eye 27.00 adjusted life-years (DALYs) and reported that a year of life for a .. . . ~~~~~~~Noncataracr intervention poorly adjusted (poorly rehabilitated) blind person is valued Passive case finding and short-term 10.00 at 0.35 DALY, whereas a year of life for a well-adjusted (well- chemotherapy for TB rehabilitated) blind person is valued at 0.48 DALY. DPT and polio immunization 20.00-40.00 For consistency with other chapters, in my comparison of Oral hypoglycemic management of NIDDM 330.00 cataract extraction with other potential health care interven- tions I have employed the DALY model in which healthy years of future life are discounted at a rate of 5 percent. I have assumed a year of life following successful bilateral surgery to Increasing Availability of Cataract Surgery equal 1.0 healthy years of life and a year of life following successful unilateral surgery to equal 0.84 healthy years of life. Although shortage of qualified surgeons is commonly invoked This is consistent with data from a U.S. study of cataract as an explanation for the backlog of blinding cataract in patients in which the improvement in visual function follow- developing countries (Wilson 1980; Foster 1987), it beats ing cataract surgery in the second eye was approximately 50 careful scrutiny. The ratio of ophthalmologists to population percent of that associated with improvement following surgery in India is 1:100,000, the same as that for Great Britain inthefirsteye (Javitt, Street, Brenner, andothers 1993).1 have (Venkataswamy 1987). Yet, in the face of an alarming preva- assumed bilateral cataract surgery costs in the developing lence of cataract blindness, Indian ophthalmologists remain world to be between $18.00 and $42.50 per eye in 1990 and concentrated in the cities, where many are underemployed, if the cost of spectacles to be $6.00. not unemployed. There remains an undersupply of ophthal- In assessing the benefits accruing from any procedure, the mologists in rural areas. The cost of starting a surgical facility survival of the underlying population must be considered. Data is beyond the resources of most individuals, and government from 330,000 U.S. Medicare patients undergoing cataract jobs that would allow one to practice in a public hospital are extraction indicate 75 percent survival at five years after scarce. Thus, the Indian subcontinent is faced more with a surgery with an absolutely straight survival curve (Street and maldistribution of surgeons than with a true shortage. Africa, Javitt 1992). Based on the U.S. data, I have assumed the to the contrary, truly presents an example of worker shortage, average patient undergoing cataract surgery in the developing where there is only one ophthalmologist per I million people, world to have a 5 percent annual mortality risk after surgery. or one ophthalmologist for every 4,000 blind individuals (Fos- Because cataract patients in the developing world tend to be ter 1987; Schwab 1987). ten years younger on average than those in the United States, however, survival may be different in that setting. MASS SURGICAL STRATEGY. High-volume surgical facilities Surgery in a single eye has an 85 percent chance of restoring have been in place in India and Pakistan for the past twenty vision, for a probable gain of 2.00 DALYS in the patient's years (Christy and Lail 1973; Liu and others 1977; Wilson expected future lifetime. On the basis of this assumption, 1980; Foster 1987; Venkataswamy 1987). They have demon- bilateral surgery has a 72 percent chance of restoring vision in strated that cataract surgery can be performed inexpensively both eyes, and a 96 percent chance of restoring vision in at and safely on an assembly-line basis. Costs have traditionally least one eye (assuming independence of eyes) for a probable been borne by a combination of government funding, private gain of 3.05 DALYs. I further assumed that a new pair of voluntary organization support, and patient fees. Not surpris- spectacles is required every five years of life. ingly, patients recognize the value of sight restoration and will Table 26-4 is a comparison between the projected cost-util- contribute to their own care within their means. Unfortu- ity ratios of unilateral and bilateral cataract surgery and several nately, when family resources are stretched thin, even the cost other health care interventions reported elsewhere in this of food during hospitalization may be too large a burden for a collection. As can be seen from the table, cataract surgery family in a subsistence economy. compares quite favorably with several interventions that are Fixed surgical facilities may succeed in areas of high popu- generally accepted without question in the public health arena. lation density, but rural areas might best be served by satellite Cataract 643 or mobile facilities. A study in Tanzania reported that only 39 gram in India (Christy and Lall 1973; Venkataswamy and percent of patients scheduled for surgery actually appeared if Brilliant 1981). The provision of free transportation to the the facility was more than twenty-five miles from home. A hospital and food while hospitalized substantially increased the decrease in the distance that patients were required to travel frequency with which patients appeared for cataract surgery meant substantially increased rates of participation (HKI 1985). (Bhatnagar and others 1988). Since blind individuals are es- Thus far, mass surgery has generally relied upon expatriate sentially immobile without assistance in that society, the pro- ophthalmologists for at least a portion of their surgical workers. vision of transportation not only relieves the patient of a This arrangement is necessary if a large volume of surgery is to monetary burden but eliminates the need for another family be performed and the backlog reduced. Because this is clearly member to be away from work during the patient's hospitaliza- not a self-sustaining arrangement, training programs must be tion. Similar findings have been reported from Nepal by Bril- devised to provide local ophthalmic surgeons who will remain liant and Brilliant (1985). in the community. Research Priorities for Ending Cataract Blindness TRAINING OF LOCAL PROVIDERS. An ongoing problem in train- ing local ophthalmologists has been a tendency for those who In 1986 the National Eye Institute co-sponsored a conference go to large urban centers for training to remain in those urban with Helen Keller International in which leading public centers. The training they receive may not be ideally suited to health ophthalmologists and policymakers articulated long- the problems they will encounter if they do return home. In term research priorities for the eradication of cataract blind- recent years, Helen Keller International has mounted a highly ness (HKI 1986). Research goals were divided into three focus successful program to train eye surgeons in the Philippine outer areas: islands. General medical officers are selected at the local level Operations research techniques for improving the efficiency and are trained in situ by visiting faculty. Their training focuses and effectiveness of cataract care in developing countries. The on the problems that face their community. Thus far the aim of specific projects would be the following: dropout rate has been minimal and the cost of training far less * To compare the effectiveness of various methods of than would be required if travel and housing were required for identifying cataract-blind people within a community and, training in urban centers. through the reduction of psychosocial and economic bar- A pilot program in Kenya has demonstrated that nonphysi- throtivate tem to of sy cian ophthalmic clinical officers can be trained to perform riers, motivate them to seek surgery cataract surgery with acceptable results (Whitfield 1987). Al- * To compare alternatives for improving access to cataract though the mechanical aspects of surgery can be leamed by surgery,such aseyecampsortheestablishmentoftemporary individuals of good dexterity, there is no evidence yet that satellite hospitals surgical judgment and the judgment required to manage com- * To evaluate altemative forms of minimum-level oph- plications can be taught without duplicating most of a formal thalmic surgical facilities in regions that are currently residency in ophthalmology. A second problem in this strategy underserved is that most developing countries have physician licensing laws * To improve operating room efficiency that are as strict as those of any Western country. Because of * To determine ways of reducing the postoperative stay the public health menace of semiskilled itinerant cataract following cataract surgery surgeons, countries such as Pakistan have laws that mandate * To increase the number of ophthalmic personnel trained jail sentences for nonphysicians who perform eye surgery. to perform cataract surgery in underserved areas Increasing Participation in Care Epidemiologic research designed to measure the magnitude of the cataract blind population in different regions, study the risk Although providing care at an affordable cost (which may factors for cataract, and evaluate methods of delaying the onset mean free of charge in some areas) is an essential step in the of cataract through controlled clinical trials. Following are eradication of cataract blindness, the simple provision of care specific priorities: is not sufficient to guarantee its acceptance. Only 20 percent of blind individuals in southern India who were offered free * Developing a standard, reproducible system for classify- cataract extraction appeared for surgery within three years ing and documenting the type and severity of cataractous (Venkataswamy and Brilliant 1981). The remaining blind change persons declined because of mistrust, lack of access to the * Identifying the risk factors for aging-related cataracts facility, inability to pay for food while hospitalized, or other in populations that have widely varying prevalence of concems. The result is that community-based facilities may disease remain underused in the face of an overwhelming backlog of * Studying migrant groups who move from high-preva- cataract blindness. Ience areas to low-prevalence areas in order to determine Several researchers have studied factors that affect rates of the relative importance of genetic and environmental fac- participation in cataract surgery at a community-based pro- tors in cataract development 644 Jonathan C. Javitt * Developing noninvasive techniques that can be used to References measure precisely the progression of human cataract * Implementing randomized controlled clinical trials to Al Salem, M., and L. Ismail. 1987. 'Factors Influencing Visual Outcome after test the potential of strategies intended to delay cataract Cataract Extraction among Arabs in Kuwait." British Journal of Ophthalmol- ogy 71:458-61. onset Bhamagar, R., K. P. West, S. Vitale, S. Joshi, G. Venkataswamy, and A. Biological research designed to elucidate the biochemical and Sommer. 1988. "Risk of Cataract and History of Severe Diarrheal Disease physiologic mechanisms of cataractogenesis. The goal of spe- in Southern India." Abstract. Investigative Ophthalmology 29:8. .ph olgc mcai s of toen e e a f se Bochow, T. W., S. K. West, A. Azar, B. Munoz, A. Sommer, and H. R. Taylor. cific projects would be the following: 1989. "Ultraviolet Light Exposure and Risk of Posterior Subcapsular Cata- * To test the hypothesis that oxidative damage to lens racts." Archives of Ophthalmology 107:369-72. proteins is a significant cause of human aging-related Brilliant, G. E., and L. B. Brilliant. 1985. "Using Social Epidemiology to cataract and to develop mechanisms for preventing that Understand Who Stays Blind and Who Gets Operated for Cataract in a catarn Rural Setting." Social Science and Medicine 21:553-58. dcamage Brilliant, G. E., R. P. Pokhrel, N. C. Grasset, and L. B. Brilliant. 1988. "The * To develop a method to grow human lens cells in tissue Epidemiology of Blindness in Nepal: Report of the 1981 Nepal Blindness culture in order to study the normal and abnormal produc- Survey." Seva Foundation, Ann Arbor, Mich. tion of lens proteins Brilliant, L. B., N. C. Grasset, R. P. Pokhrel, A. Kolstad, J. M. Lepkowski. G. * To study the molecular biology underlying the formation E. Brilliant. W. M. Hawks, and R. Pararajeskgaram. 1983. "Associations among Cataract Prevalence, Sunlight Hours, and Altitude in the Himala- of congenital and hereditary cataracts in the hope of shed- yas." AmeTicanJourmal of Epidemiology 118:250-64. ding light on the process of age-related cataractogenesis Brilliant, G. E., J. M. Lepkowski, B. Zwuta, R. D. Thulasiraj. 1991. "Social Determinants of Cataract Surgery Utilization in South India." Archives of Conclusions Ophthalmology 109(4):584-89. CDC (Centers for Disease Control). 1983. Morbidity and Mortality Weekly Report Cataract is the leading cause of blindness and disability in 32:119. developing countries. Current surgical methods of treating Chatterjee, A., R. C. Milton, and S. Thyle. 1982. "Prevalence and Aetiology catarate ating of Cataract in Punjab." BritishJournal of Ophthalmology 66:35-42. catarat aChristy, N. E., and P. Lall. 1973. "Postoperative Endophthalmitis following patient. Initial research suggests that the patient who benefits Cataract Surgery." Archives of Ophthamology 90:361-66. from sight-restoring cataract surgery may generate a 1,500 Davies, L. M., M. F. Drummond, E. G. Woodward, and R. J. Buckley. 1986. percent or greater annual return on the cost of surgery. "A Cost-Effectiveness Comparison of the Intraocular Lens and the Contact The elimination of cataract as the main cause of world Lens in Aphakia." Transactions of the Ophthamologic Society of the UK blindness requires initiatives on multiple levels in order to 105:304413. formulate short-term, intermediate-term, and long-term solu- Dawson, C. R., and 1. R. Schwab. 1981. "Epidemiology of Cataract-A Major t ions. The most pressing need is to begin immediately to reduce Cause of Preventable Blindness." BuUetin of the World Health Organization the backlog of cataract blindness through mass surgery. This 59f493o5rga Drummond, M. F. 1988. "Economic Aspects of Cataract." Ophthalmology requires a commitment of resources along with initiatives in 95:1147-53. operations research designed to reduce barriers to surgery and Drummond, M. R., G. L. Stoddart, and G. W. Torrance. 1987. Methods for the increase the effectiveness of public health programs. Economic Economic Evaluation of Health Care Programmes. Oxford: Oxford University research is similarly required to study net savings to patients, Press. their families, and society of sight restoration through cataract Ederer, F., R. Hiller, and H. R. Taylor. 1981. "Senile Lens Changes and Diabetes surgery. in Two Population Studies." Amenrican Journal of Ophthalmology 91:381-95. Although the strategy of mass surgery using expatriate oph- Ellwein, L. B., J. M. Lepkowski, R. D. Thulasiraj, and 6. E. Brilliant. 1991. thalmologists has the potential to reduce the current backlog, "The Cost Effectiveness of Strategies to Reduce Barriers to Cataract Sur- gery." InternationalOphthalmologN 15(3):175-83. only by training local ophthalmologists and ancillary person- Foster,A.1987. "Cataract Blindness inAfrica."OphthalmicSurgery 18:384-88. nel will it be self-sustaining. Additional epidemiologic HCFA (Health Care Financing Administration). 1986. Summary data on research is needed in order to elucidate better the relationship cataract surgery, drawn for Medicare Part-B claims. Distributed by Ms. between cataract and environmental factors as well as the Michael McMullen, HCFA, Oak Meadows Building, Security Blvd., Balti- effects of reducing known risk factors. more, Md. The currently used technology for cataract extraction is HKI (Helen Keller Intemational). 1985. Kongwva Prinmry Health Care Report adequate for the needs of developing countries, but eventually 1984-85. New York. the instrumentation required for extracapsular extraction and .1986. To Restore Sight: TheGlobalConquestofCataractBlindness. New lens implantation will become economically feasible in these York. countries. The new technologies of surgical extraction com- Hiller, R., L. Giacometti, and K. Yuen. 1977. "Sunlight and Cataract: An Epidemiologic Investigation." AmericanJournal of Epidemiology 105:450-59. bined with the long-term hope of affordable pharmacologic Hiller, R., R. D. Sperduto, and F. Ederer. 1983. "Epidemiologic Associations intervention may one day make the current burden of cataract with Cataract in the 1971-1972 National Health and Nutrition Examina- blindness a dim vision of the past. tion Survey." American Journal of Epidemiology 118:239-49. Cataract 645 Jacques, P. F., L. T. Chylack, Jr., R. B. McGandy, and S. C. Hartz. 1988. Sperdutr, R. D., and R. Hiller. 1984. "The Prevalence of Nuclear, Cortical, "Antioxidant Status in Persons with and without Senile Cataract." Archives and Posterior Subcapsular Lens Opacities in a General Population Sample." of Ophthalmology 106:337-40. Ophthalmology 91:815-18. Jacques, P. F., S. C. Hartz, L. T. Chylack, Jr., R. B. McGandy, and ]. A. Street,D.A.,andJ.C.Javitt. 1992. "NationalOutcomeofCataractExtraction Sadowski. 1988. "Nutritional Status in Persons with and without Senile IV: Increased Mortality following Cataract Extraction in Beneficiaries." Cataract: Blood Vitamin and Mineral Levels." AmencanJournalof Clinical Amencan J]ornal of Ophthalmologs 113:263-68. Nutntion 48:152-58. Taylor, H. R. 1980a. "The Environment and the Lens." Bnnssh Journal of Javitt, J. C., G. Venkataswamy, and A. Sommer. 1983. "The Economic and Ophthalmology 64:303-10. Social Aspect of Restoring Sight." In P. Henkind, ed., ACTA: 24th Interna- . 1980b. "Prevalence and Causes of Blindness in Australian Aborig- uonalCongress of Ophthalmology. New York: ]. B. Lippincott. ines." Medical Journal of Australia 1:71-76. javitt, J. C., S. Vitale, J. K. Canner, H. Krakauer, A. M. McBean, and A. . 1980c. "The Prevalence of Corneal Disease and Cataracts in Austra- Sommer. 1991 a. "National Outcomes of Cataract Extraction 2: Endophthalm- lian Aborigines in Northwestern Australia. " Australian Journal of Ophthal- itis following Inpatient Surgery." Archives of OphialnsoTlogy, 109:1085-89. mology 8:289-301. - 1991 b. "National Outcomes of Cataract Extraction 1: Retinal De- Taylor, H. R., S. K. West, F. S. Rosenthal, B. Munoz, H. S. Newland, H. Abbey, tachment following Inpatient Surgery." Archives oj Ophthalmology. E. A. Emmett. 1988. "Effect of Ultraviolet Radiation on Cataract Forma- Javitt, J. C., 1). A. Street, H. M. Brenner, and others. 1993. "Improvement in tion." New Englandjournal of Medicine 319:1429-33. Visual Function following Cataract Surgery in the First and the Second Torrance C. W., M. H. Boyle, S. P. Horwood. 1982. "Application of Multi- Eye." Archives of Ophthalmology. Attribute Utility Theory to Measure Social Preferences for Health States." Khan, IM. U., M. R. Khan, arid A. K. Sheikh. 1987. "Dehydrating Diarrhoea Operations Research 30:1043-69. and Cataract in Rural Bangladesh." Indian journal of Medical Research Venkataswamy, G. 1987. "Cataract in the Indian Subcontinent." Ophthalmic 85:311-15. Surgery 18:464-66. Liu, H. S., W. J. McGannon, F. 1. Tolentino, and C. L. Schepens. 1977. Venkataswamy, G., and G. E. Brilliant. 1981. "Social and Economic Barriers "MassiveCataract Relief in EyeCamps." AnnaLs of Ophthalmology 1979:503-8. to Cataract Surgery in Rural South India: A Preliminary Report." Visual Minassian, D. C., V. Mehra, and B. R. Jones. 1984. "Dehydrational Crisis Impairment and Blindness 405-508. from Severe Diarrhoea or Heatstroke and Risk of Cataract." Lancet 10: Venkataswamy, G., J. Lepkowski, R. L. Mowery, and the Operations Research 751-53. Group. 1988. "Operations Research to Reduce Barriers to Cataract Surgery Mohan, M., R. D. Sperduto, S. K. Angra, R. C. Milton, R. L. Mathur, B. A. in India." Investgative Ophthalnrology 29:8a. Underwood, N. jaffery, C. B. Pandya, Viki Chhabra, R. B. Vajpayee. 1989. Wen-shi, S. 1979. "A Survey of Senile Cataracts among High Altitude Living "India-U.S. Case-Control Study of Age-Related Cataracts." Archiv.es of Tibetans in Changdu District." ChineseJournal of Ophthalmology 15: 1CK--4. Ophthalmology 107:670-76. Whitfield, R. 1987. "Dealing with Cataract Blindness Part 3: Paramedical Rosenthal, F. S., A. E. Bakalian, and H. R. Taylor. 1986. "The Effect of Cataract Surgery in Africa." Ophthalmic Surgery 18:765-67. Prescription Eyewear on Ocular Exposure to Ultraviolet Radiation." Amer- Whitfield, R., Jr., L. Schwab, N. J. Bakker, G. G. Bisley, and D. Ross Degnan. ican Journal of Public Health 76:1216-20. 1983. "Cataract and Comeal Opacity Are the Main Causes of Blindness in Rosenthal, F. S., C. Phoon, A. E. Bakalian, and H. R. Taylor. 1988. "The the Samburu Tribe of Kenya." Ophthalmic Surgery 14:139-44. Ocular Dose of Ultraviolet Radiation to Outdoor Workers." Investiggatve WHHO (World Health Organization) Programme for the Prevention of Blind- Ophthalmology and Visual Science 29:649-56. ness. 1987. Available data on blindness. 87.14:1-23.34. Geneva. Schwab, L. 1987. "Cost-Effective Cataract Surgery in Developing Nations." Wilson, J. 1980. World Blindness and Its PrevenTon. Oxford: Oxford University Ophthalmic Surgery 18:307-9. Press. 27 Oral Health Douglas Bratthall and David E. Barmes In contrast to most other diseases, some of the oral diseases are cated operations performed by highly skilled personnel, and well known and experienced by most people, albeit perhaps some of which cannot be treated by any methods known at only to a mild degree. Those readers, however, who have had present. To this should be added the fact that considerable toothaches or jaw infections may testify that oral health prob- importance is attached to the oral cavity in many cultures in lems can be so dominant that practically all other problems developing countries; this may be illustrated by traditions such fade into insignificance until help is received-and, if profes- as grinding of healthy teeth to certain shapes or knocking out sional help is not available, removing the aching tooth may be teeth for ceremonial reasons. In industrial countries the in- another never-forgotten experience. creasing number of advertisements for dental materials, and A variety of diseases affect the oral cavity. Dental caries, the courses, in aesthetic dentistry also reflects this importance. disease causing cavities in the teeth, is common worldwide. Because the oral cavity is the means of communication, tast- Untreated caries may lead to infection in the pulp, an infection ing, eating, kissing, and so on, and because it is positioned at that may spread to the supporting tissues and the jaws, with or a level where it is easily observed, it is understandable that without pain to the individual. Other common diseases are the manv people regard oral health as very important. periodontal diseases, including inflammation of the tissues Further information regarding the etiology will be given surrounding the teeth and breakdown of bone support and loss below for some of the most prevalent oral diseases. This should of teeth. not be interpreted as an underestimation of the less prevalent Further problems affecting the teeth involve the position of diseases. Certainly, such diseases may be most inconvenient or the teeth, varying from simple conditions like too much space even fatal for those affected. But, from a global point of view, between them or overcrowding to serious lip and cleft palate any changes in the prevalence of the common oral diseases will syndromes resulting in chaos for the formation of normal be of such significance, that we, at this stage, can be excused dentition. Traffic accidents, violence, and certain sports and for making such a restriction. games often involve injuries to the teeth. Disturbances in the formation of normal tooth structures may be caused by inher- Dental Caries ited diseases or, as in the case of fluorosis, by the intake of too much fluoride through drinking water or food. A number of Dental caries is characterized by the dissolution of the hard substances may give disturbing discolorations to the teeth, tissues of the teeth (enamel, dentin), eventually leading to the either when the substances are supplied during the time of destruction of the affected tooth surface, or of the tooth itself. formation of the teeth or when they are added to already The immediate cause is the organic acids produced by certain erupted teeth. microorganisms present on the tooth. The bacteria, together The soft tissues in the oral cavity may be the site of numerous with a matrix made up mainly of extracellular polysaccharides conditions, involving oral cancers, symptoms of infection by produced from sucrose by the microorganisms, form the so- the human immunodeficiency virus (HIV), or less harmful but called dental plaque. The acids are formed when fermentable painful conditions. Disturbances in the normal saliva flow are carbohydrates are added to the plaque. Each time such a not uncommon and are most prevalent in elderly people; such process is started, the tooth will be damaged, but if the process conditions are usually very uncomfortable for the person af- does not occur too often, the natural capacity of the body to fected and may predispose him or her to further dental prob- remineralize the tooth will prevent the formation of a cavity. lems. Unnecessary or poor quality dental care can also be the From this simple description, some factors that influence the cause of oral hcalth problcms. risk of caries disease, and cavities, can be identified: From this short introduction, it is clear that the oral cavity' a The tooth surface may be more or less covered by dental is a center for a large variety of possible diseases, some of which plaque. More plaque, especially if it contains cariogenic can be prevented, some of which can be cured only by compli- 647 648 Douglas Bratthall and David E. Ba-nnes microorganisms, includes more bacteria and may result in not use fluorides, and so on. The answers to such questions may the formation of more organic acids. All methods aimed at sometimes be found through research within community den- reducing the amount of plaque, such as toothbrushing and tistry, but often they lie outside the purely odontological field. use of antiplaque substances, thus are an attempt to reduce the amount of acids to be produced. Periodontal Diseases * Dental plaque is composed of a variety of oral microor- ganisms. Some of these microorganisms have a higher car- Bacteria are the main cause also of the diseases affecting the iogenic potential than others. This potential includes factors supporting tissues of the teeth, and the host's response to the such as the ability to form acids, to form acids at low pH, to bacteria may result in more or less severe damage. Accumula- survive at low pH, to adhere to the tooth, and to form tion of bacteria on the teeth close to the tissues usually results extracellular polysaccharides. Among the many microorga- in gingivitis, characterized by a tendency of the gums to bleed, nisms identified, the so-called mutans streptococci (Strepto- especially when light pressure is applied, as, for example, at coccus murans, Streptococcus sobrinus), in particular, have toothbrushing. More severe forms involve breakdown of the been assigned an important role in the development of caries. bone support of the teeth, resulting in more or less mobile Some means are now available for combating the mutans teeth, later perhaps in the total loss of the affected teeth. streptococci, although they are not yet in common use. The processes leading to the more severe forms are the result * Composition of diet, as well as frequency of eating, are of presence of bacteria in the gingival pockets and the reaction further important factors. A diet with a low sucrose content, of the bacteria to the host defense systems. Research during or less frequent eating, will restilt in reduced formation of the last fifteen years has pointed out some bacterial species as organic acids. This knowledge has resulted in well-known being particularly associated with periodontal diseases: advice regarding the restriction of intake of products that Actinobacillus actinomycetemcomitans, Bacteroides gingivalis, contain sugar and also has led to the development of less Bacteroides intermedius, Peptostreptococcus micros, Veillonella cariogenic products, which contain other sweeteners such recta, to name some. But bacteria may be present also in as xylitol, sorbitol, or aspartame. gingival pockets where no periodontal disease appears to - Effective remineralization is another factor that has re- follow, illustrating the complex interaction with the host. ceived increased attention. This factor is dependet on ,Periodontal diseases are not considered diseases which uncon- ceived increased attention. This factor is dependent on ... saliva flow and saliva composition, but the presence of ditionallv follow gingivitis, although they may. Individuals at fluoride during the active process, in the plaque fluid, has extra risk for periodontal diseases mav be persons with im- been shown to be of utmost importance. Earlier methods of munodeficiencies, malnutrition, or diabetes, and those who employing fluorides concentrated mainly on trying to "build smoke. A thorough discussion about the possibilities of iden- in" the fluorides to make the teeth more acid resista nt,b tifying individuals at risk for periodontal disease was presented systemic addition of fluoride during the calcification of the PyJonson of p n s us. teeth. Although this strategy still has some bearing, the Preventlon of periodontal diseases usually focuses on the continuotus supply of fluorides even in low concentrations dental plaques-the effective removal of the bacterial depos- now attract most interest. The potential offluoride to reduce its, including calculuLs, on the teeth. Treatments of advanced caries has resulted in numerous attempts to use the sub- stages usually include surgical methods to get access to the , , s ..... ,, .. .. , ~~affected parts. Antibiotic therapy is sometimes introduced but stance, such as by adding fluoride to drinking water, salt, milk, tablets, fluoride rinses, vamishes, gels, and, of course, can only be looked upon as a support for the local treatment. toothpaste. * A number of individual factors may increase, or decrease, Other Oral Diseases the risk of caries. These factors operate mainly through the Because dental caries and periodontal disease are or have been saliva. Extremely low secretion rates, which, for example, so common, less attention has usually been focused on other may be an unwanted side effect of certain drugs, often result ' in hgh aris scres Aniboies n te oal fuid ar of oral health problems. It would be a great mistake, however, to in high caries scores. Antibodies in the oral fluids are of neglect these diseases and conditions, because they may, often particular interest because thev would be the main mecha- ngetteedsae n odtos eas hym fe pasmforticular interest bccase. they would be themainecha- result in severe consequences to the person affected. In a survey of the epidemiology of oral diseases other than caries and The factors mentioned above: plaque, specific bacteria, diet, periodontal disease, roughly one-quarter to one-half of popu- fluoride, and saliva are all involved in the caries process. It is lations examined were affected by conditions like masticatory important to understand what happens exactly on the tooth dysfunction, traumatic dental and maxillofacial injuries, im- where a caries lesion will, or will not, occur. Once the factors pactions, and oral mucosal disease (Andreasen and others have been identified, further questions will be raised: why do 1986). Some examples of this survey will be mentioned below. we have this particular combination of factors on this tooth, Regarding dysfunction of the masticatory apparatus, or in this person, or in this population? We then arrive at points Andreasen and others reviewed five studies, all dealing with for discussion of why a certain groups of people have a certain populations in industrial countries. About 25 to 50 percent type of diet, why they do not clean their teeth, why they do of the people had subjective symptoms. The authors of the Oral Health 649 studies pointed out that headache often follows mandibular children and adolescents age three to nineteen. The frequency dysfunction and that headache is a common cause of visits to of dental injuries varied from 8 to 35 percent. The authors of physicians. a Danish prospective study of children eligible for the pre- For traumatic dental injuries, Andreasen and others reviewed school dental service pointed out that every third child had fourteen studies that summarized investigations concerning experienced trauma of the primary dentition and every fifth child had sustained injury to the permanent dentition before Table 27-1. Prevalence of Oral Soft Tissue Lesions leaving school at age sixteen (Andreasen and Ravn 1972). Chiang Mai, Kuala Lumpur, and Sweden Data from developing countries were very sparse. ----- Few reports on mnaxilofacial injuries were available but those Prevalence (percent) reporting data from Scandinavia and England estimated the Chiang Kuala annual incidence as being I to 4 per 10,000 people, traffic Lesion Mai Lumpur Sweden accidents and assault being the main cause. A remarkable Infections higher incidence was reported from Greenland, 19 cases per Herpes labialis 0.9 0 3.1 10,000, usually caused by assault following alcohol abuse. History of herpes labialis 5.6 2.6 14.3 Tooth impactions and other eruption disturbances were fre- Intraoral herpetiform lesion 0.9 1.3 0.3 Pseudomembranous candidiasis 0 0.4 0.2 quent findings in seven studies. About 20 to 30 percent of the Angulardcheiliris 0.9 0.9 3.8 populations investigated were affected. The conditions may lead to resorption of teeth, cysts, tumors, and inflammation, in Ulcers particular, pericoronitis. Recurrent aphthae 11.1 5.1 2.0 Very few studies have presented comprehensive data on the History of recurrent aphthae 37.2 21.9 15.7 prevalence of the full range of mucosal lesions. One such study Traumatic ulcer 13.2 12.4 4.3 was performed in Sweden (Ax6ll1976), and recently data have Whitish lesions became available for most of those lesions from two more areas, Leukoplakia 1.3 1.7 3.6 Chiang Mai in Thailand and Kuala Lumpur in Malaysia PreleLikoplakia 1.7 1.7 6.4 (Axell, Bte Zain, and Siwamogstham 1990), although based Smoker'spalare 3.4 3.4 1.1 on a much smaller sample and more selected material. The Betel chewer's mucosa 0.4 1.3 0 findings are summarized in table 27-1. It should be understood Frictional lesion 3.8 5.2 5.5 that several of the conditions may be painful or even precan- Cheek and lip biting 1.7 5.6 5.1 cerous. For example, leukoplakia has been recognized as a Lichen planus 3.8 2.1 1.9 frequent precursor to oral cancer, and researchers in one study with an average observation period of 7.5 years showed that 17 Denture-related lesions percent of leukoplakias became malignant. Denture sore mouth 3.4 7.7 16.0 A chapter in this collection is devoted to cancer, but it Flabby ridge 0 0.4 8.6 should be mentioned here that the prevalence of oral cancers Denture hyperplasia 0.9 0 3.4 differs widely between different areas. In a survey of the liter- Tongue lesionm ature, the reported incidence rates of oral cancer, including Median rhomboid glossitis 1.3 1.3 1.4 vermilion border of the lip, varied from 5 to 25 cases per Geographic tongue 5.1 6.4 8.5 1 00,000 population in industrial countries and from 2 to 17 in Plicated tongue 3.4 5.2 6.5 developing countries (Andreasen and others 1986). Smok- Hairy tongue 0 0.9 0.6 ing tobacco and drinking alcohol are the major etiologic Atrophy of tongue papillae 3.0 1.3 1.1 factors. Also, it should be observed that not only the stronger types of alcoholic drinks, like whisky and vodka, are Pigmentation associated with the disease but also wines and other less alco- Melanin pigmentation 0.5 88.4 9.9 holic drinks. Chewing tobacco with or without areca (betel) Amalgam tattoo 0.9 0.4 8.2 ocdrnsChwntoacwihrwtouaea(bel is carcinogenic. Tumors and tumorlike lesions Special oral health problems are associated with HIV infec- Carcinoma 0 0.4 < 0.1 tion. Pindborg (1989) proposed a classification for lesions that Papilloma 0.9 0 0.1 included a variety of fungal infections, bacterial and viral Hemangioma 2.1 1.7 0.1 infections, neoplasms, neurological disturbances, and lesions Lipoma 0 0.4 0.1.. .. Fibroepithelial polyp 1.7 3.9 3.3 of unknown cause. The tabulation is preliminary and revisions Pyogenictgranuloma 1.3 0.9 0.1 are foreseen. It is clear, however, that the oral cavity often Mucocele 0.4 0.4 - displays symptoms of H-v infections, and some of the diseases - - Not available. that occur are very serious. - Not available.Maocuininld foexmlcrwigrspig Source Table based on study of 234 people in Chiang Mai, 233 in Kuala Malocc'usions include, for example, crowding or spacing LumpUr (Axelt, Bte Zain, and Siwarrogstham 1990), and 20,333 subjects in problems, overjet, deep or open bite, crossbite, and scissors Sweden (Axeil 1976). bite. Several problems can result from malocclusion, such as 650 Douglas Bratthall and David E. Barmes Table 27-2. Changes in Caries Prevalence compounded by the possibility of requiring specialist treat- in Twelve-Year-Olds, by Country ment, the cost of which is, of course, also an important factor. Country Year DMFT a Year DMFTa ---ny ~ z DFTa ea ~F Disease Prevalence: Current Levels and Trends Asia Bangladesh 1979 1.8 1990 3.5 Oral diseases can be measured by various indexes. The indexes China 1951 0.6 1985 0.7 may focus on be meas e by cross-secinal French Polynesia 1977 10.5 1987 2.5-3.8 may focus on the prevalence of the disease by cross-sectional Indonesia 1973 0.7 1982 2.3 studies, or the incidence, by longitudinal studies. For dental Myanmar 1977 0.8 1990 1.1 caries, a cross-sectional study will reveal lifetime caries, or the Philippines 1977 2.5 1982 5.5 amount of caries since the first permanent tooth appeared in Singapore 1970 2.9 1984 2.5 the mouth. Missing teeth pose special difficulties; a tooth may Thailand 1977 2.7 1989 1.5 be missing because advanced caries required its extraction, but Tonga 1966 0.7 1986 1.0 it may also be missing for other reasons, such as periodontitis, Industrial countnes orthodontic reasons, or aesthetic reasons. It will be appreciated Belgium 1972 3.1 1988 3.1 that age is an important factor to be taken into account when Canada 1977 6.0 1987 4.3 evaluating caries data-the older the age, the greater the risk Fapan 1975 5. 9 1987 419 of more damaged teeth. Thus, the caries index always shows New Zealand 1973 6.0 1989 214 higher individual values with age. When caries levels are Sweden 1937 7.8 1989 2.2 compared for different populations, the same age groups must United Kingdom 1973 4.7 1983 3.1 therefore be chosen. United States 1965/67 4.0 1986/87 1.8 Information regarding the prevalence of dental caries is U.S.S.R. 1972 3.5 1986 3.0 overwhelming. It stretches from ancient times, as recorded by Latin Amefica and th, Canbhean archaeological investigations of skulls, to the present, when, Argentina 1965 4.5 1987 3.4 in some areas, all children are subjected to annual checkups, Brazil 1976 8.6 1988 6.7 including x-rays, and data are fed continuously into computers Cuba 1973 5.1 1984 3.9 so that even minor changes can be seen. To obtain comparable Mexico 1972 2.7 1984 3.2 data, it is important that the same recording methods be used, Middle East and North Africa and in this field, the Oral Health Unit at the World Health Algeria 1974 1.9 1987 2.3 Organization (WHO) has prepared guidelines for oral health Israel 1966 2.4 1989 3.0 surveys, recommending indicator ages; recording forms are also Jordan 1962 0.2 1991 1.7 produced. Since the early 1970s, a powerful instrument to Syrian Arab Republic 1974 4.4 1989 1.7 monitor changes in oral health trends has been set up at WHO in Geneva: the Global Oral Data Bank. At present it contains Sub-SahaTan .Af0ica files on more than 1,000 surveys, dating from 1937, with data Mala fri 1978 0.8 1991 0.7 on caries for 148 countries and on periodontal diseases for 103 Sudan 1979 1.1 1984 2.1 countries. Each year global maps regarding caries and peri- Tanzania 1973 0.6 1989 1.0 odontal diseases are produced illustrating the latest informa- Togo 1973 1.6 1986 0.3 tion using comparable data. In table 27-2 we show examples Zaire 1970 1.0 1985 1.0 of caries data from some countries. Zambia 1971 0.1 1982 2.3 Various indexes are also used to estimate the degree of a. wMFr: Number of decayed. niissing, and filled teeth. gingival inflammation and periodontal disease. The WHO stud- Solurce: WHO Global Oral Data Bank. ies use the Community Periodontal Index ofTreatment Needs (cpiTN; see Barmes and Leous 1986; Pilot and others 1986). The index is based on three indicators: (a) presence or absence difficulties in jaw movements and temporomandibular joint ofgingival bleeding; (b) supra- or subgingival calculus; and (c) disturbances. Speech and swallowing might be affected, and periodontal pockets, subdivided into shallow (4 to 5 millime- psychosocial problems may occur if aesthetic problems are ters) and deep pockets (6 millimeters and more). The mouth apparent. In their study from the United States, Kelly and is divided into sextants and certain index teeth are registered Harvey (1977) indicate that the majority of American chil- for any of the indicators. The sextant will obtain a score from dren and adolescents have a malocclusion of some type. Actu- 0 (healthy) to 4 (> 6 millimeter pocket), and the highest score ally, they showed that 75 percent of the youths, age twelve to found on the index teeth is chosen for the sextant. The results seventeen years, had a deviation from the ideal situation and for some countries are shown in figure 27-1. Johnson (1989) about 25 to 30 percent had a severe malocclusion. For these calculated that approximately 5 to 20 percent of most popula- conditions, the treatment needs and demands vary among tions that have been adequately surveyed had destructive countries. The problem itself is an important factor, and it is periodontitis of a "clinically significant" degree. Oral Health 651 Figure 27-1. Observed Periodontal Conditions Measured by cpiTN at Age 35-44, Selected Countries Australia Bangladesh Central African Republic Italy Japan I I I Morocco __ - I Tanzania I Thailand I United Kingdom I Zimbabwe I I 0 10 20 30 40 50 60 70 80 90 100 Persons (percent) |ZIZ|I No periodontal disease I Calculus Deep pockets Lilill |Bleeding only Shallow pockets Source: WHO Global Oral Data Bank. The situation for dental caries has changed during the last treatment. A growing demand for preventive measures became decades of this century, as shown in table 27-2. A tremendous apparent and a number of methods were devised, resulting in decrease in the prevalence of caries has occurred in several a sudden decrease in caries and periodontitis. The demand for industrial countries, at least in younger age groups. Some services is now on the decrease in highly industrialized coun- developing countries, however, show the reverse trend. Some tries. Dental schools face problems because of the reduced need explanations for these facts are listed in table 27-3. for dentists, some being forced to close and others to reduce It is clear that a number of factors may have been instrumen- the intake of students. tal in the decrease in caries in industrial countries. It is not Meanwhile, developing countries are tuming to more West- possible, as yet, to grade the various factors, because the ern food. Very often these countries have no, or very limited, importance may vary from country to country and the import- resources for oral health treatment. Interest in prevention is ance of an individual factor is dependent on the original level understandably low, because oral health problems have, until of caries. recently, been more or less nonexistent. Caries is increasing, The information available as a whole may illustrate the particularly in the cities, resulting in a growing demand for situation. In historical times, the problem of caries was minor, better dental services. worldwide. Exceptions can be found in certain populations, and some groups of people showed the presence of root surface Patterns for the Years 2000 and 2015 caries. An increase was noted in restricted groups of popula- tions when refined sugars were introduced. At the beginning The World Health Organization has defined a certain number it was mostly people from higher social classes who could afford of goals for oral health for the year 2000. these products who were affected. In many Western coun- * Fifty percent of five- to six-year-olds will be caries free. tries, a "caries explosion" started during the period between * The global average will be no more than three decayed, 1850 and 1900, when industrialization and new dietary habits .imissing, or filled teeth at twelve years of age. arrived. - Eighty-five percent of the population should retain all Demand appeared for oral health services, for professional t eighty-ie ercent extractions, and for the repair of damaged teeth. Dental theirteeth at ageeighteen. schools were built, new instruments were developed, and new * A 50 percent reduction in present levels of toothlessness filling materials were progressively devised. A golden age for at age thirty-five through forty-four will be achieved. dentists began, brought about by a never-ending demand for * A twenty-five percent reduction in present levels of oral health services. Eventually, it was realized that fillings, toothlessness at age sixty-five will be achieved. crowns, bridges, and the like did not last very long and that at * A data-based system for monitoring changes in oral least one-third of dental treatment involved replacing previous health will be established. 652 Douglas Bratthall and David E. Barmes Table 27-3. Current Trends in Caries Prevalence Trend Intervention Comments Decrease in industrial countries Fluoride in toothpastes Fluoride has been added since the 1950s; 80-95 percent of toothpastes contain fluoride. It is considered a major reason for the decrease. Water fluoridation Fluoride content of drinking waters may be adjusted to optimal levels. Positive effects may be obtained even with local applications. Salt or milk fluoridation Used in some countries. Fluoride tablets, topical Used in various school-based programs, as well as individually. application, rinsing, varnishes Oral hygiene information, instruction, and supervised oral hygiene programs may have resulted in improved hygiene. Dietary advice Mainly focused on reduction of sucrose. Total sugar consumption has not dramatically changed. Use of sugar substitutes has increased. Oral microflora changes Temporary suppression through antibiotics. Possible effects on acid production through fluorides and other antimicrobial substances. Increase in developing countries Changes in diet Westernization of diet Source: Compiled by present authors. For caries, one important goal is that the mean number of after the introduction of preventive programs. These programs decayed, missing, or filled teeth should not exceed three at the have been worked out in collaboration with advanced research age of twelve. Current trends seem to indicate that many institutes in industrial countries. Knowledge has thereby been Western countries have a good chance of reaching this goal transferred and resources have been allocated to help intro- (Pilot 1988). Other countries, such as Poland, have expressed duce the programs. doubts about the possibilities of reducing caries to that extent (Ja'nczuk 1989). Many developing countries have never ex- THE PESSIMISTIC SCENARIO. To begin with, the decrease in ceeded that particular goal, but the crucial question remains: prevalence of the diseases continues for some time. This is will the increase observed in some countries be halted or not? interpreted by the decisionmakers as the elimination of dental The year 2000 is close, and for 2015, no goals have yet been diseases. The interest of the individual, the profession, and the set. Let us therefore present two "scenarios." politicians in supporting preventive efforts decreases. Preven- tive programs in schools are reduced; some cities stop adding THE OPTIMISTIC SCENARIO. The ongoing trend toward a de- fluoride to drinking water. Development of new preventive crease in oral diseases in Western countries continues. New methods are successful, but the research and development costs effective prophylactic agents are introduced, special programs are high, resulting in high prices for the products; therefore, for risk groups are designed, and oral diseases in the elderly are they are used only by a few. kept under control. The profession adjusts itself to the new The elderly population is increasing. They will have more situation. By agreement, many tasks are transferred from den- teeth than before. Root surface caries cannot be kept under tists to personnel with less costly education. The number of control. There is a shortage of personnel at homes for the dentists decreases, but the profession aims at, and succeeds in, elderly, and the care givers that remain show little interest in distributing the resources according to geographical needs. taking care of the teeth of their patients, particularly because Although the number of dental schools is reduced, the re- the task has become increasingly difficult, owing to the fact sources for dental research are not. Politicians are also aware that the "extra" teeth are situated more posterior than was the of the fact that oral diseases are not eliminated, just kept in case earlier. Many extractions follow, but the elderly demand check. Therefore, resources for continuous preventive meth- fixed teeth, and therefore treatments involving extremely ods remain intact. expensive implants and bridges increase rapidly. Because the In developing countries, the authorities realize that they number of dental schools has been reduced, the number of staff have to choose one of two ways-either the expensive one (that is, the dental researchers) has also been reduced. At the used earlier by the Western countries or the immediate intro- beginning, endless discussions between the dentists, dental duction of effective preventive programs. They choose the hygienists, nurses, and other supporting groups are held, finally latter one and are successful in educating the people in the leading to the withdrawal of the possibility for these groups to value of prevention before the diseases become prevalent. perform clinical work. All of a sudden, when it is realized that Traditional foods are retained as far as possible, and if between- there are too few dentists, it becomes a problem to cover the meal snacks containing sucrose are introduced, they will come less popular areas of the countries with professionals. Oral Health 653 The consumption of Westem type of food, including junk cover training personnel, running dental schools, dental sup- food, is rapidly increasing in developing countries. Starting in ply depots, dental research and dental journals, patient time, the cities, caries is increasing and the demand for restoration, traveling costs, and so on, plus administration of the den- not for prevention, is growing. Dental schools and dental tal programs within the government health divisions and de- clinics are being established to meet this demand. Preventive partments. In different countries, the proportion of the total programs are launched, but because of the lack of resources, costs varies according to how much is finally charged to the interest, and knowledge, the programs reach only a small patient and how much is paid through taxes. For this reason, proportion of the population. Also, the profession favors the comparisons between fees charged for a filling must be assessed reconstructive approach because it gives higher income and with caution. In addition, trying to get the values in compara- status. Dentists are gathering in and around the big cities, ble monetary values is almost impossible, but here are some leaving extractions in the underprivileged population to be examples. performed by less-qualified personnel. Advertisements for The fee for one occlusal surface filling in a molar of an adult Western food and other products are flourishing, including patient would, in the United Kingdom, be about 5 (corre- cigarettes. In many developing countries, smoking is on the sponding to about us$8), of which the patient would pay increase, resulting in more and more cases of oral cancer. directly 75 percent, if the dentist is working within the Na- In both developing and industrial countries, HIV infections tional Health Service. In Sudan, the patient would have to pay are spreading rapidly, and patients with oral manifestations are about 12 Sudanese pounds, or about us$1 to $2, and in a frequently seen in clinics. Resources have to be set aside for Brazilian city, us$25 could be charged. The figures are easily these patients and, because special, often time-consuming misleading, however, and a WHO/HDI (World Health Organiza- arrangements for infection control have to be made, these tion/Federation Dentaire Internationale) joint working group, patients take more and more time to treat. Due to general JWG 9, proposed instead a relative value system, in which environmental problems such as pollution, climatic change, different treatments were related to a particular item, giving drug abuse, and even war, smaller and smaller proportions of the value 100. The item chosen was a one-surface amalgam national budgets are being set aside for health purposes, in filling, in a first molar, including anesthesia and lining. particular, for oral health. Data from the following countries are available: Austria, The coming decades will be extremely important, because Denmark, Finland, France, Hong Kong, Japan, the Nether- the decisions to be taken will influence oral health for many lands, and Sweden. The fee index indicates that the price for years to come. At the present time, it is possible to have some upper and lower dentures would be about twenty to ninety hope for a positive outcome. The pessimistic scenario, how- times more expensive than the filling. One porcelain or metal ever, was very easy to imagine. Some questions to consider crown was estimated to cost seven to forty times more than the when trying to decide which scenario will come out on top are, filling, and a simple extraction was estimated to cost from 33 for example, is it usual that decisions are made which favor the to 148 percent of the fee for the filling. health of the total population' Is present knowledge about Here is another way to illustrate the cost of treatment. In a prevention being used properly? How many years does it take provincial hospital in a district in Zimbabwe, 48 kilometers to introduce new ideas to an entire population? Will environ- from the capital, Harare, a simple filling would cost Z$5.00, mental and economic crises be solved? What are the resources equivalent at that moment to about us$2.50. For the same for introduction of the negative factors in relation to the amount of money, 1 to 2 kilograms of meat could be obtained, health-supporting activities? or two tubes of toothpaste, or 7 kilograms of brown sugar. Or, All the facts mentioned in the two scenarios will probably the patient could have traveled by bus to Harare, enjoying 100 happen, somewhere in the world. We will have various com- pieces of candy during the trip. At a private clinic in the binations of the pessimistic and optimistic scenarios. Some capital, the patient could be charged four to five times more countries will be winners, some losers. The regular monitoring for the filling. reports from the WHO Oral Data Bank will reflect the net Total costs for dentistry may be very low at present in some outcome worldwide. developing countries, where, in some cases, the total budget for health is just a few dollars per person per year. In industrial Economic Costs countries, dental costs may often reach about 10 percent of the health budget. This is the case in Switzerland, for example (6.5 A patient has a cavity that needs restoration. How much does million inhabitants), where total costs have been calculated at it cost to get it filled?The directs costs involve time for drilling, about 1.7 billion Swiss francs (SFr), corresponding to about filling, and polishing, plus the materials. Further costs include SFr 400,000 per dental surgery (Meier 1988).1 Ninety-four those of office space, dental equipment, salaries for personnel, percent of these costs are paid by the patients, the remainder office supplies, telephone, and so on. If the tooth had been by the state. In Sweden (about 8.5 million inhabitants), the more seriously damaged, perhaps a crown would have been total cost for dental care was 6,505 million Swedish krona necessary, which would have meant additional costs for im- (Skr) in 1985, of which 44 percent was covered by the state, pression materials, labor of technicians, gold, x-rays, transpor- 25 percent by county councils, and 31 percent by patients. tation between dentist and technician, and so on. Other costs These are direct costs only and do not cover such expenses as 654 Douglas Brat"hol and David E. Barmnes education of dental personnel or loss of working hours for the * Level I . Prevention of onset of disease (prepathogenic), patients. Current direct dental costs per inhabitant of Sweden including integration of general and oral health; informa- would amount to approximately us$125 per year. tion, advice, control; diet and nutrition; general and oral For the purpose of comparison, it is interesting to study the hygiene; fluorides. costs for an advanced form of therapy, implants. This treat- * Level 2. Prevention of further development of disease, ment is used for patients who do not have teeth in one or both including simple fillings; fissure sealing for high-risk pa- jaws. In 1986, a thorough cost evaluation (Karlsson 1986) of tients; professional hygiene. such care, the Branemark method, was made for ten patients who had received treatment in specialist clinics in a county of * Level 3. Relief of pain and debilitating consequences of Sweden. The treatment involves the following procedures: disease, including emergency treatment, extractions, and after detailed examination, titanium fixtures are implanted in medication; referral. the patient's jaws by a surgical operation. After a healing period of three to six months, "distances" are applied, and after Cost and Effectiveness a further two weeks, bridges can be inserted, fixed to the distances. This procedute is followed by several checkups. An example from Switzerland may illustrate cost and effective- Thus, the result is "fixed" teeth for a patient, for whom the ness in an industrial country. The estimations have been made previous altemative was removable dentures. available by Dr. M. Bittner (personal communication, 1989). In the study referred to, Karlsson (1986) calculated direct In the canton of Basel, 30.4 tooth surfaces per fifteen-year-old costs (for treatment time for dental personnel and for equip- child had to be restored in 1961. The cost for these treatments ment) and indirect costs (resources lost because of missing was estimated at SFr 2.8 million per 1,500 children. In 1988, teeth, such as work problems, time, and travel for the patient). only 3.7 surfaces per child had to be restored, at a cost of SFr Furthermore, the author discussed calculation of costs of a 0.3 million, thus a difference of SFr 2.5 million. more undefined character, such as the monetary value of the In 1962 a water fluoridation program was implemented in problem of having no teeth. Because of the great difficulties of Basel, and in 1970 a school-based preventive dentistry program obtaining such values, these costs were not included in the was introduced. During the past two decades, fluoridated results. The results showed that the total cost for one jaw was toothpaste has become more and more prevalent in the mar- about Skr 40,000 to 60,000 (about us$6,150 to $9,230). This ketplace, and approximately 90 percent of the population now cost is about seven times higher than for removable dentures, uses such toothpaste. Seventy percent of the population uses under similar conditions. fluoridated salt. Certainly, these calculations are relevant only for this par- The school-based progtam, which actually starts when the ticular group of patients, and similar treatment in other coun- children are three years of age and follows the children tries would give other results because of varying costs for throughout secondary school, includes topical fluoride appli- dentists' time, transportation, and so on. Still, the calculations cations and oral hygiene instructions as well as dietary advice. illustrate the enormous costs that could follow if no preventive Most tasks are perftormed by auxiliaries. The cost for this programs that protect against loss of teeth are implemented program is estimated at SFr 16 per child per year. Thus, for and if the population demands fixed teeth. 1,500 children who participate in the program for ten years, the cost would be SFr 240,000. The cost for the fluoride used Elements of Preventive Strategy for water fluoridation and maintenance in Basel is now esti- mated to only SFr 0.5 per person per year. The basic elements of a preventive strategy consist of identifi- Some further data on the effectiveness of the program can cation of, and action against, the etiological factors for each be mentioned: of the fifteen-year-old children leaving school oral disease. In table 27-4 we illustrate these elements for some in 1967, 0 percent had no caries. In 1988, 34 percent had no diseases. As can be seen, information and motivation are caries. For the same age group, fifteen teeth wete affected, as a important preventive methods for several conditions. The use mean, in 1961, whereas only three teeth were affected in 1988. of these methods, however, requires that the patient (popula- It is not possible to estimate exactly how much of the caries tion) follow certain rules. One definite advantage of a method reductions are due to the various components in the program. such as water fluoridation is that people will benefit even if In countries without fluoridated water or salt, substantial re- they are unable to follow advice because of sickness, failing ductions in caries have also been observed, if fluoridated interest, lack of education, and the like. toothpaste and school programs are used. The elements of prevention must be presented within a system, and the industrial countries have developed more or The Future of Prevention Technology less effective frameworks. In developing countries, this work is still in the beginning stages, and includes a strategy with Can we expect new methods that will change, dramatically, primary, secondary, and tertiary prevention. One concept for the basis for preventive treatment of the oral diseases? Dental primary oral health care was recently described by Jeboda and research tries to find new ways, in several directions. Intensive Eriksen (1989). They suggested these levels: studies focus, for example, on plaque-inhibiting substances. Oral Health 655 Table 27-4. Preventive Strategies for Oral Health Condition Cause Action needed Methods Dental caries Increased sugar intake Reduce sugars Information, recommendations; increase price of sugars, taxes; sugar substitutes Plaque present Reduce plaque Information and instruction; inore effective tooth cleaning; flossing, toothpicks; professional tooth cleaning; less sugars in diet; fissure sealants; antimicrobial products High mnutans streptococci Reduce mutans streptococci More effective tooth cleaning; less sugars levels in diet; antinmicrobial vamish; chlorhexidine gels and rinses; vaccination Insufficient fluoride in Increase fluorides Water fluoridation; daily use of fluoride relation to etiological toothpastes; fluoride-containing drinks and factors foodstuffs; vamishes, rinsing programs; fluoride tablets, gels Reduced saliva secretion Compensate Find out reasons and take appropriate actions Periodontal disease Plaque present Reduce plaque Information and instruction; more effective tooth cleaning; flossing, toothpicks; antimicrobial products; professional tooth cleaning; scaling Specific pathogenic Reduce bacteria Antibiotics bacteria Oral cancer and precancer Increased risk factors, Reduce risk factors Information, motivation including tobacco, alcohol AIDS-related symptoms Risk behavior Treatment and prevention Information, motivation Lip and palate clefts Present from birth Surgical treatment Care in dental hospital Fluorosis High fluoride intake Reduce intake Defluoridation of drinking waters; identify other sources; information Orthodontic problems Genetic Treatment Care by specialized personnel for advanced cases Other dental diseases, Prevention Care by specialized personnel for advanced caries, or periodontal cases disease Temporomandibular joint problems Stress Treatment and prevention Individual treatment, information, clinical care Other dental diseases, Prevention Individual treatment, information, clinical caries, or peri(odontal care disease Traumatic injuries Sports Treatment and supervision Individual treatment, mouth guards Source: Authors' compilation. Such products could have an effect on the tooth surface, the Whereas plaque-inhibiting substances act on most bacterial saliva pellicle covering the teeth, or the oral bacteria. Several types forming the dental plaque, other substances try to find products seem to be promising. The products would be admin- means of attacking the pathogenic bacteria. Most stidies con- istered through mouth rinses, topical application, or tablets. cern the mutans streptococci. Very promising results have Prices, number of applications, and long-term effects are as yet been reported by Sandham and others (1988). They have unknown, and it can he assumed that these products will first found that a vamish, containing chlorhexidine, can eliminate be used in industrial countries. the mutans streptococci in many patients. The use of this 656 Douglas Brattuhall arLd David E. Barmes varnish could be combined with methods which identify per- habits and products that increase the risk of disease, such as sons carrying large amounts of these bacteria. The system frequent intake of products that contain sugar. Of course, if seems to be ready for large-scale clinical studies. It is important, sugar-rich products are necessary for survival, that would however, to evaluate the effects on caries, not only on the take precedence. bacteria. * If pain is present, possibilities for pain relief must exist. Vaccination against dental caries has been discussed for * If there is any risk of caries occurring, or if caries is already decades, and vaccines that have good protective effects in present, a fluoride program should be implemented, and if animals have been produced. Hesitation to test these products toothpaste is used, it should contain fluoride. If toothpaste on humans stems from concerns for safety. Because caries is not is too expensive, school-based fluoride programs should be a deadly disease, not one single case of fatal side effect should started. For cities, water fluoridation can be considered, be accepted. Suspicion of possible unwanted cross-reactions of provided good technical management and safety can be certain caries vaccine preparations with human tissues has guaranteed. been substantiated, and, therefore, the first generation of these * Although heavy formation of dental calculus is common vaccines will probably not come into use. Genetic engineering, icta poul ationsoi d eloin c us it mihb however, is making possible the developmnent of new versions, in certain populations In developing countries, it might be in which antigens of the mutans streptococci are transferred unrealistic to suggest services like professional removal of into which l antigescfrhe anisms.t istrptoocie tr a nsferr calculus for these populations at present. It should be possi- vaccine will reduce the m utans levels on the teeth with ble, however, to introduce oral hygiene programs among vaccinewill reuce themutans evels o the teth, wit children, which would place the next generation in a better concomitant reduction of caries. Many years of research will . . still be needed, however, if effective vaccination is to be of use position. to developing countries. * Oral manifestations of cancers, or precancerous lesions, Although fluorides have been investigated intensively for are common in some populations. These populations should decades, new products can still be expected. For example, be, or have a chance to be, inspected for early signs of such bgaard and others ( 1988) have suggested that a hydrogen manifestations at the same time that acceptable services for fluoride solution with low pH, resulting in calcium fluoride- treatment are built up. If pathogenic processes are diag- formation on the teeth, could be very effective, and some pilot nosed, it should be possible to obtain surgical or other types experiments have supported their hypothesis. More clinical of treatment. experiments are necessary, however, to prove its usefulness * Serious cleft syndromes in newborn babies are not com- under field conditions. mon, but if present, they require advanced treatment. A Dietary modifications through the use of sugar substitutes referral centerfor these conditions should be available. Such may be effective, but so far the products are, in general, a center should also be able to take care of severe fractures comparatively expensive and present some side effects if con- from accidents and other causes. sumed in large amounts. Industrial countries have more oppor- * If fluorosis is present to a disturbing extent, defluorida- tunity to use the products currently available. It is possible, tion should be performed, if at all possible. however, that new products, or modifications of sucrose, will .. however, avaiatblew products, will modiions o sui frdeveloping As a basis for all their efforts, the authorities should have an become available which will also be suitable for developing oral health program which includes surveys and other data countries. collection to clarify the distribution of oral diseases within the These few examples serve to illustrate the fact that new population and incidence of the diseases. The main "local" strategies may be devised, some within a fairly short time, but etiological factors for the diseases should be known. The effects those particularly useful for developing countries may yet be of preventive programs already implemented should be moni- years away. Still, that technology will change is now apparent; tored. Changes in diet should be observed, in particular, any it will leave drilling and filling for more advanced technology, increase in sugar, and sugar-rich snacks, between meals. Na- based on microbiology and chemistry. The examples, however, tional authorities should work toward the training of dental focus on caries and periodontal diseases. The spectrum of other oral diseases may become affected by the changes exemplified personnel and the development of appropriate technology. only to a certain degree. Rather, there may be a risk of increase Those who select personnel should bear in mind geographical , . , , . . , ~~~~~~distribution. Goals can be formulated: all persons should be in some conditions in developing countries, such as traumatic able to eat, drink, and talk without discomfort. In addition, the injuries from traffic accidents as traffic increases. mouth and teeth should have an acceptable appearance. Expected Minimum Standards Oral Health Personnel For a developing country, it is reasonable to expect that the Oral health services are expensive, as revealed earlier. Man- fol-)Ilowing standards be achieved: power and education account for three-quarters to four-fifths * The population should receive information about the of these costs. Therefore, resource planning should aim at principal oral diseases, thus allowing the individual to avoid ensuring that both the quantity and type of oral health person- Oral Health, 657 nel are adequate. It is certain that savings are possible if range of strategies from interventive to noninterventive and less-educated personnel can be used. It is apparent that several displayed in three main oral health personnel combinations. functions mentioned above could be performed by personnel Clearly, great differences in economic and health conse- other than dentists. It is also clear, however, that certain quences are embodied in these comparative figures, for exam- conditions demand highly skilled personnel. What, then, ple, from 665 oral health personnel per million for the situation would be the best strategy for a given situation and the best that includes high caries prevalence and an interventive strat- combination of personnel types? egy to 95 per million for the situation that embodies low caries One should not expect a simple answer to such a question prevalence and a noninterventive strategy. Similarly, within because the basis for decisions differs so widely among coun- one caries prevalence level the differences are considerable; for tries. But the question has been studied by WHO, and a planning example, the interventive strategy in the high caries preva- instrument has been drawn up. Using the Lotus computer lence level calls for 665 dentists per million, whereas the program, the WHO method for estimating oral health personnel noninterventive strategy for that same prevalence level re- requirements provides a procedure for rapid calculation on the quires only 175 dentists plus 50 auxiliaries plus 40 primary basis of data or estimates for: health workers if the three-level combination of personnel * Oral disease levels is used. As large as these differences are, they can be even larger if * Need for oral care and retreatment frequency one varies such items as demand for care, time per item, and * Demand for care hours worked. The figures in table 27-5 are based on a 2,000- * Average time per item of care hour year. If this is changed to a more likely 1,500-hour year, * Average hours worked or even a 1,200-hour year, which is a reality in many countries and not even a minimum, the lowest and highest estimates of For any single level of disease one can alter the assumptions personnel needed per million become 130 and 890, respec- on the demand for care, on treatment strategies which are more tively, for the 1,500-hour year and 160 and 1,110, respectively, or less interventive, on retreatment frequency, and on time for the 1,200-hour year. allocations per item for time worked. Thus, one can have The full versatility of the system extends to extrapolating widely different estimates of personnel needs for a single oral observed trends in the oral health status and to entering data health status. Also, overall personnel needs can be subdivided appropriate to the achievement of stated, measurable goals in on the basis of use of dental professionals only, professionals the medium to long term. This system is already a powerful and auxiliaries, and professionals and auxiliaries and primary planning tool, and the WHO Oral Health Unit hopes to expand health workers. In the latter case, especially, the numbers it further in the near future with more specific measurement of should read as full-time equivalents rather than simply head economic factors. The importance of the system is also re- counts. Subclassification, on the basis of intervention, allows flected when the number of dentists in relation to the number for an aggressive treatment service, for which a prime example of inhabitants in some countries is considered (table 27-6). is one in which every tooth recorded as carious would be filled, Hypothetical cases have been taken from the list in table 27-6 throughamoderateservice, toanoninterventive system which to highlight contrasts based on actual situations for which uses high-risk assessments to defer filling any tooth as long as assumptions which seem most appropriate have been made. some form of prevention, simple surface care, or even "wait and Applying the WHO method and assuming the partially inter- see," can substitute. ventive strategy, a 1 ,500-hour working year, and a stable caries In table 27-5 we give an example of this type of exercise for level, we conclude that one dental operator is needed in the three distinct levels of dental caries, each managed through a following countries for the populations indicated: Brazil, Table 27-5. Oral Health Personnel Needed per Million Population Dentists plus auxiliaries plus Level of caries Strategy Dentists only Dentists plus auxiliaries pnrimary health workers Low Intervenrive 145 75+70 75+30+30 Partially interventive 105 40+65 40+25+40 Noninterventive 95 35+60 35+15+45 Moderate Interventive 255 165+90 165+65+25 Partially interventive 195 125+70 125+45+25 Noninterventive 160 95+65 95+35+30 High Interventive 665 465+200 465+170+30 Partially inrerventive 365 245+120 245+85+35 Noninterventive 265 175+90 175+50+40 Source: Oral Health Unit, WHO, Geneva. Calculations based on 2,000 hours worked per vear. 658 Douglas Bratthall and David E. Barmes Table 27-6. Dentists per 10,000 Population, * An effective primary oral health organization for by Country prevention and care. If possible, the activities should CountrY Dentists be combined with other health services or school-based programs. Asia * Referral centers as mentioned above. Also, developing Indonesia 0.1 countries are in need of some highly trained experts in Mongolia 0.4 various fields. SriLanka 0.2 * Appropriate technology. It is important that equip- Thailand 0.3 ment, materials, and methods work under field condi- IrLdustrialcountries tions. There are numerous examples of sophisticated and Belgium 6.1 expensive dental units that are not operating in develop- Canada 4.9 ing countries. Denmark 8.8 * Resources for oral health planning, preventive programs, France 7.2 and education. They include surveys, epidemiological stud- Germany 5.7 ies, and the establishment of national or local registries, with Italy 0.6 efficient use of informatics and electronics. Netherlands 4.9 We do not think that the present budgets for oral health in United Kingdom 3.1 most developing countries are sufficient for the addition or United Stares 5.9 enhancement of these items. New resources are necessary, but Latin Anerica and the Caribbean the many demands already present, as well as those expected Argentina 2.2 in the near future, urgently need to be met. Bolivia 0.6 Brazil 1.3 Priorities for Operational Research Colombia 3.6 Costa Rica 3.1 Many important questions and problems for oral health need CUuule y.6 attention. The list below illustrates some of them as seen from Uruguay 7.7 a global standpoint. Another order of priority may very well be Middle East and North America chosen locally, which is why the list should not, by all means, Algeria 1.2 be regarded as the universal truth. Jordan 2.4 Iran 0.5 * Prevalence and incidence of oral diseases. Epidemiological Israel 7.1 investigations should be performed regularly and records Oman 0.6 analyzed to monitor the oral health situation. They should, Pakistan 0.1 Saudi Arabia 1.0 of course, include caries and periodontal disease, but it is Yemen, Rep. of 0 important that they be extended to cover also the more common discomforting, premalignant or malignant oral Sub-Saharan Afca diseases. Children and representative samples of indicator Kenya 0.1Q ages among the elderly should be included. Mali ages Senegal 0.1 * Etiological factors. Concomitant with epidemiological Sierra l eone 0.1 surveys, one should try to identify the main factors causing Togo 00c disease in the population under investigation. Studies a. In Democratic Republic of Yemen, only eighteen dentists in countrv. should be performed aimed at finding effective means of b. Only fifteen dentists in country. reducing such factors or risk behaviors. c. Ony five dentists in countr. * Treatment needs. Improved methods and tools to evalu- ate treatment needs and demand for oral care are necessary. The methods should include an attempt to determine the 2,990; Costa Rica, 5,488; Germany. 2,941; India, 5,696; Jor- total needs of the population, taking into consideration also dan, 8,036; Pakistan, 4,036; Sierra Leone, 10,976; Sweden, the less prevalent diseases. 2,601; the United Kingdom, 2,616. * Preventive programs. A variety of preventive programs should be installed and new strategies should be tested. The Priorities for Resource Allocation possibilities of selecting risk groups should be further inves- tigated, and various combinations of oral health personnel We believe resources in developing countries should be allo- tested. In particular, it is important to establish the effect of cated to the following: primary health care workers during longitudinal studies. Oral Health 659 Also, collaboration with nondental personnel within pre- Ax6ll, T. E. 1976. "A Prevalence Study of Oral Mucosal Lesions in an Adult vention needs further study. Swedish Population." Odontologisk Revw 27(supplement 36):1-103. ToothpasteandotheroralhealthpToducts It is believed that Ax6ll, T. E., R. Bte Zain, and P. Siwamogstham. 1990. "Prevalence of Oral Soft Tissue Lesions in Out-Patients at Two Malaysian and Thai Dental fluoride toothpaste and other fluoride-containing products Schools." Community Dennstry and Oral Epidemiology 18:95-99. have been of great importance in decreasing the prevalence Barmes. t). E., and P. Leous. 1986. "Assessment of Periodontal Status by of caries in industrial countries. The use of such products in cPITN and Its Applicability to the Development of Long-Term Goals on developing countries should be promoted, but the outcome Periodontal Health of the Population." International Dental Journal of such activities is also dependent on the price of the 36:177-81. products. New products, more afforable than those produced Ja'nczuk, Zbigniew. 1989. "Oral HIealth ofPolish Children and WHO/FDI Goals today, must become available, for the Year 2000." CommunitY Dentstry and Oral Epidemiology 17:75-78. Jeboda. S., and Harald Eriksen. 1989. "Primary Oral Health Care: The Con- Transfer of kwwkedge. The results of various oral hiealth cept and Suggestions for Practical Approach." Odonto-StomatologieTropicale projects must be readily communicated so that good ideas can 11:121--26. be picked up quickly by other countries or communities. Re- Johnson, N. W. 1989. "Detection of High Risk Groups and Individuals for search should advise on improved methods to transfer knowl- Periodontal Diseases." International DentalJournal 39:33-47. edge and skills to developing countries. In particular, the Karlsson. Goran. 1986. "Samhallsekonomisk utvardering av kakbens- possibilities offered by new technologies should be explored. forankrade broaren forstudie." Report 8. Center for Medical Technology Assessment, University of Linkoiping, Sweden 1-78. Kelly, J., and C. Harvey. 1977. "An Assessment of the Teeth of Youths 12-17 Notes Years." DHEW Publication (HRPA) 77-1644. National Center for Health Statistics, U.S. Public Health Service, Washington, D.C. During the preparation ofthischapter, we have received valuable comments Meier, C. 1988. "Wieviel ist Prophylaxe wert ?" Schweizerische Monatsschriftfiir and suggestions from a number of colleguies who have read parts of the Zahnmedizin 99:647. manuscript or draft versions. We would like to thank all these persons for their Ogaard, B., Gunnar Rblla, Joop Arends, and J. M. ten Cate. 1988. "Orthodon- help and for the time they have spent with the issue. In particular, we want to tic Appliances and Enamel Deminerahzation. 2. Prevention and Treatment thank A. Adeyinka, J. Ahlgren, T. Axdl. G. Bratthall, J. R. Freed,J. Frencken, of Lesions." Amenican Journal of Orthodontics and Dentofacial Orthopedics D. T. Jamison. J. McCombie, Ff. Miyazaki, W. H. Mosley, T. Pilot, 94:123-28. S. Schweitzer, R. SerinirachA Y. Songpaisan' and W.D. Sithole. Pilot, Taco. 1988. "Trends in Oral Health: A Global Perspective." New l. A billion is 1,000 million. Zealand Dental]ournal 84:40-45. Pilot. Taco, D. E. Barmes, M.-H. Leclercq, B. McComble, and Jennifer Sardo References Infirri. 1986. "Periodontal Conditions in Adults, 35-44 Years of Age: An Overview of cpirN Data in the WHO Global Oral Data Bank." Community Andreasen, J. O., J. J. Pindborg, E. Hjorting-Hansen, and T. E. Axell. 1986. Dentistry and Oral Epidemiology 14:310-12. "Oral Health Care: More than Caries and Periodontal Disease. A Survey of Pindborg, J. J. 1989. "Classification of Oral Lesions Associated with HIV Epidemiological Studies on Oral Disease." International Dentaljournal 36:207-14 Infection." Oral Surgery, Oral Medicine, Oral Patholgy 67:292-95. Andreasen, J. O.. and J. J. Ravn. 1972. "Epidemiology of Traumatic Dental Sandham, J., J. Brown, H. Phillips, and K. Chan. 1988. "A Preliminary Report Injuries to Primary and Permanent Teeth in a Danish Population Sample." on Longterm Elimination of Detectable Mutansstreptococci in Man."journal Internawonal Journal of Oral Surgery 1:235-39. of Dental Research 67:9-14. 28 Schizophrenia and Manic-Depressive Illness Peter Cowley and Richard Jed Wyatt Schizophrenia and manic-depressive illness constitute a tre- state prevalence; therefore, it includes both old patients and mendous health burden. They affect 2 percent of the world's new patients. population during life's most productive years, in turn straining family and other resources (Goodwin and Jamison 1990; Risk Factors Jablensky and others 1992). The diagnoses of schizophrenia and manic-depressive illness rely on objective criteria that can There are few discernible risk factors for schizophrenia and be used by trained health professionals. When broadly defined, manic-depressive illness; nonetheless, twin and adoption stud- schizophrenia can include both brief and chronic forms of the ies indicate that schizophrenia and manic-depressive illness illness. Manic-depressive illness can occur in milder (bipolar have a genetic component (Gottesman and Shields 1982; 11) and severe forms (bipolar 1). In both forms of bipolar Goodwin and Jamison 1990). Schizophrenia and manic- disorder an individual's mood, energy level, and cognition vary depressive illness have approximately the same incidence and greatly over shorter or longer time periods. In order to limit the prevalence in males and females (Robins and Regier 1991; scope of this project, manic-depressive illness is used synony- Jablensky and others 1992). The peak age for developing mously with bipolar I and 11 disorders. schizophrenia and manic-depressive illness is approximately Although most epidemiological studies of schizophrenia are twenty; all but a few cases develop initial symptoms before age from industrial countries, a recent World Health Organization thirty-five (Goodwin and Jamison 1990; Jablensky and others (WHO) study provides epidemiological information from devel- 1992). oping communities (Jablensky and others 1992). There is, Socioeconomic class is a possible risk factor for schizophre- however, painfully little known about manic-depressive illness nia, those living in poor socioeconomic conditions having in developing countries (Goodwin and Jamison 1990; Robins high incidence and prevalence rates (Robinson and Regier and Regier 1991). 1991; Jablensky and others 1992). It is not known if the poor The treatment of schizophrenia serves as the primary tern- socioeconomic conditions in urban areas pose a risk or if plate for the cost-effectiveness calculations of how the program persons with schizophrenia, because of their disease, migrate might work for manic-depressive illness. By showing how the into these socioeconomic conditions. There is less of a "down- cost-effectiveness template can be used for more than one ward" social drift among persons with manic-depressive illness, mental illness, we hope that further research can be carried out which may be because manic-depressive individuals often have on the cost-effectiveness of treating other mental illnesses, high energy and enthusiasm, which are correlated with social such as unipolar depression. The treatment protocol presented achievement (Bagley 1973). here has been simplified from what would be expected in the most industrialized parts of the world, making it practical and Incidence and Prevalence: Schizophrenia less costly, if also less exacting. Plans include a referral base (general medical practioners, psychiatrists, families, tradi- Studies done by WHO indicate that the incidence (derived from tional healers, and herbalists) from which individuals with the time of the initial provider contact and interviews with such symptoms as "odd behavior," nontraditional violent out- family members about a patient's psychiatric history) of bursts, and delusional thinking can be sent to a clinic for "broadly defined" schizophrenic patients who seek treatment evaluation and possible medical treatment. is between 15 and 52 per 100,000 people (age fifteen through The clinic will be staffed by trained auxiliary health workers fifty-four), the developing world reporting the higher figures and nurses; the client will visit the clinic once a month, and a (Wig 1982; Jablensky and others 1992). "Broadly defined" trained psychiatrist will visit the clinic once a week to attend schizophrenic patients include those with evidence of a psy- patient review sessions. Finally, the model assumes a steady- chotic state, such as nuclear schizophrenia, paranoid state, 661 662 Peter Cowley and Richardjed Wyatt acute paranoid reaction, alcohol- or drug-induced hallucinosis, Table 28-1. Course of Schizophrenic Disease unspecified psychosis, probable and borderline psychosis in Developing and Industrial Countries (Jablensky and others 1992). (percent) Approximately 30 to 40 percent of individuals in the devel- Location Acute Chronic Intermittent Total oping world who have experienced psychotic episodes feel b persecuted or neglect daily tasks such as personal cleanliness Developingregions' 27.5 37.1 32.7 97.3 (Jablensky and others 1992). If only those patients who show Industrial regions' 8.5 63.8 25.3 98.6 nuclear symptoms of schizophrenia (delusions of control, feel- Note: Acute coturse: full remission, no firther episodes. Chronic course: at ings of someone inserting thoughts, or auditory hallucinations) least (ne subsequent psychotic episode, with incomplete remission between are tabulated, the incidence decreases to between 7 and 14 per episodes, or continuous psychotic episodes. Intermittent: partial remission, no are ~~~~~~~~~~~~~~~~~~~further episodcs, or at least one subsequent episode with full remission between 100,000 (age fifteen to fifty-four); such individuals usually episodes show at least intermittent symptoms for the rest of their life a. Agra. India; Cali, Colombia; and Ibadan, Nigeria. (Jablensky atnd others 1992). Comparison with other annual b. The course of illness in some patients did not fit the definition of acute, chronic, or intermittent. incidence rates from developing countries is difficult because c. Aarhus. Denmark; London. England; Moscow, Russia; Prague, Czecho- of differing diagnostic criteria. Nonetheless, other studies from o'ovakia; and Washington, D.C., United States. Asia indicate an incidence of between 2 and 11 per 100,000 Source: Leff and others 1992. in those age fifteen and above; Beijing, China, reported an incidence of schizophrenia of 11 per 100,000 in persons age Morbidity and Mortality: Manic-Depressive Illness fifteen and above (Yucun and others 1981; Wig 1982). Retrospective studies indicate that 0 to 55 percent of non- Incidence and Prevalence: Manic-Depressive Illness prophylactically treated manic-depressive patients had only one episode, whereas 13 to 42 percent had two to three The authors of three studies in northern Europe have found an episodes, 8 to 40 percent had four to six episodes, and 2 to 69 annual incidence of manic-depressive illness of between 11 percent had more than seven episodes (reviewed in Goodwin and 21 per 100,000 for persons age fifteen and older who seek and Jamison 1990). It has been estimated that 22 percent of treatment (reviewed in Goodwin and Jamison 1990). Individ- manic-depressive patients (mainly women) have a chronic uals with manic-depressive illness often present with: inflated course with virtually no normal intervals between mania and self-esteem, distractibility, increased pleasure, decreased sleep depression (Tsuang, Woolson, and Fleming 1979). patterns or signs of fatigue, feelings of worthlessness, or suicidal ideation. Commonly, the disease course is cyclical in nature, Burden of Schizophrenia and the threat of a manic or depressive episode continues throughout the rest of the individual's life. It is suspected that At fifteen-year follow-up, persons with schizophrenia and manic-depressive illness is not as prevalent in developing other categories of psychosis reported mortality rates 1.8 times countries as it is in industrial ones (Goodwin and Jamison greater than the general population, with 5 percent of patients 1990). adequately treated (Lin and others 1989). Approximately 8 percent of schizophrenic patients in the United States kill Morbidity and Mortality: Schizophrenia themselves, at an average age of thirty-one; similar rates of suicide are suspected among individuals with schizophrenia It has been reported that the course of schizophrenia is less from the developing world (Roy 1986; Leff and others 1992). severe in developing countries (jablensky and Sartorius 1988). Psychiatric disorders such as schizophrenia and manic- The differing disease course patterns for suboptimally treated depressive illness exert a tremendous toll on the emotional and patients from developing and industrial countries are shown in socioeconomic capabilities of both patient and caretaker. It table 28-1. Data in this table indicate that patients from has been reported that between 17 and 50 percent of schizo- developing countries are more likely to have a brief psychosis. phrenic patients being followed up had severe social impair- World Health Organization research indicates that several ment (Leff and others 1992). Other research indicates that factors are associated with a goodl prognosis (being female, more than 25 percent of the schizophrenic population in the married, and having an acute onset); however, these prognos- United States is unable to work or perform homemaker respon- tic indicators and the presence or absence of effective treat- sibilities and that those capable of working or performing ment do not completely explain the differences in outcome housekeeping have at least a 25 percent disability (ongoing between patients from the developing and industrial worlds research by Wyatt). (Jablensky and others 1992; Leff and others 1992). The better prognosis for schizophrenic patients from the developing world Burden of Manic-Depressive Illness may be influenced by the use of a broad case definition or differing demands posed by the particular society (Wyatt and Untreated manic-depressive illness is reported approximately Stevens 1987; Jablensky and Sartorius 1988). to double the yearly risk of dying, with suicide causing the SchizophreTua and Manic-Depressive lUness 663 majority of the excess mortality (Goodwin and Jamison 1990). reducing the intensity of mania or depression; correspondingly, It is estimated that one in five manic-depressive patients there is a decreased episode frequency (Goodwin and Jamison commits suicide, often within the first five years of the disease 1990). By reducing the intensity of mania or depression, lith- onset (Goodwin and Jamison 1990). Suicide is suspected to be ium also helps reduce suicide. Research has shown that 1.17 less common among manic-depressive patients in the develop- suicide attempts per patient were made before lithium and 0.18 ing world, which may be a result of lower incidence rates, suicide attempts per patient were made after lithium was differing cultural manifestations of the disease, or a cultural introduced (Causemann and Muller-Oberlinghausen 1988). bias against reporting suicide (Wittkower and Rin 1965). Due to their illness, approximately 20 to 30 percent of individuals Cost-Effective Schizophrenia Case Management with manic-depressive illness in the United States are unable to work (ongoing research by Wyatt). There is no model available regarding the cost-effectiveness of either a schizophrenia or manic-depressive outpatient medical Therapeutic Strategy for Schizophrenia program in the developing world. The following model is an attempt to provide estimates of the cost and effectiveness of a Such antipsychotic medications as fluphenazine, haloperidol, case management program based on a best-case scenario; it is and chlorpromazine reduce the length of psychotic episodes in presented in equation format with results in table 28-2. schizophrenic patients and can prevent relapses. Within six The annual cost of a medication treatment program for weeks after starting treatment with antipsychotic medication, individuals with schizophrenia per million population age 40 to 50 percent of active and chronic schizophrenic patients fifteen and above is the sum of: experience remission (Rifkin and others 1991). In addition, * Estimate of tTeatment costs per acutely psychotic patient maintaining patients on antipsychotic medication produces treated successfully (Cost,e ). The outpatient costs per a 50 to 60 percent reduction in relapse rates (Baldessarini, acutely psychotic patient who responds to medication and Cohen, and Teicher 1990; Rifkin and others 1991). wll receve antipsychotc edication for one year is equal Anripsychoric medication has been reported to cause a 50u ilrcieatpyhtcmdcto o n eri qa Antipychoic mdicaion as ben reorte to ausea 50 to: (Nacute x V ) + Cacute, where Nacute number of outpatient to 60 percent decrease in the severity of illness for both acute ts neex V ) + each acutely of outphrent and chronic psychotic patients as rated by standard clinical visits needed per year by each acutely e il schizophrenic rating scales (Santos and others 1989; Baldessarini, Cohen, patient; V cost of one outpatient visit; and Cacute = yearly and Teicher 1990). Relapse rates can be further reduced by c antipschot ic a atic nec ication for using long-acting injectable antipsychotic medications (Bal- e ate il schienic patient dessarini, Cohen, and Teicher 1990). When a schizophrenic * Estimate of treatment cost a puer acutely psychotic patient patient improves clinically, social outcome also improves. It treatedunsuccessfully(Costyuchoti t ess)n Becausew a fraction of has been reported that there is a 0.5 correlation between the acutely ill psychotic patient pool who will be given clinical and social improvement as measured by interpersonal antipsychotic medications for a period of up to six months relationships, sociability, leisure activity, and work activity will not respond, the following equation is needed: (Shepard and others 1989). (Nacute/unsuccess x V) +e ban where Nacurelunsuccesp = Research indicates that once patients have discontinued the number of visits used by an acutely psychotic patient antipsychotic medication, they tend to have a much poorer who isdroppedefromthe programeafter three monthsbecause social outcome, including a decrease in work productivity and of nonresponsiveness to medication, and atre/unsuccess = an increase in sociallydisruptive behavior (Johnson and others cost of medication in acutely psychotic patients who are 1983). There is an apparent decline in the suicide risk among dropped from the program because of nonresponsiveness to treated schizophrenic patients, compared with the risk among medication. those who are untreated; one study indicated that 27 percent * Estimate of treatment costs per chronically ill patient served of schizophrenic patients attempted suicide when neuroleptic (COstch,o1c) . The outpatient costs for patients with schizo- medication was discontinued, whereas only 11 percent at- phrenia who will receive long-term antipsychotic therapy tempted suicide while on medication (Johnson and others is: (Nchronic x V) + Cchronic, where Nchronic = number of 1983). outpatient visits needed per year by each chronically ill schizophrenic patient, and Cchrontc = yearly cost of medica- Therapeutic Strategy for Manic-Depressive Illness tion for a chronically ill schizophrenic patient.I * Estimate of treatment costs per intermittently ill patient Roughly 80 to 90 percent of manic-depressive patients on treated (Costin ermitten, ). The outpatient costs for patients lithium respond favorably, resulting in a 60 to 80 percent who will receive intermittent antipsychotic medication reduction in relapses (Goodwin and Jamison 1990). Further- is: (Ninrermittent x V) + Cintermittent, where Nintermittent = more, it has been reported that both manic and depressive number of outpatient visits needed per year for each episodes are 80 percent less frequent in patients treated with intermittently ill schizophrenic patient, and Cintermittent = lithium (Holinger and Wolpert 1979; Rybakowski and others yearly cost of medication for an intermittently ill schizo- 1980). Lithium helps to decrease the duration of an episode by phrenic patient. 664 Peter Cowley and Richard Jed W-yatt In order to simplify the equations, the estimated number Manic-Depressive Illness Case Management of treated patients in the treatment groups (number- schizophrenia) is: (Ischizophrenia x P x SS) x (PT x E), where The cost-effectiveness of a lithium medication program can be Ischizophrenia = incidence of broadly defined schizophrenia calculated using exactly the same equations as those used per I million population age fifteen through fifty-four;2 for the neuroleptic medication program. The input functions P = proportion of broadly defined acutely ill schizophrenic for the case management program for lithium treatment of patients of any age (Ischizophrenia) who will fall into either an manic-depressive illness are noted in table 28-4. The manic- acute, a chronic, or an intermittent disease outcome; depressiveillnessmedicationinterventionprogram,withtotal SS = multiplication factor to account for preexisting pa- cost figures of $268.00 per disability-adjusted life-year gained tients; PT = proportion of population of each disease and $0.092 per person (age eighteen and above), like the outcome who will be correctly diagnosed and appropri- program for schizophrenic patients, is affordable compared ately treated at the health center (does not include those with other adult chronic disease interventions, and it has a who dropped out because of medication side-effects); and similar distribution of benefits, as does the schizophrenia treat- E = percentage of each pool of patients that responds to ment program (Jamison, chapter 1, this collection). ant ipsychotic medication. 3 * Estirnate of disability-adjusted life-years (DALYS) gained per Gaps between Good and Actual Practice acutely psychotic patient successfully treated (DALY acue ). The number of DALYs gained per acutely ill patient successfully Bothersome but nonlethal side effects from antipsychotic med- treated can be calculated as follows: R x (Hacute + CTacute ication used to treat schizophrenia are numerous and include + Macuce), where R = percentage reduction in patient and dry mouth, constipation, decreased libido, and tremors, all of caretaker's disabilities and patient's excess mortality with which often can be treated by lowering the dose; tardive medication; Hacute = acutely ill patient's quality-of-life dis- dyskinesia (repetitive involuntary movements) and neurolep- ability; CTacute = disability per acutely ill patient from his or tic malignant syndrome (high fever and blood abnormalities), her caretaker's reduced quality of life; and Macure = acutely which are potentially more serious, can contribute to the ill patient's yearly increase in the risk of dying. already serious morbidity (Baldessarini, Cohen, and Teicher * Estimate of DALYS gained per chronically ill patient treated 1990). Many patients who are given antipsychotic medication, (DALY chronic ) The number of DALYs gained per treated chron- particularly early in their treatment course, are also given ically ill patient treated is: R x (HChronc + CTchr0IC + M , anticholinergic and antiparkinsonian medication to prevent where Hchron,c = chronically ill patient's disability from his acute dystonic reactions (rigidity of muscle groups, especially or her reduced quality of life; CTchrontc = disability per facial), which, when they occur, are both painful and fright- chronically ill patient from his or her caretaker's reduced ening (McKane and others 1987; Santos and others 1989). quality of life; Mchronic = chronically ill patient's yearly Compliance by the patient in taking antipsychotic medication increase in the risk of dying. is another important issue, and it has been estimated that as few as 50 percent of schizophrenic patients take their medica- * Estimate Of DALYS gained per intermittently ill patient treated tion as prescribed (Wilcox, Gilian, and Hare 1965). As with (DALY itermittert ). The number of DALYS gained per intermit- all medications, describing potential side effects before they tently ill patient treated is: R x (H,nterinittent + CTintermittent + all and making ertain tentialnsid eff e nt are Mintermittent), where HLntermitent = intermittently ill patient's occur and maklng certain rhe diagnosis and treatment are disability from his or her reduced quality of life; CTintermittent = appropriate sho2ul increase compliance. disability per intermittently ill patient from his or her Uprto i compain of ithium for or caretaker's reduced quality of life; and Mintermittent = chron- depressive illness complain of excessive thirst, tremor, or icaly ll chzoprenc atint' yerl inreae n te rsk memory problems (Goodwinand Jamison 1990). Lithium may ically ill schizophrenic patient's yearly increase in the risk be harmful to fetuses and can also cause decreased thyroid of dying. activity (Mannisto 1980). Toxicity due to lithium overdose In table 28-2 we display the equations used in determin- can cause serious kidney damage and is often fatal (Goodwin ing the cost-effectiveness of the model, and in table 28-3 we and Jamison 1990). Noncompliance of lithium users in taking present inputs used for the schizophrenia medical treatment their medication ranges between 18 and 53 percent and is often program.4 The $223.00 per disability-adjusted life-year thought to be connected to the denial of the disease as well gained is less expensive than most of the adult chronic as the side effects (Goodwin and Jamison 1990). Manic- disease interventions, such as coronary artery bypass surgery depressive patients need to be monitored for suicidal ideation, and cancer treatment programs (Jamison, chapter 1, this particularly during the first few years of the disease, when the collection). The program would cost approximately $0.104 risk is the highest. per person (age fifteen or older), not an unreasonable bur- Ethical and legal issues regarding giving medication to pa- den for most health care systems. Over 95 percent of the tients who cannot give consent (particularly those in acute case management benefits are from reductions in the patient psychotic states) need to be considered in relation to local and caretaker's quality-of-life disability, with little contri- standards, and possible dispositions for aggressive patients also bution from averted mortality. need to be explored. If antipsychotic medication or lithium is Table 28-2. Derived Variables and Their Most Likely Values Name or symbol Variable Derivation Most likely value Number-schizophrenia acutc Number of acutely ill patients treated successfully (Ich,i-iVh-enia X Pacut. x PTacute) x Eacute 18 Number-schizophreniaacute/unsuce. Number of acutely ill patients treated unsuccessfully (hch,inphrn. x PX ute/unwccem x PTatie/unsuccets) x( - Eacute) 15 Number-schizophrenia chronic Number of chronically ill patients treated ('schizophrenia X Pchronc X PTchr-n) x Echronic 696 Number-schizophrenia ntireni Number of intermittently ill patients treated (Is-htzphrenia x Pinrermirrent x PT,ntermirrenr) x Eintermirtent 141 Cost ate Cost per acutely ill patient treated successfully (N_,0. x V) + Caci.rc $160 Cost acorn/unioccem- Cost per acutely ill patient treated unsuccessfully (Nacureluncuc- x V) + Cacureiun-uc-t $63 Cost chronic Cost per chronically ill patient treated (Nchronic X V) + Cchronic $117 Cost inrermi,eni Cost per internittently ill patient treated (Ninterirretr x V) + C,nrerinir,ent $140 DALY Ate Disability-adjusted life-years gained per acutely ill patient R x (Hactue + CT-ue + Ma0u0e) 0.30 treated successfully DALY chronic Disability-adjusted life-years gained per chronically ill R x (Hchronic + CTchranic + Mchntitic) 0.60 patient treated DALY intermitent Disability-adjusted life-years gained per intermittently ill R x (Hinte,iieni + CTitermirrent + Minteriiatent) 0.30 patient treated U Total number ofdisability-adjusted life-years gained from (Number-schizophrenia acute X DALY acute) + 465 case management program (Number-schizophrenia chronc X DALY chronic) + (Number-schizophrenia inteittirent X DALY inteeurenr) Z Total cost of schizophrenia program (Number-schizophrenia acure X Cost acute) + $104,999 (Number-schizophrenia acure/uroucceuu x Cost aculefunnucce.a) + (Number-schizophreniachronic x Costchronic) + (Number-schizophreniaintermittent x COStintermitent) Q Cost per disability-adjusted life-year gained Z/U $223 Source: Aithors. 666 Peter Cowley and Richard Jed Wyatt Table 28-3. Variables for Model of Schizophrenia Case Management Value Symbol Variable Acute Chronic Intermittent P Proportion of schizophrenic population 0.28 0.37 0.33 PT Proportion who will receive correct treatment 0.40 0.60 0.50 SS Steady-state factor to account for preexisting patients 1.00 17.4 5.8 N Number of clinic visits needed per year 302 12 21 V Cost of one outpatient visit $2.25b $2.25 $2.25 C Yearly medication costs $92 $90 $92 E Percentage of patients responding to medication 0.50 0.60 0.50 H Patient's decreased quality-of-life disability 0.40 0.80 0.40 CT Caretaker's decreased quality-of-life disability 0.20 0.40 0.20 M Patient's yearly increase in risk of dying 0.0035 0.007 0.0035 R Percentage reduction in patient's/caretaker's disabilities and 50 50 50 mortality risk when patient taking medication Note: Total incidence (I) of schizophrenia (broadly defined) is 300 cases per million people age fifteen to fifty-four. a. Twenty-one visits for acute patient unsuccessfully treated. b. $16 for acute patient unsuccessfully treated. Source: Authors. given by nonpsychiatrists and without serum monitoring, psychotherapy aimed at stress reduction and education about missed opportunities for dosage adjustment and reducing side the illness together with antipsychotic medication for schizo- effects will be increased. Furthermore, the use of antipsychotic phrenia and lithium for manic-depressive illness appear to medication or lithium in the developing world needs to be make both illnesses more manageable (Baldessarini, Cohen, investigated, because medication may work better or worse in and Teicher 1990). those areas. The cost-effectiveness of both programs relies Research on the epidemiology of schizophrenia and manic- heavily on community and family support. If the community depressive illness is of importance; of particular importance does not respond, effectiveness will decrease as the costs in- is a determination of associated disability. There are numer- crease because patient compliance will not be encouraged and ous rating scales which measure social activities, level of assistance in helping the patient remain in the program (trans- anxiety, activity, ability to perform work, self-esteem, and port, jobs, housing, and so on) will be lacking. same and opposite gender friendship that could be used with local modification to help determine disability (Weiss and Priorities for Control others 1985). A treatment protocol that would enable su- pervised primary-level health workers to identify and cor- Case management with medication is not the only alternative rectly treat psychotic and bipolar disorders needs to be for these patients, but it may be the most cost-effective alter- developed. Research also needs to identify contact points native. Psychotherapy alone has been proven to be of limited (general practitioners, herbalists and traditional healers) benefit compared with medical treatment, but some forms of that would act as bases for referral of psychiatric patients to Table 28-4. Variables for Model of Manic-Depressive Illness Case Management Value Symbol Variable Acute Chronic Intermittent P Proportion of manic-depressive population 0.22 0.55 0.33 PT Proportion who will receive correct treatment 0.30 0.50 0.40 SS Steady-state factor to account for preexisting patients 1.00 16.7 4.2 N Numberofclinicvisitsneededperyear 30a 12 21 V Cost of one outpatient visit $2.25b $2.25 $2.25 C Yearly medication costs $92 $90 $92 E Percentage of patients responding to medication 70 60 70 H Patient's decreased quality-of-life disability 0.30 0.60 0.30 CT Caretaker's decreased quality-of-life disability 0.10 0.20 0.10 M Patient's yearly increase in risk of dying 0.004 0.008 0.004 R Percentage reduction in patient's/caretaker's disabilities and 60 60 60 mortality risk when patient taking medication Note: Incidence (I) is eighty cases per million people age eighteen and older. a. Twenty-one visits for acute patient unsuccessfully treated. b. $16 for acute patient unsuccessfully treated. Source: Authors. Schizophreria and Manic-Depressive Illness 667 a clinic-based medicine intervention program and to ensure followed by injections once each month of generic long-acting patient compliance with the program. fluphenazine, and that one-half of the patients need 50 milli- grams of generic diphenhydramine hydrochloride to prevent dystonic side effects. The medication costs per year for the Appendix 28A. Sources Used to Obtain acutely or intermittently ill schizophrenic patient (Cacute or Cost-Effectiveness Estimates CinteTmittent) is thus $92 (wholesale price from survey of produc- ers). The yearly cost of medication for the acutely ill patient This section discusses sources for cost-effective case manage- who does not respond to treatment (Cacute/nonresp) is $16 for the ment of schizophrenia and manic-depressive illness. initial two weeks of oral medication plus two months of inject- able medication. The cost of the chronically ill patient's med- Schizophrenia ication (Cchrontc) is $90, because these patients do not need the initial stabilizing doses of fluphenazine. The yearly incidence rate of broadly defined schizophrenia The effectiveness values of antipsychotic medications (E) (Ischizop,hren,a) of 300 per 1 million population, age fifteen are derived from data that show that the decrease in relapse through fifty-four, is based on published results from the devel- rates when using long-acting antipsychotic medications, in oping world (Jablensky and others 1992). The proportion of patients who show an initial response to antipsychotics, is individuals with schizophrenia who fall into the three types of approximately 60 percent (E,h10ni,) and that 50 percent of disease outcome categories (P) is derived from table 28-1 actively psychotic patients (or relapsing intermittently ill pa- (developing world data with broad case definition of schizo- tients; Eacute or Einrerm,ttent) will remit within six weeks after phrenia). Pacute patients are projected to need antipsychotics beginning antipsychotic therapy (Johnson and others 1983; for only one year with no other treatment necessary, whereas Baldessarini, Cohen, and Teicher 1990; Van Putten, Marder, Pinterm,ttent patients are assumed to need antipsychotics for a and Mintz 1990). year's duration every third year, and Pchronic patients will need The reduction of clinical and social impairment, or quality- antipsychotics for the remainder of their lives. of-life disabilities (H), for the acute (40 percent), chronic (80 The 17.4 steady-state factor for the chronic (SSchronic) pa- percent), and intermittent patients (40 percent) through med- tient pool and the 5.8 factor for the intermittent (SSIntermittent) ication is based on various clinical rating scales; the caretaker's pool are based on a cohort calculated to remain in the program decreased quality-of-life disability (CT) is projected to be for thirty-nine years with a 5 percent yearly dropout rate.5 one-half of the patient's quality-of-life disability.6 The steady-state factor of intermittently ill patients is one- The yearly increase in the risk of dying (M) is from research third that of chronically ill ones, because they are projected to showing that individuals with schizophrenia in Taiwan need medication only every third year. The proportions ofeach (China) had approximately a 200 percent standardized mor- disease pool treated (PT), ranging from 0.40 to 0.60, are clin- tality ratio in a fifteen-year follow-up (Lin and others 1989). ically based; those in the acute pool are the ones least likely to Survival statistics for Mexican males in 1980 from age twenty enter the program, because their illness is of shorter duration. to sixty were then used to calculate a 0.007 yearly increase in Thirty outpatient contacts per year (N) are needed to treat the risk of dying (M) for the schizophrenic population (United an acutely ill schizophrenic patient. This estimate is based on Nations 1982). the assumption that the patients will need fourteen consecu- The reductions in quality-of-life disabilities for both patient tive daily visits for initial stabilization, followed by weekly and caretaker and the reduction in excess mortality risk seen visits for the next six weeks, then monthly visits for the for the patient (R) from the patient's taking antipsychotic remainder of the year. The acutely ill patient who does not medication assume that once the patient's life improves, the respond to treatment will be treated for two and a half months caretaker's life will improve at the same rate. The reductions (twenty-one visits). The intermittently ill schizophrenic also assume a 60 percent clinical improvement by clinical patient's course of illness would probably be known to the rating scales and a published correlation between clinical and health center staff, making initial stabilization less time con- social improvement, resulting in a projected 50 percent reduc- suming; thus only twenty-one visits will be needed (seven tion figure (Santos and others 1989; Shepard and others 1989; consecutive daily visits, three weekly visits, and eleven Baldessarini, Cohen, and Teicher 1990). The same reduction monthly visits). The chronic pool of schizophrenic patients input is used for decreasing the patient's excess mortality risk was projected to need outpatient visits once a month for and is based on research indicating a 50 percent reduction in antipsychotic medication injections each year. The $2.25 out- suicide attempts with medication and an assumption that if patient cost per visit (V) is derived from calculations in table suicidal behavior is minimized so will other dangerous behav- 28A-1 showing a hypothetical summary of capital and annual ior which causes the high mortality risk (Johnson and others costs of health clinics and outposts that have a capacity of 1983). approximately 30,000 mental health visits per year. The disability associated with decreased quality of life for It is projected that all patients diagnosed with schizophrenia patient and caretaker and the increase in the risk of dying for are placed on antipsychotic medication, beginning with two the acute and intermittent groups (when relapsed) are pro- weeks of 2 milligrams of generic oral fluphenazine per day, jected to be one-halfthose ofthe chronic group. The reasoning 668 Peter Cowle-v and Richard Jed Wyatt Table 28A-I. Annual Costs of Schizophrenia and Manic-Depressive Case Management Programs Vanrable Assumpions Cost (dollars) Operating costs Salaries: case management officers Fifteen officers per center; $1,000 per officer per year 45,000 Salaries: nurse practitioners One per center; $1,500 per nurse per year 4,500 Retirement pensions 5 percent of salaries 2,475 Supplies $250 per center, not including drugs 750 Utilities Gas, water, and electric; $150 per center 450 In-service training $500 per center 1,500 Supervision One full day each month by psychiatrist at each center; per diem and transportation 2,000 Maintenance and repair: buildings 1.5 percent of construction price per year 900 Maintenance and repair: equipment 15 percent of purchase price per year 112 Transportation between centers and Twelve visits per year by motorcycle to each village; twenty villages per villages center; fifteen miles round trip; 20 cents per mile for gas 2,160 Public education None 2,000 Contingencies None 3,000 Total 64,847 Capital costs Buildings Annualized with thirty-year life span; $20,000 per center original cost 2,000 Equipment Annualized with ten-year life span; $250 per center 75 Vehicles One motorcycle per center; annualized with five-year life span; $1,000 per motorcycle 600 Total operating and capital costs 67,522 Cost per clinic visit 30,000 patient visits per year 2.25 Note: Assumes three centers. Source: Authors; Over 1991. for this is straightforward: the length of a psychotic episode depressive population than the schizophrenic population will based on data concerning hospital stays is six months; thus, the undergo treatment because the manic-depressive disease can disability of these two groups exists for only one-half of the year often be "hidden" with greater ease in society. The proportions (Shepard and others 1989). vary between 0.30 and 0.50, with the chronic patients most likely to enter because of the length of illness. Manic-Depressive Illness The number of outpatient visits needed per year (N) and the cost of one outpatient visit (V) are projected to be the same The incidence of manic-depressive illness (Imanic-depressive) in for the course of each subtype of manic-depressive illness as the developing world is projected to be 80 per 1 million people they were for the course of the subtype pools of schizophrenia. aged eighteen and older, substantially less than the 150 per The cost of $92 per year for lithium (C) assumes an average million reported from the industrial world (Goodwin and daily dose of between 900 and 1,000 milligrams, which is the Jamison 1990). The proportion of manic-depressive individu- traditional dose of lithium for the manic-depressive patient als (P) who fall into the acute (only 1 manic-depressive epi- (Goodwin and Jamison 1990: wholesale price from survey of sode), chronic (7 episodes), and intermittent (2-7 episodes) producers). This price includes, for the acute and intermittent disease categories is a derivative from research with a fifteen- patient (but not for the chronic patient), a two-week 2-milli- year follow-up period (Goodwin and Jamison 1990). gram daily dose of oral fluphenazine. As with the schizophrenia For chronic manic-depressive patients the steady-state fac- case management plan, the acutely ill manic-depressive pa- tor is 16.7 (SSchronc) and is formulated much like the program tient unsuccessfully treated will be in the program for only two for schizophrenic patients, except the cohort was followed for and a half months, needing twenty-one clinic visits and med- only thirty-four years (average age of onset: twenty-five) and ication costing $16. the yearly dropout rate because of side effects of medication Lithium has been reported to reduce relapses by 60 to 80 (such as tremors and gastrointestinal effects) was 5 percent. percent in the industrial world and it is projected that lithium Calculations for the intermittently ill pool of manic-depressive will prevent relapses 6 to 60 percent (Echronic) in the develop- patients use the same data, except this pool of patients will ing world (Rybakowski and others 1980). Lithium has also receive lithium for only one out of every four years. been reported to be 80 to 90 percent effective in lowering the The proportions of the disease category subtype popula- severity of illness in acutely ill patients (Eacure or Eintermittent), tion who will receive care at a health system (PT) are partially but it is projected to be only 70 percent effective in the based on an assumption that a lower percentage of the manic- developing world (Goodwin and Jamison 1990). Schizophrenia and Manic-Depressive Illness 669 Patient (H) and caretaker (CT) quality-of-life disabilities bottom of table 28-3. The same rationale applies to the manic-depressive are clinical judgments based on clinical and social rating scales. patient treatment program (table 28-4). 5. The contributions to the total for each of the last thirty-nine years As with the schizophrenia case management model, the acute were added to this year's patient pool to give an approximate number of and intermittently ill manic-depressive patient (30 percent patients in treatment in that disease outcome category. This total number of disabled) is projected to suffer from one episode lasting half a patients treated (in that disease outcome category) divided by this year's year, resulting in both one-half the disability and one-half the contribution to the patient pool (in that disease outcome category) is the mortality increase suffered by the chronically ill patient who steady-state factor. Because the average age of schizophrenia onset is twenty and the general Is 60 percent disabled the entire year (Goodwin and Jamison life expectancy is predicted to be sixty (individuals with schizophrenia gener- 1990). Correspondingly, it is projected that the caretaker's ally have a higher standardized mortaliry rate), a program of thirty-nine years' quality-of-life disability is 20 percent when he or she is charged duration is appropriate. with caring for a chronically ill patient, and 10 percent when 6. Completely (100 percent) disabled = 1.0 disability quotient; 40 percent an acutely or intertnittently ill patient is involved. disabled = 0.4 disability quotient; and so on. Research indicates that mortality rates are 2.2 times greater for manic-depressive individuals than the general population, which, if one uses the previously mentioned survival tables of References Mexican males in 1980, translates into a yearly increase in the risk of dying of 0.008 for manic-depressive individuals (M) Bagley, C. 1973. "Occupation Status and Symptoms of Depression." Social (United Nations 1982; Goodwin and Jamison 1990). Science and Medicine 7(5):327-39. The patient and caretaker's projected 60 percent reduction Baldessarim, R., B. Cohen, and M. Teicher. 1990. "Pharmacological Treat- ment." In S. Levy and P. Ninan, eds.. Schizophrenia Treatmnent. New York: in quality-of-life disability with lithium usage (R) is based on AmerILan Psychiarric Press. the impressions of clinicians which indicate that the intensity Causemann, B.. and B. Muller-Oberlinghausen. 1988. "Does Lithium Prevent of manic-depressive episodes decreases significantly. The de- Suicides and Suicide Attempts!" In N. J. Birch. ed., Lithium: Inorganic crease in the excess mortality for the lithium-treated manic- Pharmacolog' and Psychiartic Use. Oxford: IRL Press. depressive patient (R) is derived from published reports Goodwin, K., and K. Jamison. 1990. Manic-Depressive Illness. New York: showing a 60 percent decrease in suicides in manic-depressive Oxford University Press. patients who are taking lithium (Goodwin andJamison 1990). Gottesman, I.. and 1. Shields. 1982. Schizophrenia: The Epigeneac Puzzle. If the suicide rate decreases 60 percent, it is assumed that the Cambridge: Cambridge University Press. generalized excess mortality will also decrease 60 percent. Holinger, P., and E. Wolpert. 1979. "A Ten Year Follow-Up of Lithium Use." IMJ 156:99-104. Jablensky, A., and N. Sartorius. 1988. "Is Schizophrenia Universal!" Acta Notes Psychiatria Scandanavia 344(supplement):65-70. Jablensky, A., N. Sartortus, G. Emberg, M. Anker, A. Korten. J. E. Cooper, R. Day, and A. Bertelsen. 1992. "Schizophrenia: Manifestations, Incidence, An earlier version of this paper was presented at the WHO/World Bank and Course In Different Cultures: A World Health Organization Ten- consultation on interventionsfor Nervous System Disorders held in Washington, Coury Dy."ePs tColo red:cine u ea rgai1-97. D.C. on July 6-7, 1992. We wish to thank Thomas McGuire for his assistance Country Study.c PsychologicalMedicine 20(supplement):1-97. at the consultation and later, when he reviewed this chapter. Johnson, D., G. Pasterski, J. Ludlow, K. Street, and R. Taylor. 1983. "The 1. In contrast to the acutely ill patient pool, there is no need for an Discontinuance of Maintenance Neuroleptic Therapy in Chronic Schlzo- adjustment for patients who do not respond to antipsychotics, because the phrenic Patients: Drug and Social Consequences." Acta Psychiatria Scan- treatment responses of the chronic and intermittently ill groups of patients danavia 67:339-52. will be known to the clinic staff. Leff, J., N. Sartorius, A. Jablensky, A. Korten, and G. Emberg. 1992. "The 2. Incidence reported in the WHO study (jablensky and others 1992) was for Intemational Pilot StudyofSchizophrenia: Five-Year Follow-Up Findings." a population of 100,000; for the present circumstances we have adjusted that PsychologicalMedicine 22:131-45. figure to a population of 1 million. Incidence refers co individuals of fifteen Lin, T., H. Chu, H. Rin, C. Hsu, E. Yeh, and C. Chen. 1989. "Effects of through fifty-four years who experienced clearly psychotic symptoms, had Social Change on Mental Disorders in Taiwan: Observations Based on a never made contact with a helping agency in the past, and were residents of 15-Year Follow-Up Survey." Acta Psychiatna Scandanavia 348(supple- the catchment area. It was also felt that the risk for schizophrenia was ment): 1-34. negligible after age fifty-four; thus, the incidence rates are for those age fifteen McKane, J., D. Robinson, D. Wiles, R. McCreadie, and G. Stirlng. 1987. and older. "Haloperidol Decanoate v. Fluphenazine Deanoate as Maintenance Ther- 3. Eitherofthe twopoolsof acutely ill patients (respondersand nonresponders) apy in Chronic Schiz0phrenic In-Patients." Bntish Journal of Psychuarrz will not have a steady-state factor by definition. `lncidencesch-,ph,eni" is 151:333-36. total incidence and will be the same value for each disease outcome. Mannisto, P. T. 1980. "Endocrine Side-Effects of Lithium." In F. N. Johnson, "Effectiveness <,t,i..,.c, ,," is (I - Effectiveness_c,e) because it is the pool of ed., Handbook of Lithium Therapy. Baltimore: University Park Press. acute patients who are treated unsuccessfullv (the response of acute patients Over, Mead. 1991. Economics for Health Sector Analysis: Concepts and Cases. to medication is uinknown because they have never participated in the W a d. W orl eank, EcomcDvlontstitute. program). W~~~~~~~~~~~~~~~Nashington, D.C-: WVorld Bank, Economic Development Institute. program) . 4. The equations do not include an input for the number of disability- Rifkin, A., S. Doddi, K. Basawaraj, M. Borenstein, and M. Wachspress. 1991. adjusted life-years gained for the acute nonresponding patient pool by defini- "Dosage of Haloperidol for Schizophrenia." ArchiVes of General Psychiantr tion. In addition, the inputs for this pool are different from the acute respond- 48:166-70. ing pool with respect to number of visits needed (number-visits) and the cost Robins, L., and D. Regier. 1991. Psychiatric Disorder in America: The Epidemi- of medication (yearly cost-medication_Ce/non-re,) and are included at the ologic Catchment Area Study New York: Free Press. 670 Peter Cowley and Richard Jed Wyatt Roy, Alec. 1986. "Depression, Attempted Suicide, and Suicide in Patients Van Putten, T., S. Marder, and J. Mintz. 1990. "A Controlled Dose Compar- with Chronic Schizophrenia." Clinics of North America 2(): 193-205. ison of Halopenidol in Newly Admitted Schizophrenic Patients." Archives Rybakowski, J., M. Chtopocka-Wozniak, Z. Kapelski, and W. Stryzewski. of General Psychiatry 47:754-58. 1980. "The Relative Prophylactic Efficacy of Lithium against Mania and Weiss, D., K. DeWiLt, N. Kaltreider, and M. Horowitz. 1985. "A Proposed Depressive Recurrences in Bipolar Patients." International Pharraco- Method for Measuring Change Beyond Symptoms." Archives of General psychiatry 15:86-90. Psychiatry 42:703-8. Santos,J.,J.Cabranes, C. Vazquez,F. Fuentenbro, 1. Almoguera,and). Ramos. Wig, N. 1982. "Methodology of Data Collection in Field Surveys." Acta 1989. "Clinical Response and Plasma Haloperidol Levels in Chronic and Psychiania Scandanavia 296(supplement):77-86. Subchronic Schizophrenia." Biological Psychiatry 26:381-88. Wilcox, D., R. Gilian, and E. Hare. 1965. "Do Psychiatric Outpatients Take Shepard, M., D. Watt, l. Fallon, and N. Smetton. 1989. "The Natural History Their Drugs?" British Medical Joumral 2:790-92. of Schizophrenia: A Five- Year Follow-Up Study of Outcome and Prediction in a Representative Sample of Schizophrenics." Psychological Medicine 15 Wittkower, E, and H. Rin. 1965. "Transcultural Psychiatry." Archives of (supplement): 1-46. G P - Tsuang, M.,R. Woolson, and]. Fleming. 1979. "Long-TermOutcome ofMajor Wyatt, R., and J. Stevens. 1987. "Similar Incidence Worldwide of Schizophre- Psychoses: 1. Schizophrenia and Affective Disorders Compared with Psy- nia: Case Not Proven." BriashJournalofPsychiarry 151:131-32. chiatrically Symptom-Free Surgical Conditions." Archives of General Psychi- Yucun, S., A. Weixi, S. Liang, Y. Xiaoling, C. Yuhua, and Z. Dongfeng. 1981. atry 36:1295-1301. "Investigation of Mental Disorders in Beijing Suburban District." Chinese United Nations. 1982. Demogrraphic Yearbook. New York. Medical]ournal 94(3):1 53-56. PART FIVE Conclusion The Healdh Transition: Implicawo for Healdi Policy mn Dewlopmg Cowunes 29 The Health Transition: Implications for Health Policy in Developing Countries W. Henry Mosley, Jose Luis Bobadilla, and Dean T. Jamison Changing epidemiologic profiles of developing countries are interventions for communicable childhood disease led, leading-in many countries quite rapidly-to fundamental through experience, to a reasonable sense of a cost-effective changes in the volume and composition of demand for health mix of interventions. The situation becomes vastly more com- services and needs for health promotion. The purpose of this plex with the emergence, as quantitatively important, of a collection has been to attempt to take stock, in a systematic broad range of additional conditions; hence the motivation for disease-by-disease manner, of the potential for cost-effective the systematic analyses reported in this collection. responses to this changing pattern of needs. Although Chapter 1 described the approach taken in the chapter- considerations of intervention cost-effectiveness (or value- specific analyses and summarized the resulting conclusions for-money) were important even before rapid change in epide- concerning cost-effectiveness. Our purpose in this chapter is miologic profiles, the relatively limited range of key to explore, in a more general way, the implications of the Figure 29-1. Relationships among Demographic, Epidemiologic, and Health Transitions Health transition Demographic transition Epidemiologic transition Infectious Chronic and Urbanization disease Fertility Population noncom municable mortality declines ages diseases Industrialization declines emerge Rising incomes Expansion of education Improved medical and public health Economic Persistence or technology - recession and reemergence of increasing communicable inequality diseases Protracted-polarized epidemiologic transition Source: Authors. 673 674 W. Henry Mosley, Jose Luis Bobadilla, and Dean T. Jamison epidemiologic and health transition for health policy. We Figure 29-2. Regional Projections of Life begin by reviewing the global health transition and its constit- Expectancy and Fertility uent demographic and epidemiologic transitions. We then turn to discussion of the implications of these transitions for Total fertility rate (Births per woman) national governments and, in closing, we explore implications 7 for intemational aid. The Health Transition 6 - Essential to an understanding of the evolution of disease con- trol priorities in developing countries is a reasonable projec- 5 - tion of probable changes in the pattern of disease. These changes are likely to be profound. Our discussion of these changes divides naturally into four parts-the first deals with 4 - the demographic transition, the second deals with the epide- miologic transition, the third deals with the changing risk environment that has been occurring, and the fourth deals 3 - with the widening gap in health problems and health needs across social and economic classes. Collectively these changes ' . are coming to be referred to as the "health transition," and 2 _ figure 29-1 illustrates relations among the demographic, epi- demiologic, and health transitions. *1985 02000 02015 1_ The Demographic Transition and Population Aging Health patterns in the developing world during the next three 0 I I I l l decades will be profoundly influenced by recent and projected 50 60 70 80 future declines in fertility and mortality as these nations pass Life expectancy (years) through the demographic transition. Figure 29-2, which is drawn from the demographic analyses prepared for this collec- tion, projects declines in the total fertility rates and the gains Middle East/North Africa - S- - b- Asia in life expectancy that might be expected in each of four regions of the developing world during the thirty years 1985 to 2015, assuming reasonable and achievable continuation of estab- Source: Bulatao and Stephens 1990. lished trends. (Table 29-1 provides detail by region on the demographic parameters estimated for 1985-90 and projected It is commonly assumed that the changing health picture for 2000-2005; definitions of the regional groupings that are seen in populations undergoing the demographic transition is used may be found in chapter 1, table A-1.) The projected primarily a function of the declines in mortality. In fact, declines in fertility for Sub-Saharan Africa and the Middle East however, the age structure and, correspondingly, the cause are substantial, averaging 50 percent, whereas the gains pro- structure of death during the course of the demographic tran- jected for life expectancy are more modest, ranging from 10 sition is strongly influenced by the rapid decline in fertility. percent in Latin America to 25 percent in Sub-Saharan Africa. (The role of mortality decline in creating preconditions for Long-term projections are inevitably tentative; nonethe- fertility decline nonetheless leaves mortality decline as a cen- less, it should be noted that fertility changes of this magnitude tral indirect cause of epidemiologic change. Figure 29-1 in a thirty-year period are not unprecedented. The total fertil- illustrates this point.) This occurs because of a phenomenon ity rates in the Latin American and Asian regions ranged from that is described by demographers as the "momentum" of 5.5 to 6.0 births per woman in the late 1950s and declined to population growth. To explain simply, with high fertility the their present levels of 3.3 to 3.5 in less than thirty years. age structure of a population is highly skewed toward the Perhaps more problematical are the projected mortality de- young, irrespective of the level of mortality (figure 29-3). With clines. These do not yet take into account the acquired im- sustained high birth rates and larger numbers of women enter- munodeficiency syndrome (AIDS) epidemic, which has ing the reproductive ages every year, the base of the population assumed significant proportions in many countries in Sub- is continually expanding as more births are added every year. Saharan Africa and Latin America. Nonetheless, much of With the onset of the fertility transition and rapidly declining the developing world is now well through a transition from birth rates, however, the number of births added each year may high mortality and fertility rates to low ones; this demo- remain unchanged or even decline. Consequently the age graphic transition sets the stage for epidemiologic change. structure of the population will be progressively transformed Table 29-1. Demographic Parameters, Globally and by Region, 1985-1990 (Estimates) and 2000-2005 (Projections) Population Crude birth rate Crude death rate Life expectancy at birth (millions) (per 1,000 population per year) (per 1 ,000 population per year) Total fertility ratet (ears)' Region' 1985 2000 1985-90 2000-5 1985-90 2000-5 1985-90 2000-5 1985-90 2000-5 Industrializedmarketeconomies 760 810 13 12 9 10 1.7 1.8 76 78 Industrialized transition economies 416 453 17 15 11 10 2.3 2.1 70 73 Subtotal, industrialized economies 1,176 1,263 15 13 10 10 1.9 1.9 74 76 Latin American and Carribean 402 529 29 21 7 6 3.6 2.5 67 71 Sub-Saharan Africa 556 720 46 40 15 11 6.4 5.4 52 57 Middle East and North Africa 376 573 40 32 10 8 5.6 4.3 60 65 AsiaandthePacific 2434 3,118 27 21 9 8 3.3 2.6 64 68 Subtotal, developing countries 3,668 4,940 31 25 10 8 3.9 3.1 62 66 World total 4,844 6,203 27 23 10 8 3.4 2.9 65 68 a. Appendix 29A lists countries in each grotiping. b. Number of children a woman would be expected to bear during her reproductive years, based on the age-specific fertility rates prevailing in that period. c. Average of the male and female life expectancies reported in the source table. Source: Bulatao and Stephens 1990. 676 W. Henry Mosley, Jose Luis Bobadila, and Dean T. Jamison Figure 29-3. Age Distribution of the Population (figures based on altemative mortality and fertility assumptions) Percent 25 - A 20 - 15 00 5 0 b. oc Age group Note: Percent distribution of the female population by age groups in West model life tables. Curves A and B represent a gross reproduction rate of 4.0 (eight births per woman) with life expectancies of seventy five years (curve A and forty years (curve B). Curves C and D represent a gross reproduction rate of 1.5 (three births per woman) with life expectancies of seventy-five year (curve C and forty years (curve D). Source: Coale and Demeny 1983. from the shape of a broad-based triangle to a rectangular or cause structure that has been termed the "epidemiologic tran- even trapezoidal shape with a narrowing of the base (figure sition" by Omran (1971). Omran identifies three phases in this 29-3). The pace of fertility decline will be directly reflected in transition: the age of pestilence and famine, the age of receding an immediate slowing (and even reversal) in the growth of the epidemics, and the age of degenerative and man-made diseases. youngest age groups. The adult population will, however, In table l-I (and table 29-7), the "unfinished agenda" and the continue to grow for several decades because of the continuing "emerging problems" illustrate the health conditions that are aging of the larger cohorts of persons already born. typically prominent, for children and for adults, in the pre- and Figure 29-4 illustrates this phenomenon for Latin America. postepidemiologic transition environment, respectively. (We Although the size of the age cohort under five years old consider phase three to be the posttransition phase.) changes very little during the thirty-year period, there is a Olshansky and Ault (1986) proposed a fourth phase in the dramatic increase from ages forty-five through sixty-four. In epidemiologic transition-the age of "delayed degenerative the very long run (more than a century) the numbers of the diseases." This phase was proposed because of the progressive elderly in the rapidly growing developing countries can in- decline in the death rates from some chronic diseases associ- crease insize bymore than 100times (Chesnais 1990). Kinsella ated with steady gains in life expectancy among the aged in (1988) examines a broad range of consequences, in addition the United States and some other industrial countries. Crim- to the health ones we address here, of population aging in mins, Saito, and Ingegneri (1989) have reported for the United developing countries during the next several decades. States that these gains in survival among the aged have in large measure been in "disabled" years rather than "healthy" life. In The Epidemiologic Transition this circumstance, improved survival among the aged implies that there will be an increasing, not lessening, demand for The transformation in the age structure of mortality associated health services (Verbrugge 1984, 1989). Fries (1989), how- with the demographic transition leads to a transition in its ever, notes that in the United States in recent years there have The Health Transition: Implicatons for Health Pobcy in Developing Countries 677 Figure 29-4. Estimates of Male Population of Latin during the thirty-year period, reflecting the momentum of America, Age Groups 0-4 and 45-64, 1970-2015 population growth that follows the historical pattems of high fertility. Population (in millions) At the other extreme, Sub-Saharan Africa, which experi- 70 enced no significant fertility declines prior to 1985, will con- tinue to show large increases in the population down to age 60 - fifteen and then smaller increases in the younger ages, reflect- so - ing the much later (probable) onset of a fertility decline. In the Middle East and North Africa, a somewhat earlier onset of 40 - fertility decline is projected. Again, the large increases in the older age groups in these regions reflect the momentum of 30 - population growth. It should be clear from table 29-2 that, even if there were no 20 _ change in age-specific morbidity and mortality rates, projected declines in fertility would have a significant effect on the age 10 structure and, therefore, on the relative frequency of different _ _ _ _ causes of death simply because the population is aging. For 0 _ _ - example, in Latin America, all things being equal, we could 1970 1985 2000 2015 expect that this change in age structure would be accompanied by more than a doubling of chronic disease among adults in Age 0-4 Age 45-64 relation to acute diseases among infants and children. In fact, however, mortality rates from these conditions probably will Source: Bulatao and Stephens 1990. decline. As shown in table 29-2 these projected mortality declines are greatest (decreases of 60 to 70 percent) in the been substantial declines in the incidence of such conditions youngest age groups and least (decreases of 7 to 18 percent) in as heart disease, lung cancer, and automobile accidents. He the oldest age groups. observes that "successful aging" with lessened infirmity can be Interactions of changes in mortality with the changes in age achieved if medical systems pursue vigorously the path of structure will result in an even more drastic transformation of prevention rather than concentrating on developing sophisti- the health picture. This is also shown in table 29-2, which cated means of treating diseases after they are recognized. gives the percentage changes in numbers of deaths that are Because of the central role of population dynamics in shap- projected to occur within each age group by the year 2015. ing the profile of illness and the pattern of cause of death, it is Again taking Latin America as an example, among children worth discussing these matters somewhat further. Preston under five projections show only a 2 percent increase in the (1986) has shown that from the time a population comes down population size but a 62 percent decline in the age-specific to and maintains replacement reproduction levels (total fertil- mortality rate, resulting in a 61 percent decline in number of ity rates approximately 2.1) the entire growth of the popula- deaths. By contrast, in the oldest age group the projections tion occurs only in the population segment beyond the mean show a 141 percent increase in the population size but only a age of childbearing (approximately twenty-eight years). He 12 percent decline in mortality, resulting in more than a concludes: "Population momentum turns out to be momentous doubling of the number of deaths and an even greater increase only for mature adult ages where productivity is typically in chronic disability. This epidemiologic transition will have highest and where concerns regarding the economic effects of important consequences for the organization and delivery of rapid growth are probably least" (p. 349). The implications of health services in the future. population momentum in older age groups for the health system are, however, dramatic. Table 29-2, which is derived Changing Patterns of Risk from World Bank projections, shows the percentage changes in the population size by age group projected to occur in each In addition to changes in population age structure, which is of the four developing country regions over the thirty-year the primary determinant of epidemiologic transition, there are period 1985 to 2015.1 This illustrates the marked changes in global social and economic trends which are transforming the the age structure of these populations that will result, primarily risk factors for different diseases (Kjellstrom and Rosenstock as a consequence of dramatic declines in fertility. In the cases 1990). The most obvious global shift is from rural to urban of Asia and Latin America, where fertility declines have been living. In 1985, only 31 percent of the population of the well under way for the last twenty years, there would be very developing regions of the world resided in urban areas, in little change in the size of the populations under age fifteen in comparison with 72 percent in the industrial regions (United the next thirty years; by contrast, the populations over age Nations 1989). But the urban growth rate in the developing forty-five would increase by over 130 percent. These increases regions is projected at 3.6 percent per year through the end of would represent growth rates of 2.8 to 3.2 percent per year the century, so its urban population will reach 40 percent by 678 W. Henry Mosley, Jose Luis Bobadilla, and Dean T. Jamison Table 29-2. Projected Change in Population, Mortality Rates, and Deaths between 1985 and 2015, by Age (Group (percent) Age group Region 0-4 5-14 15-44 45-64 65+ All ages Asia Population 5 7 53 131 134 51 Mortality rates -70 -58 -43 -30 -7 -18 Deaths -68 -55 -13 62 118 23 Latin American and Caribbean Population 2 18 57 159 141 63 Mortality rates -62 -60 -47 -29 -12 -17 Deaths -61 -53 -21 84 112 35 MLddle East and North Africa Population 38 70 127 175 150 102 Mortality rates -64 -65 -57 -36 -18 -43 Deaths -50 -44 -2 76 105 15 Sub-Saharan Africa Population 70 116 163 151 161 132 Mortality rates -64 -64 -53 -29 -11 -45 Deaths -40 -22 24 79 132 28 Source: Calculated from Bulatao and Stephens 1990. the year 2000 and 50 percent by 2015. This rapid shift from a drivebelts resulted in an increased number of serious injuries. rural subsistence economy to an urban, market-oriented, in- Injuries are particularly a problem in many poor countries, dustrial economy brings with it a range of new health problems which lack the resources and institutions to establish and (Susser 1981). At the same time, economic growth brings with enforce safety measures. Noteworthy, throughout the world in it the wherewithal and knowledge for populations to acquire developing and industrial countries alike, injuries are now the the nourishment and sanitation that can reduce the incidence leading cause of death during half the human life span (Stans- of and fatality rates from communicable disease. Reductions in field, Smith, and McGreevey, chapter 25). risk for communicable disease combined with increases in Chronic conditions such as cardiovascular disease, cancer, other risks, further discussed below, have the potential to and chronic obstructive pulmonary disease are also recognized amplify the effects of demographic trends. An important gen- to be substantially influenced by economic and environmental eral conclusion from a recent study of adult health in develop- factors, some of which are amenable to modification by the ing countries by Feachem and others (1992), however, is that health system in ways that are reviewed in the chapters in part the overall effect of development on age-, sex-, and cause- 4 of this collection. For example, the U.S. Department of specific mortality rates for noncommunicable disease is to Health, Education, and Welfare identified lifestyle and the lower them, despite the often-increasing prevalence of well- environment as the primary determinants of mortality for all established risk factors of modern society. but one of the ten leading causes of death over age one in the High rates of injuries related to motor vehicles, industrial United States in 1975 (USDHEW 1978). A more recent analysis accidents, and toxic chemicals (for example, pesticides) are indicated that just three preventable precursors to premature one consequence of rapid urbanization, industrialization, and death in the United States-alcohol, tobacco, and injury mechanization of agriculture. Stansfield, Smith, and risks-accounted for 59 percent of all preventable years of life McGreevey (chapter 25) provide extensive documentation of lost before age sixty-five and 54 percent of all preventable days the dramatic increase of these categories of injuries in devel- of hospital care (Amler and Eddins 1987). oping countries. For example, in Thailand in the age group of Smoking provides an excellent illustration of an emerging one to forty-four years, motor vehicle mortality has been health problem (Zaridze and Peto 1986; Stanley, appendix A). increasing at 30 percent annually, moving from sixth place to In the United States it is now well established that tobacco use first place among all causes of death between 1947 and 1980. is currently responsible for more than 30 percent of all cancer In 1978, mortality rates per vehicle were fifty times higher in deaths, including cancers of the lung, larynx, oral cavity, Ethiopia and Nigeria than in the United States or the United pharynx, pancreas, kidney, and bladder (Ernster 1988). The Kingdom. Pesticides in Sri Lanka in 1978 caused almost twice most dramatic evidence of this is the tenfold increase in lung as many deaths as occurred as a consequence of polio, diphthe- cancer mortality in the United States since the turn of this ria, tetanus, and pertussis combined. In India, mechanization century, with differential trends in males and females reflecting of grain mills without appropriate protective shields over differences in smoking habits (Lopez, chapter 2). Similar dra- The Health Transition: Implications for Health Policy in Developing Countries 679 matic increases in lung cancer mortality associated with ciga- d iet with over 30 percent energy from fat, whereas close to half rette consumption have been observed in Japan since World (47.8 percent) of the lowest tercile consumed a diet with less War 11, and rising rates have also been reported from Singapore than 10 percent of energy from fat (Popkin and others 1992). and Shanghai in recent years (Lee and others 1988; Barnum An increase in the rate of ischemic heart disease is already and Greenberg, chapter 21). In some Latin American and being seen in some developing countries as they proceed Caribbean cities more than half of the young people smoke; by through the epidemiologic transition. Singapore has experi- the mid-1980s it is estimated that at least 100,000 deaths in enced a doubling of the heart disease mortality rates during the this region were caused by smoking (USDHHS 1992). The rapid past two decades in older age groups, although there appears increase in smoking in China has the potential of leading to to be a leveling off in the increase among younger males in actual increases in age-specific mortality rates, which would recent years (Hughes 1986). As shown by the experience of run counter to standard demographic assumptions as reflected, Japan, however, which has a low rate of ischemic heart disease for example, in table 29-1 (Yu and others 1990). Bumgarner (although not stroke), economic development need not be and Speizer, chapter 24) have illustrated the plausibility of associated with the disease patterns seen in most Western this outcome with quantitative projection models, and Lopez populations. Developing countries generally have fewer risk (chapter 2) summarizes analogous predictions for other parts factorsforsomeofthediseasesassociatedwithWesternculture of the world. (Rose 1985). The course of these chronic diseases in the future The relation between cigarette smoking and lung cancer will depend on the choices made by developing countries as illustrates a vitally important feature of chronic noncommuni- they consider alternative health development strategies while cable diseases-the long latent period between exposure and proceeding through the epidemiologic transition. onset of the disease. Smoking is also one of the strongest risk The onset of the global AlDs epidemic has brought sexually factors for chronic obstructive pulmonary disease (COPD) and transmitted diseases (STDs) to the forefront of the health ischemic heart disease (Bumgarner and Speizer, chapter 24, agenda of many developing countries. The risk factors for STDS and Pearson, Jamison, and Trejo-Gutierrez, chapter 23); as are directly related to patterns of sexual behavior. These, in with cancer, the latency of effect is long for COPD and irrevers- turn, are often related to the development process. For exam- ible; but with ischemic heart disease, cessation confers substan- ple, in many developing countries, factors contributing to high tial reduction in risk within a year. These data highlight the rates of STDs include increasing urbanization with disruption of importance of taking action to prevent chronic diseases de- traditional social structures, increased mobility for political or cades before the epidemic appears. Regrettably, a recent review economic reasons, poor medical facilities, and high unemploy- (Masironi and Rothwell 1988) found that the rate of tobacco ment rates (Piot and Holmes 1989). Over and Piot (chap- consumption in developing countries is increasing; consistent ter 20) present data which show that the high rate of human with this, Barnum and Greenberg (chapter 21) have found a immunodeficiency virus (HIV) infection in eighteen African strong relationship between tobacco consumption and higher cities can be correlated with a low ratio of females to males in levels of national income among developing countries. urban centers, creating a high demand for prostitutes. An Ischemic heart disease and stroke, other major causes of associated factor of significance is the relatively low level of death among adults, are potentially amenable to early preven- female education, which suggests that where there are fewer tive interventions. Although all the determinants of ischemic alternative economic opportunities for women, prostitution is heart disease remain to be defined, and patterns of attributable more frequent. Reducing the risk of HIV in these circumstances risk will certainly differ in developing countries from those of will probably require significant social changes relating to the the industrial countries, where epidemiologic data are avail- role and status of women (including increasing female educa- able, several behavioral risk factors are well established. These tion) as well as promoting the use of condoms and treating include smoking, sedentary lifestyle, and high saturated fat coexisting STDs. diets (Pearson, Jamison, and Trejo-Gutierrez, chapter 23). In general these behaviors are strongly associated with urban- Epidemiologic Polarization ization in low-income countries. For example, Popkin and Bisgrove (1988, p. 9) reported that "urban residents consume During the next three decades the most dramatic declines in increased amounts of processed foods, meats, fats, sugar and mortality in the developing regions of the world are projected dairy products while rural residents consume more coarse for the infectious and parasitic diseases that primarily affect grains, roots and tubers and pulses." infants and young children; relatively modest changes are China provides a remarkable example of the transition to projected for the death rates for conditions such as cardiovas- unhealthy diets. The rapid increase in income and related cular disease, cancer, and other chronic diseases.2 As a result, social and agricultural advances has led to a rapid increase in as tables 29-3 and 29-4 suggest, these chronic, noncommuni- the proportion of obese Chinese and the proportion of Chinese cable diseases of adults will rapidly emerge as the leading causes consuming a very high-fat diet, although a large proportion of death in developing countries. Table 29-3 shows estimates still have a very low-fat diet and are exceptionally lean. For and projections of mortality by broad category of cause based instance, a nationwide survey found that close to one-fifth on model life tables; table 29-4 shows, for 1985 only, the best (18.3 percent) of the highest tercile of Chinese consumed a available empirical estimates, using vital statistics and epide- 680 W. Henry Mosley, Jose Luis Bobadilla, and Dean T. Jamison Table 29-3. Major Causes of Death in Industrial and Developing Countries, 1985 and 2015 (percent) Industrial countries Developing countries Cause of death 1985 2015 1985 2015 Infection 9 7 36 19 Neoplasms 18 18 7 14 Circulatory problems 50 53 19 35 Pregnancy-related deaths 0 0 1 1 Perinatal problems 1 1 8 5 Injuries 6 5 8 7 Other 15 16 21 19 Total number of deaths (millions) 12.0 14.5 37.9 47.8 Note: These estimates (1985) and projections (2015) are based on assumptions about changes in total mortality rates built into the World Bank's demographic projections model and on historically based asuumptions about the relationship between mortality by cause and mortality level. Countries included in the "Industrialized" and "Developing" categories are found in appendix 29A. Source: Bulatao and Stephens 1990. miologic data. Both tables, and the preceding discussion, point decay of the health advances that had been achieved in recent toward the same conclusion: a health transition of massive decades in some countries, which is often reflected in a rising proportions is well under way in the developing world, and it occurrence of childhood malnutrition (Albanez and others will continue for several more decades at least. 1989; Cornia, Jolly, and Stewart 1987; and Bell and Reich A critically important feature of the health transition is the 1988). These setbacks, combined with a wide disparity in emergence of epidemiologic polarization within and between health conditions of different social classes, have been charac- countries of the developing world. In recent years economic terized by Bobadilla, Frenk, and their colleagues (Frenk and growth in the developing world has not been steady. The others 1989; Bobadilla and others, chapter 3) as "epidemio- worldwide recession, poor economic management, and the logic polarization." excessive accumulation of debt have led to serious setbacks in Available evidence suggests that these setbacks in prog- the economic circumstances of many developing countries. ress and continuation of polarization rarely result in a rever- One potential consequence has been the stagnation or even sal in the pace of mortality decline (Hill and Pebley 1990), Table 29-4. Deaths by Cause, Industrial and Developing Countries, 1985 Industrial countries Developing countries Cause of death Number (000) Percent Number (000) Percent Infectious and parasitic diseases 506 4.6 17,000 45.0 Diarrheal diseases - - 5,000 13.0 Tuberculosis 40 0.4 3,000 7.9 Acute respiratory illness 368 3.3 6,300 16.6 Measles, pertusis, diphtheria - - 1,500 4.0 Other 4,800 12.7 Other measles and pertusis .. .. 700 1.8 Malaria .. .. 1,000 2.6 Schistosomiasis .. .. 200 0.5 Other .. .. 800 2.1 Matemal causes 5 0.05 500 1.3 Perinatal causes 100 0.9 3,200 8.4 Cancers 2,293 20.8 2,500 6.6 Chronic obstructive pulmonary disease 385 3.5 2,300 6.1 Circulatory and certain degenerative diseases 5,930 53.7 6,500 17.1 Ischemic heart disease 2,392 21.7 Cerebrovascular disease 1,504 13.6 Diabetes 153 1.4 - - Extemal causes (injuries) 772 7.0 2,400 6.3 Other and unknown 1,054 9.5 3,500 9.2 Total 11,045 100.0 37,900 100.0 - Not available. .. Negligible. Source: Lopez (chapter 2, chis collection). The Health Transition: Implications for Health PolUcy in Developing Countries 681 even for Africa (Feachem, Jamison, and Bos 1991) . This is best Table 29-5. Probability of Dying by Age Five, by documented in the recent report of the demographic and Country, 1980-85 health surveys (DHS) in twenty-seven countries in Africa, Probability of Asia, and Latin America, carried out between 1986 and dying by age five 1990 (Sullivan 1991).3 They showed declines in mortality (percent) Latin America Asia Afrca of children under age five in every country; by region the 25-30 none none Mali average percentage declines were: North Africa, 46 percent; Latin America, 32 percent; Asia, 28 percent; and Sub- 20-25 none Bangladesh Liberia Saharan Africa, 12 percent. Although effective health Senegal interventions may have blunted the potential mortality 15-20 Haiti India Ghana consequences of economic stagnation, evidence is emerging Uganda from several countries that low child mortality levels can 10-15 Guatemala Turkey Egypt now be maintained even in the presence of sustained high Peru levels of malnutrition and morbidity (concerning Sri Lanka, 5-10 Brazil Philippines Botswana see Gunatilleke 1989; concerning Zimbabwe, see Sanders Dominican Thailand and Davis 1988). It is important that the persistence of these Republic undesirable states, as is emphasized in chapter 3, not be Ecuador masked by undue focus on (relatively) favorable mortality El Salvador statistics. Mexico Table 29-5 illustrates the wide disparity in levels of child 2-5-5 Argentina Kuwait none mortality (under five years) seen among developing countries Chile Malaysia in every region of the world in the 1980s. Mortality levels range Colombia Sri Lanka from under 2.5 percent in Costa Rica and Cuba to over 20 Panama percent in Bangladesh and Mali. Within countries as well, Uruguay significant disparities in health conditions are found among subgroups of the population. Some of the most recent estimates <2.5 Costa Rica Hong Kong of these conditions are from the recent DHS (Rutstein 1992). Cuba Singapore Tabulations of the levels and differentials in mortality of Source: Hill and Pebley 1990; updated by Hill, personal communication. children under age five, when grouped according to urban as against rural residence and the mother's level of education, are presented in table 29-6. In general the data indicate child countries. Where data are available, the patterns are similar to mortality rates 30 percent to 50 percent lower in urban than those well documented in the industrial countries; mortality in rural areas, and a two- to threefold difference between rates for most chronic diseases among adults are higher among women with no education and those with seven or more years the lower social classes than among the upper classes (Kaplan of education. and others 1987; Feachem and others 1992; World Bank The relation between matemal education and child survival 1989). For example, in a study of mortality in the rich and poor in developing countries has been observed in multiple studies areas of Porto Alegre, Brazil, Barcellos and others (1986b, during the past decade (Caldwell and McDonald, 1981; p. 206) found that death rates for men between forty-five and Cochrane, Leslie, and O'Hara 1982; Hobcraft, McDonald, and sixty-four were 50 percent higher among the poor, with death Rutstein 1984) and has led some demographers to observe that rates for cancers, cardiovascular diseases, respiratory diseases, what counts in child survival is not just the overall health and and injuries all higher among men living in poor neighbor- socioeconomic condition of the country where one resides but hoods. The reasons, as in the industrial countries, relate to a the individual's (or family's) social and economic resources. high-risk lifestyle that includes alcohol consumption, smok- The urban-rural mortality differentials in table 29-6 provide ing, lack of exercise, and obesity, as well as poor living and one indicator of the disparity among families in different working conditions. settings. In this context, because social and economic devel- In many countries of the world, particularly across Asia, opment usually does not occur uniformly throughout all areas women experience excess mortality as compared with men of a country, one will frequently see important differentials in because of their marginalized position in society (Das Gupta mortality rates in different geographic regions within coun- 1987). These excesses are most evident in the higher rates of tries. Examples include Mexico, Brazil, Kenya, Nigeria, India, infant and childhood mortality among females. Another re- and Indonesia. Thus, epidemiologic polarization occurs not flection of the disadvantaged position of women is the extraor- only across social classes but in regional mortality differentials dinarily high rate of preventable maternal mortality in many as well. developing countries, which is 100 to 500 times higher than The analysis above is limited to infant and childhood mor- in the industrial countries (Walsh and others, chapter 17). tality primarily because there are relatively few data on mor- Among the surviving women, studies in many parts of the tality differentials by social class among adults in developing world have documented a higher prevalence of stunting and 682 W. Henry Mosley, Jose Luis Bobadilla, and Dean T. Jamison Table 29-6. Mortality Rates of Children Younger than Five Years, by Residence and Mother's Education, Selected Countries, 1986-90 (deaths under age 5 per 1,000 births) Type of residence Mother's education Country Rural Urban None 1-3 years 4-6 years 7 or more years Africa Egypt 164 88 161 116 108 48 Morocco 137 81 125 76 66 50 Tunisia 88 62 84 71 58 39 Mali 303 203 290 244 214 112 Liberia 239 216 242 * * 177 Senegal 250 135 226 179 123 75 Uganda 191 164 195 222 173 144 Togo 169 131 170 163 131 89 Ghana 163 131 175 119 169 125 Burundi 186 163 191 167 141 90 Kenya 91 89 109 101 92 70 Zimbabwe 98 55 120 94 83 60 Botswana 56 57 62 52 47 53 Asia Indonesia 124 78 144 139 99 48 Thailand 52 35 76 88 47 19 Sri Lanka 43 40 72 47 41 35 Latn America Bolivia 168 114 180 166 141 70 Peru 153 74 169 147 113 53 Guatemala 130 99 136 122 84 44 Brazil 121 88 136 137 70 40 Mexico 104 59 112 91 54 29 Ecuadlor 99a 63a 159 121 74 49 Dominican Republic 66a 69d 136 100 96 66 Paraguay 47 43 65c SI 41 27c Colombia 32 35 74 - - 26 -Not available. * Less than 500 children exposed. a. Based on last five years. b. Calculated from urban breakdown. c. Fducation categories are 0-2 years, 3-5 years, primary complete, and secondary or higher. d. Secondary onlyv Source. RuLIstein 1991. micronutrient deficiency (Leslie 1991). The enormity of this 2.5 percent, a level 50 to 200 times greater than in the problem has recently been shown by Coale (1991), who ana- industrial countries. The estimates of Murray, Styblo, and lyzed the estimated deficits in the female population of several Rouillon (chapter 11) indicate that this will result in approx- Asian countries; the deficits were derived from a comparison imately 7.3 million new cases and 2.7 million deaths in 1990. of the actual ratio of males to females with the expected ratio More than two-thirds of these deaths will be among produc- if there were no excess female mortality. For China and India, tive adults (ages fifteen through fifty-nine), primarily the from 5.3 to 5.6 percent of females were "missing," indicative poor. Significantly, this disease alone accounts for about 26 of a deficit of 52 million in these countries. percent of an estimated 7 million avoidable adult deaths in Although noncommunicable diseases and injury will be- the developing world. Assuming no change in the present come more prominent with the epidemiologic transition, the trends of decline and no improvement in case detection and infectious diseases, malnutrition, and excess (unwanted) fer- treatments, Murray, Styblo, and Rouillon project as many tility cannot be ignored. These will, however, become even as 2.9 million tuberculosis deaths still occurring by 2015. more concentrated among the poor, leading to the phenome- Because of the contribution of the HiV epidemic to tubercu- non of epidemiologic polarization. Tuberculosis is illustrative losis, coupled with the rate of population growth in Sub- of a leading disease that remains on the unfinished agenda Saharan Africa, the number of deaths projected could of developing country health problems. In most developing increase by more than 100 percent in that region during the countries, the annual risk of infection ranges now from 0.5 to next twenty-five years. The Health Transition: Implications for Health Policy in Developing Countries 683 Consequences for the Health System cost-effectiveness is an essential first step to meeting that challenge. A central consequence of the health transition for health The composite effect of the demographic transition and the policy is that in most developing countries, pre- and post- socioeconomic changes on the health system that are foresee- epidemiologic transition problems will coexist. Omran (1971) able for the next thirty-five years-mainly urbanization and predicted evolution toward this state of epidemiologic diver- higher levels of education-will be formidable. Four main sity in his original essay on the epidemiologic transition; Evans, effects are highlighted as they apply to most of the developing Hall, and Warford (1981) and Hiroshi Nakajima (World countries. Health Organization, 1988) further described the trend; and First, the total burden of disease, measured by the number the authors of several World Bank country-specific analyses of days that people suffer from acute episodes of disease, (Jamison and others 1984; World Bank 1989 and 1990a) have chronic disabilities, and days lost as a result of premature death, attempted to draw appropriate implications for policy. In will increase. This is not only because the population will table 29-7 we attempt to summarize, for each of the age cate- continue to grow but also because the prevalence of disease gories we are using, which health problems on the current will increase as more chronic diseases predominate in the agenda need continued attention and which neglected or health profile. The adult population suffers more diseases emerging problems are likely to require substantial increases in simultaneously and these tend to last longer, as compared with effort. The latter are considered in the light of the indicated child morbidity. In addition the emergence of new health risks change in the age distribution of mortality. As Foege and as described in the previous section will lead to higher rates of Henderson (1986, p. 321) have observed, these countries "will incidence of some conditions, particularly lung and breast not have the luxury of dealing with two kinds of problems cancer, some accidents and violence, and AIDS. sequentially. For the remainder of this century they will be Second, the demand for health services will be greater. dealing with both simultaneously." Health systems of develop- Demand is a direct function of three factors that tend to move ing countries will, then, be facing unprecedented increases in in the same direction: (a) health needs that were described the volume and diversity of problems they must address; the before; (b) the threshold for converting need into demand, challenge is to respond with maximal effectiveness, given the which will decrease as a result of the higher levels of income sharp constraints on their resources. Assessing intervention and education of populations and the accessibility to informa- Table 29,7. Health Problems Affecting Various Age Groups in Developing Countries Population (millions) Deaths (millions) Important health problems Age group 1985 2015 1985 2015 Unfinished agenda Emerging problems Young children 490 626 14.6 7.5 Acute respiratory infection Injury (0-4 vears) Diarrheal disease Leaming disability Leaming disability Malaria Measles, tetanus, polio Micronutrient deficiencies Protein-energy malnutrition School-agechildren 885 1,196 1.6 1.3 Geohelminthinfection Leamingdisability (5-14 years) Micronutrient deficiencies Schistosomiasis Young adults 1,667 2,918 5.0 6.0 Excessfertility AIDS ( 15-44 years) Malaria Injury Matemal mortality Mental illness Tuberculosis Sexually transmitted diseases Middle-aged 474 1,131 5.9 10.4 None Cancers (45-64 years) Cardiovascular disease Chronic obstructive pulmonary disease Diabetes Elderly 153 358 11.0 22.5 None Cataracts (65+ years) Depression Disability Total 3,669 6,229 37.9 47.7 Note: Many conditions for older age groups manifest themselves clinically iong after the processes leading to the clinical condition have been initiated; preventive intervention wiil, therefore, need to be directed to vounger populations. Source: Figures for population and deaths calculated from Bulatao and Stephens 1990. 684 W. Henry Mosley, Jose Luis BobadiUa, and Dean T. Jamison tion acquired in urban areas through radio and other mass associated with modifying the incentives and knowledge of media; (c) finally, the supply of services, particularly those patients and providers. These are listed as the instruments of provided by hospitals, that is, the increased proportion of policy in table 1-2; these instruments are further discussed populations living in urban areas will improve the physical below. access to health facilities and therefore will boost the demand a . for services. * ProvidIng informtiouln. Fundamental to any improve- Third, the emergence of noncommunicable diseases and ments in health behavior among the population are infor- Third the mergece ofnoncomunicble dsease and mation, knowledge, and skills, ideally reinforced with social disabilities due to injury will increase considerably the com- spt.on ree, ars, tseave begun to se the plexity of the health care services required. In general, health media and modemn communication technologies to reach personnel will require higher qualifications and probably some the public with information to promote good health behav- level of specialization. The technology for diagnosis, treat- iors through programs of "information, education and com- ment, and rehabilitation will be more sophisticated, and the munication." Often this is done effectively in partnership organizational arrangements to ensure minimum standards of with Ofte th aps the most n pamples care will also increase in complexity. with the private sector. Perhaps the most notable examples careth,ll allstheopreviouslydescribedeffec increase incomplof this in a number of developing countries are mass mobi- Fourth, all the previously described effects will increase lization efforts in support of immunization campaigns, as expenditure for health care. The one that has the greatest well as communication programs to improve maternal relevance is probably the higher cost of medical care that will weaning practices and promote the practice of family plan- result from the greater complexity of services, particularly the ning. In the United States, recent publication of selected introduction of new health technologies. The greatest effect operative mortality and success rates, by hospital, has al- of the health transition is likely to be seen in hospitals (Barnum lowed more informed consumer choice to stimulate quality and Kutzin 1993). Most developing countries provide hospital assessme andorole i hospitole i services for only a fraction of the population. The demand for assessment and control i hospitals. services is already greater than the supply. The shortage of * Regulation/Legislation. Health ministries generally have hospital beds will, according to the effects described above, be considerable regulatory powers, for example, in licensure of exacerbated. Three primary causes for hospital admission are practitioners, and in food and drug control and sanitation, likely to grow: childbirth, noncommunicable diseases, and though resources for inspection and enforcement are often injuries. limited. A central regulatory power of govemments lies in the determination of which health services will (can) be Policy Implications for National Governments privately provided (through market mechanisms or through nongovernmental organizations) and which will be pro- Chapter 1 assembled the cost-effectiveness findings from vided by the state. When coupled with effective public each of the disease-oriented chapters. These findings were education to reach a social consensus, regulatory authority grouped as population-based interventions and facility- can be an effective tool for health promotion, as evidenced based interventions as described in table 1-2. Notably, this by the ability of some governments to limit pollution levels analysis does not weigh preventive as opposed to curative and to restrict the advertising ofcigarettes or the protnotion strategies but, rather, considers the cost-effectiveness of the of infant formula and baby bottles. full range of interventions-primary prevention, secondary a Taxes, Subsidies. Taxes or price subsidies can be an prevention, curative, rehabilitation, and palliation-on the important tool available to governments to promote or same scale. Our purpose in this section is to explore the discourage various practices related to health. The judicious implications for policy. We stress here that conclusions for application of high taxes can discourage consumption of policy are highly dependent on the local epidemiological, cigarettes and excess consumption of alcohol, whereas sub- administrative, and financial context; it is within such con- sidized prices-for example, for contraceptives-can be a texts (at the national or district level) that policy is shaped. tool to promote desirable behaviors. Fuel taxes can reduce All that can be done in a general overview, such as this, is motor vehicle use, thereby decreasing pollution and vehicle to point to policies that appear approximately valid for a accidents, to take another example. Similarly, reduction of range of countries and that, therefore, are likely to serve as subsidies of some very high-fat food products can discourage a useful starting point for country- (or district-) specific consumption of fat. analysis of policy. * Direct Investwents. In many circumstances the only (or best) recourse for the govemment may be direct investments Policy Instruments of Government perhaps with policies of partial cost recovery.6 Immuniza- tion programs and vector control are two examples on the When intervention is desirable, govemments have available a prevention side. The complexity and relative infrequency variety of measures to promote health and prevent disease that of many case management procedures, combined with the include but extend far beyond the usual activities of ministries absence of informed consumers, suggest that a prominent of health.5 Governmental interventions may be usefully role for govemment in the financing of a basic level of grouped into five broad categories, the first three of which are hospital services may be desirable. The Health Transition: Implications for Health Policy in Developing Countries 685 * Research. Even if research results are protected by be addressed by policymakers in designing cost-effective patent, it can be difficult for the private sector to recoup intervention programs. the cost of research investment, and, when it is recov- ered, it is at the expense of fully appropriate use of the BEHAVIORAL CHANGE. Behavioral change includes personal research product. (Comanor 11986] reviews an extensive behaviors related to diet, hygiene and sanitation, personal literature on these issues in the context of the pharma- health habits, reproduction, and self-care or self-referral for ceutical industry.) The economic case is typically strong, illnesses. Worldwide experiences with agriculture, nutrition, then, for heavy contribution by government to finance family planning, and child survival programs have made it research. The purpose, of course, is to lengthen the menu clear that effective population-based health care requires ac- for intervention choice. tive and informed participation by families and communities (Homik 1988). Programs to prevent deaths from diarrhea and An Integrated Approach to Policies and Strategies respiratory infections in infants and children require that mothers be motivated and trained to become informed diag- In this collection we look at diseases or related conditions (for nosticians and managers of home therapy (Berman, Kendall, example, cancers, helminthic infections) one at a time. Al- and Bhattacharyya 1989; Mosley 1989). Correspondingly, though this disease-by-disease approach facilitates the techni- family and community involvement is essential for appropriate cal analyses of costs and effectiveness of specific interventions, antenatal and childbirth care (Walsh and others, chapter 17), in reality, policymakers and health planners must use a more for effective nutrition intervention programs (Pinstrup- integrated strategy and consider packages of interventions. In Andersen and others, chapter 18), for early diagnosis (Stans- this situation, issues of feasibility and sustainability arise field and Shepard, chapter 4), and for compliance with (Vilnius and Dandoy 1990). treatments for chronic conditions such as tuberculosis or hy- Feasibility encompasses political, administrative, and logis- pertension among adults. tical considerations. Some policies, such as raising the age of Each of the first four of the policy instruments of govem- marriage, may not yet be politically acceptable; others, such as ment noted above may need to be invoked to promote desired establishing environmental monitoring, may not be adminis- behavioral changes. The most direct approach is through mass tratively feasible because of lack of legal authority or trained media. Most governments are using radio and television broad- personnel. Lack of a well-functioning health infrastruc- casts and print mass media to reach the general population ture or an efficient distribution system may be a logistical with health information and promotional messages. In recent barrier reducing the cost-effectiveness of some strategies years there have been important developments in mass com- in the short run. munication strategies that are greatly enhancing their effec- Sustainability is a particularly serious concern, since be- tiveness in creating public awareness of health problems and fore this decade, few developing countries had seriously supporting appropriate behavioral changes (Church and attempted to implement a health care program with their Geller 1990; Gilluly and Moore 1986). Key elements of effec- own resources, where total population coverage was the tive communication programs include identifying the target objective. Consequently, the international community and audiences and conducting preliminary research to tailor the national governments are learning that even highly cost- message to their specific needs. Also, the media chosen must effective interventions like immunization programs may ex- be able to reach the target group. Most important, imple- ceed the available resources of some developing countries in mentation of a communication program must be a leamning the current economic climate. An advantage of the analytical process-all materials must be pretested and modified, and approach taken in this collection is that it identifies a range of the effect of the program must be carefully monitored and health policies and strategies, some of which may require only evaluated. minimal government resources (for example, regulations) and Entertainment for social change is a new concept in health some that can even generate revenue (for example, taxation communications that is rapidlygainingworldwide prominence on tobacco). This approach does, however, require health (Coleman 1988;Colemanand Meyer 1990). This method uses ministries to transcend their traditional bounds and look at the the universal appeal of entertainment by bringing together entire national development strategy with regard to its conse- popular entertainers, skilled producers, and health profession- quences for health. als to show people how they can live safer and healthier lives. In chapter 1, health interventions are classified as popu- To date, this strategy has been used most successfully in the lation-based or clinical (tables 1-5 and 1-8). The population- field of family planning, with productions ranging from music based interventions encompass five strategies: (a) change of videos in Latin America and the Philippines to television personal behavior; (b) control of environmental hazards; (c) dramas and soap operas in Nigeria, Egypt, India, and Mexico. immunization; (d) mass chemoprophylaxis; and (e) screening Often these productions are of such high quality that they gain and referral. Clinical interventions are assumed to occur, for top ratings on popularity charts. An advantage of this is that simplicity, at three levels: (a) the clinic; (b) the district they are often broadcast on commercial channels at no charge, hospital; and (c) the referral hospital. The discussion below therefore providing a major subsidy to the health education follows this same framework, elaborating on issues that must program. 686 W. Henrv Mosley, Jose Luis Bobadilla, and Dean T. Janiison Although the mass media are useful in introducing new environmental protection agencies may be set up indepen- ideas and providing information in support of health programs, dently of ministries of health. However the administrative the production of sustained behavioral changes in the popula- structure is organized, because of the nature of environmental tion generally requires a more comprehensive strategy, incor- hazards control, the activities must be administered in a way porating the more persuasive instruments of government to that facilitates maximum coordination and collaboration consolidate behavioral change. Perhaps the most neglected among diverse government agencies whose operations will tool here is taxation: estimated price elasticities of demand for directly, or indirectly, impinge on environmental health. alcohol and tobacco products are substantial.7 Such integrated Table 29-8 indicates the range of government agencies and strategies are discussed in detail in the chapters on injury programs outside the health sector that may need to be in- (Stansfield, Smith, and McGreevey, chapter 25), protein- volved in implementing environmental health activities. energy malnutrition (Pinstrup-Andersen and others, chap- Resources are limited in developing countries, but the multi- ter 18), and cancers (Barnum and Greenberg, chapter 21). sectoral character of environmental control programs means that their cost may be spread across government agencies and, ENVIRONMENTAL HAZARDS CONTROL. Environmental health by regulation and taxation, through the private sector. Thus, and safety is largely a matter of engineering and regulation to environmental improvements, although still constrained by reduce health risks from known environmental hazards, even overall national resources, are not dependent upon the budget when occurrences of the hazard may be increasing. For trans- of a single ministry such as the ministry of health. A broad port-related injuries, although the number of motor vehicles discussion of approaches to environmental improvement is and the distance driven are increasing, it has been shown that available in the World Bank's World Development Report the combined effect of seat belts, speed limits, safer roads, for 1992. better vehicles, drunk driving prevention, and so forth, has Consider motor vehicle injuries as an example. The health been to reduce the health risk (Kjellstrom and Rosenstock ministry may take the leading role in surveillance and identi- 1990). This pattern has been observed in industrial countries fying the growing problem. But a policy recommendation and in a few developing countries in which safety programs limited toestablishingemergencycare unitswithout involving have been implemented and data are available. As with motor vehicles, one can project a rapid increase in Table 29-8. Agencies Responsible for modern environmental health hazards associated with indus- Health-Related Environmental Improvements trialiZation and urbanization in developing countries. The Environmental concern Relevant agencies problems of environmental control will be compounded in many countries, however, because the low incomes and stan- Water availability Public works dards of living mean that the traditional hazards associated Water quality/fluoridation Industry with poor sanitation will remain. Waste disposal Agriculture Forestrv The underpinnings of environmental hazards control are: Urban development epidemiologic surveillance to detect illnesses or injuries re- Rural development lated to environmental risks swiftly; regular monitoring of Food safety Agriculture potentially hazardous environmental conditions; and regula- Food fortification Industry tory or taxation authority to ensure that appropriate risk re- Trade duction actions are taken. Traditionally, environmental Vector control Agriculture control programs in ministries of health have been limited to Urban development water, food, and sanitation inspection to reduce infectious diseases. Government capabilities and authority in this area M i soad and I must be greatly expanded to monitor and control a much Alcohol control broader range of environmental risks, including air pollution, Transportation toxic wastes, traffic hazards, occupational safety, unsafe man- Occupational safety Labor ufactured goods, and other health risks. Some of the profes- Industrv sional and technical capacities reqtiired to monitor and A'ricuIrure regulate environmental hazards may exist in different minis- Alcohol control tries in government; however, their functions are often limited Transportation by insufficient technically trained personnel, litnited re- Air quality Motor vehicle control sources, and, particularly, lack of statutory authority. Industry Significant government initiatives in environmental haz- Power developmiient ards control must begin with broad and detailed statutory Housing quality Urban development regulations empowering one or more agencies to take effective Rural development actions. Given the scope and magnitude of the tasks to he Housing carried out-which will encompass areas as diverse as law, Public works engineering, medicine, economics, physics, and chemistry- Soutrce: Authors. The Health 7Transition: Implications for Health Policy in Developing Countries 687 other relevant sectors in activities such as initiating measures in Tamil Nadu State in India, substantial program efficiencies to upgrade roads, highways, and intersections; improve motor can be achieved (World Bank 1990). vehicle safety; and reduce drunkdrivingandpedestrianhazards Cost savings to government for these mass interventions would rapidly produce diminishing returns. An intersectoral may be achieved through the use of mass mobilization cam- effort toward the prevention of motor vehicle injuries can be paigns, in which a substantial contribution in kind may be expected to have a synergistic effect, thus making the overall provided by the private sector. This has been the case with strategy more cost-effective. This example can be multiplied polio immunization mass campaigns conducted at intervals of with many environmental approaches to health interventions, six months in some Latin American countries. Program effi- as table 29-8 indicates. It reinforces the rationale for an envi- ciencies can also be achieved by focusing efforts in places ronmental protection agency with broad authority to monitor where the target population will be concentrated. Warren and hazards and take legal action. others (chapter 7) propose school-based delivery of "targeted mass chemotherapy" for intermittent (six-month or annual) IMMUNIZATION, MASS CHEMOPROPHYLAXIS, AND SCREENING. mass treatment for helminth infections witlh the objective of The population-based interventions included under the head- reducing worm burdens, and hence morbidity, without neces- ings immunization, mass chemoprophylaxis, and screening all sarily eliminating infections. The rationale is that in heavily share certain characteristics: (a) they involve the direct ad- infected populations it is not the acute effects of infection that ministration of a specific technical intervention to individuals are the major public health concem but the chronic insidious on a one-by-one basis; (b) they are directed to certain target effects of continuous moderate to heavy infection throughout populations; and (c) coverage of the target population is im- childhood, which reduces the growth and intellectual devel- portant to producing the desired effect. Technically, each of opment of children. Immunizations are among the most cost- these intervention strategies is highly effective when correctly effective of interventions discussed in this collection, and applied to a compliant subject, but their actual effectiveness school-based anthelmintic chemotherapy also appears highly in developing countries is strongly conditioned by the local attractive. administrative, managerial, and logistical capabilities, by tra- Where provision for treatment and follow-up is available, ditional cultural constraints, and by epidemiologic factors. screening selected populations for infectious diseases is also It is particularly with the interventions in this category that cost-effective, for example, miners for tuberculosis and com- the decision criteria noted in the section "An Integrated mercial sex workers for STDs. The latter strategy has recently Approach to Policies and Strategies," above, need to be care- gained in significance as an important public health interven- fully applied by policymakers, because the one-on-one charac- tion for two reasons. First, there is evidence that some STDs play ter of these interventions means that they are intrinsically a role in the transmission of AIDS. Second, theoretical work demanding of resources in terms of personnel and logistics. suggests that reducing the risk of HIV transmission by a small Even if the criteria are satisfied at the planninig stage, these core group of infected carriers with multiple sex partners is a interventions require careful monitoring and evaluation for much more effective means of limiting the epidemic than their effect during implementation; any breakdown in the treating a much larger group of people with few sex partners technical requirements of the intervention, any failure to (Over and Piot, chapter 20). Other diseases for which screen- reach the target population, or inadequate compliance with ing and referral are at least moderately cost-effective are breast required procedures by the recipients can greatly reduce their and cervical cancer (Bamum and Greenberg, clhapter 21). cost-effectiveness. Murray, Styblo, and Rouillon, however, the authors of the Traditionally child survival interventions which are tar- chapter on tuberculosis (chapter 11), do not recommend ac- geted to the same group (such as immunizations, micronutrient tive case finding. supplementation [particularly vitamin Al and growth monitor- ing) are combined into an intervention package to make more CLINICAL INTERVENTIONS Chapter I summarizes the range of efficient use of limited resources. This strategy can be very health interventions that require medical facilities, the level cost-effective if each of the specific interventions is carefully of facility required, and the estimates of cost-effectiveness. As monitored and regularly evaluated to see that it meets the with population-based interventions, national policymakers standards required for an effective programl and that it produces should have an informed epidemiologic analysis in the local the desired effect on health in the population. If, however, an context, along the lines of the framework given here, to guide activity is simply added to an operating program and no pro- the allocation of resources across clinical facilities of varying cedures have been properly established to ensure a health complexity and cost (Barnum and Kutzin 1993). Operation- effect, efficiency will decline. This has frequently been the case ally, the choices for clinical interventions should actually be when growth monitoring (a screening tool) has been intro- for packages of activities, because once certain institutional duced into child survival programs without any provision to resources are established (for example, a surgical suite with attend to children with faltering growth (Gopalan, cited by blood bank), many procedures can be performed at marginal Pinstrup-Andersen and others, chapter 18). Conversely, when cost. As emphasized in the discussion in chapter 1, however, growth monitoring is used as a tool to manage a population- the factor to consider is not just the marginal cost of the based nutritional supplementation program, as has been done procedure but its cost-effectiveness with regard to disability- 688 W. HenrY Mosley, Jose Lws Bobadilla. and Dean T. Jamison adjusted life-years gained. Institutional capacities will be lim- is the reorienting and retraining of health care providers. The ited, and the time and resources spent on relatively ineffective worldwide experience in initiating national programs to pro- procedures, such as surgery for lung cancer, will be taken away vide family planning services and child survival technologies from resources that could be spent on highly costeffective has revealed that the vast majority of doctors, nurses, and other interventions, such as cesarean section for obstructed labor. health care providers do not have the necessary training and From the perspective of developing public policy, the extent technical skills to provide even basic contraceptive technolo- to which economies of scale will (or will not) result from a gies. Before the introduction of oral rehydration therapy, packaging of services or from delivering required volumes of which rationalized diarrhea management, hospitals in many procedures determines the extent to which competitively pro- developing countries were experiencing acute diarrhea case- vided services can be efficient in any given demand environ- fatality rates as high as 10 to 30 percent; 99 percent of these ment. There is much anecdotal evidence to suggest that actual ought not to have occurred, given the availability of intrave- economies of scale sharply limit the scope for competition to nous fluids. Even now, in countries in which oral rehydration be efficient, suggesting the importance of government in fi- therapy has been introduced and available for five to ten years, nancing a basic level of care or regulating hospital services. many physicians typically administer unnecessary and, at Barnum and Kutzin (1993) provide a comprehensive anal- times, dangerous drugs to patients with diarrhea (Mamdani ysis of the economic and financial issues surrounding resource and Walker 1986; Martines, Phillips, and Feachem, chapter allocation to hospitals in the public sector. They note that 5). Similarly, in cases of acute respiratory illnesses medical hospital operating expenses, which commonly absorb from 40 practitioners may prescribe as many as three to six drugs, to 80 percent of public sector health resources, are at the core including more than one antibiotic, often in ineffective doses of the gap between required and available health resources in (Quick and others 1988). Furthermore, systematic patient many countries. In addressing this issue, thev make several follow-up is rarely carried out in primary health care facilities observations that are relevant to this collection. First, they to see if the treatment has been effective. confirm that in low-income economies nonhospital interven- Many of the limitations described above are the result of tions are more efficient in dealing with prevalent health con- resource constraints; however, it should be apparent that there ditions. They point out, however, that in countries with highly will be no cost-saving by poorly trained personnel dispensing successful primary health care programs (China and Sri Lanka) ineffective treatments (Stansfield 1990). Rather, for health a substantial proportion of health resources (above 60 percent) care to be cost-effective, health professionals and their support is spent on hospitals. Still, these are not large. tertiary facilities staff must be trained and motivated to diagnose and treat with sophisticated high-technology equipment, but district- properly the diseases they see. Much more use of practical level hospitals. diagnostic algorithms, continuing education programs, and This leads to Bamum and Kutzin's second point relating to careful supervision are essential to achieve this goal. Good efficiencies within the hospitals. In many low-income coun- records and case follow-up must become an integral part of tries, a high proportion of hospital expenditure is for person- treatment programs, since most of the benefits of therapeutic nel. With fiscal constraints, hospitals, particularly lower-level regimens are lost without proper patient compliance. With facilities, may be inadequately provided with drugs and other limited resources, gains in cost-effectiveness can be achieved essential supplies, resulting in low admission and turnover only by limiting the range of conditions to be cared for, and by rates. This can lead to misuse of tertiary facilities for minor doing the job well. Although the selective disease-specific illnesses and can contribute to the inappropriate provision by strategy of vertical programs is commonly criticized because of all hospitals of extended care or convalescence in order to the presumed lack of efficiency in the use of medical man- maintain bed occupancy levels. The implication of Barnum power, it has had the advantage of focusing attention on each and Kutzin's analysis is that hospital efficiency can be im- critical step necessary to make an intervention effective (Tay- proved by more effective allocation of resources to increase the lor and Jolly 1988; Mosley 1988). At the same time it builds a quality of care. This means strengthening both technical and base of practitioner competence that can later be extended to managerial skills, as well as providing for sufficient drugs and providing a broader range of services. supplies to care effectively for a selected group of conditions An important step in the process of improving the qualifi- for which interventions can be cost-effective. cations of health providers is strengthening professional asso- In considering specific activities that may be carried out in ciations. Presently, professional associations in many different levels of facilities, we briefly discuss certain generic developing countries are heavily dependent on commercial issues related to continuing education for health providers and enterprises (primarily the pharmaceutical industry) for na- the assessment, development, and control of technology. We tional meetings, publications, and continuingeducation. With deal with these issues here because they are essential ingre- this limited exposure to technical developments, health pro- dients in the process of selecting and implementing cost- viders are in no position to judge the merits (much less the effective interventions in hospitals and other facilities. cost-effectiveness) of new products for patient care. Continu- ing education with recertification of competence, which is a CONTINUING EDUCATION FOR HEALTH CARE PROVIDERS. A crit- requirement in highly developed countries like the United ical element in developing cost-effective health care systems States, where physicians have virtually unlimited access to the The Health Transition: Implications for Health Policy in Developing Countries 689 medical literature, is essential in developing countries, where Analytic Capacity Building resources are severely constrained. National professional asso- ciations could play a vitally important role in this area, but The fundamental underpinning of any health intervention government financial support is likely to be required to facili- program is measurement and evaluation. Without measure- tate provision of unbiased information. ment of the nature and magnitude of the health problem in a population and its trends and determinants, it is impossible to TECHNOLOGY ASSESSMENT, DEVELOPMENT, AND CONTROL. Even design intervention strategies that maximize the effectiveness in a wealthy nation like the United States, private hospitals of the health technologies. Correspondingly, in the absence of are not permitted to introduce expensive high-technology quantitative indicators of program performance, it is impossi- procedures (for example, open-heart surgery) without permis- ble to assess the efficiency of an intervention strategy, much sion of a government-mandated review board, which assesses less undertake analyses of the cost-effectiveness ratios of alter- the demand for heart surgery and the availability of the proce- native policy options. Managers of the smallpox eradication dure in other hospitals in the area. The hospitals themselves program stress the central role that outcome measurements are also shortening the duration of inpatient stay and moving played in the success of that program (Fenner and others many procedures to the outpatient facilities to cut costs. And 1988). consumer groups are demanding cost-saving innovations like In health intervention programs, measurement problems are the availability of less expensive generic drugs instead of the complex, but work has begun in developing the survey tools costly proprietary products. and analytical methods (White 1985; Gray 1987). Feachem, In the financially constrained environment of developing Jamison, and Bos (1991, p. 45) have reviewed experience with countries, cost containment is even more essential. To (and findings from) a range of analytical advances as applied move in this direction requires institutional capabilities as in Africa and conclude that "several new and powerful ap- described in the section "Policy Implementation and Health proaches-use of indirect demographic methods, case-control System Responses," below. But beyond a control function is epidemiologic techniques, and particularly, expanded use of the critical need for research to adapt highly effective tech- sentinel districts and facilities-offer highly cost-effective ways nologies to developing countries. This was done with oral for health ministries to meet an important part of their infor- rehydration therapy for diarrheal diseases, and steps are mation needs." being taken to simplify the diagnostic requirements for The microcomputer is the most important technical ad- effective treatment of acute respiratory infections (Martines, vance supporting the development of strengthened informa- Phillips, and Feachem, chapter 5; Stansfield and Shepard, tion systems (Berge, Ingle, and Hamilton 1986). Micro- chapter 4). computers have now been adapted for a wide range of health Adaptations of medical technology to developing country care applications in developing countries by the World Health settings have not only involved prevention and medical treat- Organization and the U.S. Centers for Disease Control, in- ments but surgery as well. Female surgical sterilization tradi- cluding managing primary health care programs and drug tionally had been an inpatient procedure done under general supply systems, monitoring immunization coverage, and stan- anesthesia. Experience has accumulated over the past twenty dardizing nutrition surveys (Victora 1986; Wilson and others years with performance of a mini-laparotomy that can be done 1988; Hogerzeil and Manell 1989; Babikir, Dodge, and Pett under local anesthesia on an outpatient basis (Liskin and 1989). Specialized software packages are also available for Rinehart 1985). Cataract extraction is another procedure demographic data analyses, field survey research applications, adapted to conditions in developing countries. High-volume health and population program planning, and so forth. Wide surgical facilities have been in place in India and Pakistan for applicability of epidemiologic and economic analyses will re- the past twenty years. In these settings, cataract surgery can be quire much more trained manpower in developing countries. performed inexpensively and safely on an assembly-line basis. Many governments will need to encourage the creation of More recently, a pilot program in Kenya has demonstrated new institutions or reconfigure old ones in order to address the that nonphysician ophthalmic clinical officers can be trained issues identified here. Critically needed capacities include: to perform cataract surgery with acceptable results (Javitt, chapter 26). * Demographic Analysis. These capabilities provide the An important case study of the adaptation of technology on fundamental underpinnings of a population-based health a national scale is the "simplified surgery system" developed in system. There must be accurate measures of the numbers and Colombia (Velez Gil and others 1983; Yankauer 1983). Con- distribution of the population, its social and economic char- trolled trials of selected surgical procedures were conducted, acteristics, and the trends and determinants of population comparing their safety and effectiveness when performed as change. These data will provide rhe basis for designing ambulatory procedures with that when performed as inpatient intervention strategies as well as for assessing the effect of procedures. The results indicated that 75 percent of surgical the disease burden on the population. interventions did not require hospitalization. The govern- * Epidemiologic Surveillance. This capacity is essential to ment has now instituted a nationwide program of ambulatory assess the magnitude of health problems, define their deter- surgery. minants, and monitor the effect of health program interven- 690 W. Henry Mosley, Jose Luis Bobadilla, and Dean T. Jamison tions. At the present time in most developing countries, program inputs (and their costs), outputs, and expected surveillance is limited to measuring the performance of a few outcomes. infectious disease control programs. Epidemiologic capaci- * Information systems must be established that provide ties will need to be greatly strengthened as health program timely data on health outcomes, intermediate objectives, strategies move more toward regulation, taxation, subsidies, and program inputs and costs. and information programs in order to reduce acute and * Regular analyses of input-output-outcome relationships chronic disease risks by changing behaviors and improving w regular ale inpt-utptoutome relionships envionmnta saey with respect to the instruments of government policy must environmental safety. be carried out to ensure that the instrument mix is, in fact, * EconomicAnalysis. Thedemographic and epidemiologic inducing the desired level of operation of the range of capacities will only measure the burden of disease, its trends, interventions. determinants, and the effect of interventions. Economic analysis will be essential to measure the cost-effectiveness Selecting health care priorities in a given setting is only of alternative intervention strategies as well as to assess the the first step toward improving the allocation of resources overall claim of the health sector on scarce development in the health sector (Murray 1990). The analysis of the resources (Bamum and Kutzin 1993). Building capacities in burden of disease would ideally lead to a list of health this area involve strengthening health service information problems ranked by order of importance. But clearly, the systems to measure more effectively the resource inputs, the fact that a health problem is high priority does not lead operations of the service delivery system, and its program automatically to the decision that the government should outputs. Continuing comprehensive analyses of these data invest in prevention or case management. As has been will be required to determine the cost-effectiveness of vari- shown in this collection, the role of cost-effectiveness anal- ous operational programs. This activity must encompass the ysis is to inform decisionmakers what interventions are private as well as the public health care sector. likely to yield more years of healthy life and therefore are * Health Technology Assessment. One aspect of cost- preferable. The results from the cost-effectiveness analysis effectiveness analysis has become known as health tech- can be used to make decisions at two different levels: first, nology assessment; institutional capabilities in this area to set priorities among the alternative interventions avail- must include not only the assessment of the effectiveness able to control a specific disease (for example, measles) or of new drugs, vaccines, or equipment but also their costs to reduce the exposure of the population to a specific risk and benefits when introduced into the health system. For tactor (for example, tobacco); second, to set priorities within example, there may need to be some control of the intro- the health sector, selecting the most cost-effective interven- duction of expensive high-technology health care inter- tions for those health problems that produce the greatest ventions such as computerized axial tomography, or C-AT, burden of disease. scanners or open-heart surgery in order to control health The idenitification of these high-priority interventions still service delivery costs. More important, because drugs is insufficient to justify public investment. For example, it is account for 40 to 60 percent of the health budget in many clear that in many countries, some of the cost-effective inter- developing countries (not including private expenditure). ventions are already being delivered by the private sector there is an urgent need to build up the institutional (including traditional practitioners) or by voluntary organiza- resources to assess these products, not only with regard to tions and, therefore, intervention by the government is not safety and effectiveness, but with regard to use and cost justified. In family planning, for example, there is strong in- (Mamdani and Walker 1986). volvement of the private sector in many countries (Lande and Geller 1991). Policymplemetationd Health Systemn Responses High-priority interventions, for which government involve- Pnment is justified, deserve a level of investment to achieve the The discussion above leads to five central conclusions for greatest possible coverage and the highest quality standards. Policy: But to achieve these goals and the ultimate outcome on health status, the health system needs to have the infrastructure and Acomprehensivehealthpolicyshouldmoveonmultiple organization to deliver the services. Table 29-9 shows a frame- fronts simultan.eously, considering the full range of facility- work that integrates three criteria-burden of disease, cost- based and population-based options for any problem being etfectiveness, and health system strength-to set priorities and addressed. ai plt-ae posoanpoe tg define more specifically the response from the health system. addressed. Strategies are suggested to strengthen the health system, if * Health strategies should be goal oriented, with specitic the system is weak, including development of trained staff quantitative intermediate objectives against which programn and necessary infrastructure. Other possible combinations achievements can be measured. of these three criteria are also shown, and possible responses * Planning the appropriate intervention mix should spec- from the health system are suggested. Importantly, this frame- ify as far as possible the quantitative relationships between work indicates that an intervention with a very unfavorable The Health Transition: Implicatons for Health Policy in Developing Countries 691 Table 29-9. Responses of Strong and Weak Health Care Systems to Burden of Disease Intervennon cost- Burden of disease effectiveness S_tronghealth systems Weak health systems High High Aim for full population coverage Reorient/train existing staff Improve quality of services provided Develop technical/ management systems Establish infrastructure Low Research to improve interventions Research to improve interventions Do not expand services Restrict or eliminate services Institute cost recovery Low High Target high-risk groups Provide services on demand Low Restrict services or provide cost recovery Eliminate services Source: Authors' design. cost-effectiveness ratio that is aimed at controlling a disease Despite the importance of using explicit criteria to set with low prevalence and low lethality clearly is a good candi- healthprioritiesforpublic investment indevelopingcountries, date for rationing or elimination. there is not much experience with country-level applications. Another combination of criteria of interest to developing An important exercise was undertaken in the late 1970s in countries passing through the later phases of the epidemiologic Ghana, however, which used the number of healthy days of transition is that of available interventions for high-priority life lost to assess the effect of diseases on health, and cost- diseases that have unfavorable cost-effectiveness ratios. The effectiveness analysis to assess the appropriateness of altema- most obvious example of this situation at present relates to tive interventions (Ghana Health Assessment Project Team AIDS, where control of transmission in the core population 1981). Five disease conditions were considered, namely, ma- offers the potential for being cost-effective (Over and Pior, laria, measles, childhood pneumonia, sickle-cell disease, and chapter 20) but actual application of the interventions is severe malnutrition. The results were used in the design of the difficult. The proposed response given in table 29-9 is research; Ghanian primary health care program, and the methodology operational research aimed at improving the cost-effectiveness proposed has served as a yardstick for subsequent developments of current approaches to patient care and to promoting and in the assessment of the burden of disease. supporting behavioral change, and basic research directed to developing new interventions, that is, vaccines or better drugs International Aid which will cost less and become more effective. As noted earlier in the discussion of clinical interventions, the control We have shown some of the implications of selecting disease of the emerging noncommunicable diseases, where tertiary control priorities, from the perspective of national govern- hospital care is not now cost-effective, will probably depend ments, for the process of designing and implementing pro- on the development of lower-cost interventions that can be grams. This section deals with implications for the instruments provided in district hospitals and health centers or through through which agencies providing development assistance in population-based programs. the health sector can channel their aid. We begin by catego- Probably one of the earliest attempts to incorporate the rizing these instruments and then turn to the implications of analysis of the burden of disease and the cost-effectiveness of this review's finding-conceming the health transition and interventions into the process of health planning was devel- intervention cost-effectiveness-for future directions of assis- oped by the Pan-American Health Organization, in collabo- tance in the health sector. ration with the Center for Development Studies at Caracas, Venezuela, in 1965. This is, by and large, the most com- Instruments of Aid prehensive planning methodology proposed for developing countries. It provides details of the planning process and the One reasonable categorization of aid follows from whether the requirements of information. Although the principles pro- objective is one of assisting in the provision of services, of posed were accepted, decisionmakers found them difficult to helping to improve the policy environment, or of expanding apply, mainly because the information available was inade- the research base underlying new interventions or improved quate and the complexity of the estimates demanded expertise resource allocation (table 29-10). These instruments of aid not always available. Fortunately, in recent years important relate closely to the instruments of government discussed in developments in informatics, epidemiology, and economic the previous section. (See also chapter 1.) Many of the success- evaluation of health services are making more accessible the ful experiences with aid in the health sector have had as their advanced methods proposed by the Center for Development objective the provision of services where no services, or only Studies planning model (PAHC/WHO 1965). inadequate services, were available. The smallpox eradication 692 W. Henry Mosley, Jose Luis Bobadilla, and Dean T. Jamison Table 29-10. Instruments of Aid Modality of assistance Objective Program implementation Capacity strengthening Service delivery Support acquisition of drugs, equipment, and Invest in institutional development and staff technical assistance for delivery of expanded training to improve efficacy of service program of immunization (EPI), vector delivery, for example, through improved control programs, hospital services logistics and supply systems Policy improvement Identify specific areas of policy improvement Invest in development of policy and (such as ban on tobacco advertising or planning departments in ministries or introduction of cost-recovery mechanisms) universities; invest in staff training and and include them (usually conditional) as advanced education part of an assistance package Undertaking research (including Conduct research or analyses (perhaps with Invest in national and international capacity epidemiologic, evaluational, and involvement of aid agency or expatriate staff) for undertaking research relevant to economic analyses) to strengthen formulation of policy or epidemiologic and economic conditions of delivery of service developing countries, both institutional and human resource development Source: Authors' design. effort had this objective, as does its successor, the Expanded important role in health sector operations at the World Bank, Programme on Immunization. Mission hospitals, too, are ori- sometimes closely tied to provision of service in so-called ented toward provision of service, as are many other forms of hybrid projects. assistance. Some aid to the health sector is channeled to research and to The capacity of a country to deliver services will, it is in- development of research capacity in recipient countries (see table creasingly recognized, depend a great deal on the policy environ- 29-11). Among the programs: ment in which systems for delivering services must function. The policy environment defines a range of key structural condi- * The Programme for Research and Training on Tropical tions: the mandated division of labor among public, private, and Diseases supports biological and operational research on five nonprofit nongovernmental organization sectors; cost-recovery major parasitic diseases and one bacterial disease that affect policy (and financing policy more generally); referral policy; more than 600 million individuals; it is currently expending pharmaceutical policy; policy toward prevention; policy toward about $40 million per year. taxation or subsidization ofhealth-influencing processes or com- * The Human Reproduction Program deals with biological modities; and policy toward distribution of access to services. and social aspects of fertility and its regulation; it currently Obviously some policy environments will be conducive to operates at a budget of about $23 million per year. inefficiency or inequity; others less so. * TheintemationalClinicalEpidemiologyProgram(INc(LEN) The potential importance of aid in assisting with improving and blindness-related programs on trachoma and oncho- policy has been the subject of much attention and debate in cerciasis represent efforts of foundations (the Rockefeller the past five to ten years. Policy-oriented aid inevitably has the Foundation and the Edna McConnell Clark Foundation, flavor of exchange of policy reform for financial assistance. The respectively) in two quite different domains. extent to which such exchange is productive depends greatly on the strength of those factions in the country who are Other important programs are well established-many of intellectually (or otherwise) committed to reform, on the them, like the Programme for Research and Training on Trop- substance and style of the discussions leading to agreement, ical Diseases and the Human Reproduction Program, managed and on the inherent viability of the measures adopted. Most by the World Health Organization and funded by multiple policy-based aid to date has been concerned with improving donors. The influential Commission on Health Research for macroeconomic policy; more than $1.5 billion of World Bank Development has been convened over the past several years (and International Development Association) lending in fis- and recently completed its work. The commission provided an cal year 1989, for example, was for "structural adjustment extraordinarily thorough critical review of current efforts and loans" (or SALs) involving fast disbursing resource transfer and capacities (Murray and others 1990) and of desirable directions macroeconomic conditionality. The World Bank is now also for future effort (Commission on Health Research 1990). The using "sector adjustment lending" instruments; incremental, Commission on Health Research has labeled this "essential highly flexible resources are made available to a sector in national health research" and identified support of such re- tranches released on certification of specified progress in policy search as a high priority. A follow-on secretariat to the com- improvement. Efforts to help improve the policy environment mission has been established in Geneva to facilitate research through sector adjustment lending are playing an increasingly efforts of individual countries. The Health Transition: Implicatons for Health Policy in Developing Countries 693 Much research important for resource allocation is relatively ity at the national or subnational level) is essential, and, nontransportable-local epidemiologic and operational anal- increasingly, assistance programs include substantial resources yses being important examples. Many research results, how- for capacity strengthening through institutional development. ever, are transportable; lessons from Senegal and the Gambia Often this involves direct assistance to an institution-for about the effectiveness in the field of oral and injectable polio example, a ministry headquarters or a hospital-designed to vaccine, for example, are probably almost as relevant in South improve its overall functioning. Relevant efforts may include Asia as they are in West Africa. The transportability of re- staff training, reorganizational advice, or support for develop- search does vary, of course. Little of use to Zaire in controlling ment of information systems. Often of particular importance AIDS is likely to emerge from study of sexual practices in San for capacity strengthening is investment in education and Francisco. Still, it is clear that much in the way of research training facilities for health professionals, including nursing, output is transportable, leading (in economists' jargon) to medical, and public health faculties. To be effective, such important informational externalities. Existence of these ex- investment may require a long time horizon; but the payoff can ternalities creates conditions in which any individual country be very substantial indeed. The Rockefeller Foundation's more is unlikely to invest fully in (nonpatentable) research because than thirty-year involvement with the Peking Union Medical that country reaps only a fraction of a research project's bene- College, for example, has had an influence on health policy in fits, yet it must pay the full cost. The existence of these informa- China, including Taiwan, that extends from the 1920s to the tional externalities, combined with substantial research capacity in present day (Bullock 1980). donor countries, makes research a particularlv viable domain for aid. In addition, the requirement for a substantial critical mass Conclusions of highly qualified (and, therefore, highly paid) scientists for much research points toward internationalization of the con- Estinmates of the levels and structure of cause of mortality duct of that research (as well as of its finances). This suggests reviewed in this chapter strongly suggest that, not only will the desirability of relatively few (but productive) venues with the number of deaths rapidly increase in developing coun- broad participation on the staff. tries, but there will also be a substantial (although incom- Two additional comments are worth making about re- plete) slhift in the distribution of causes to the relativelv search. First, despite the existence of programs that were expenisive noncommunicable diseases of adults and the el- described in the preceding paragraph, the current volumnie of derly. This shift, and the epidemiologic diversity likely to resources going into research is quite limited. The Commis- result from a lingering heavy burden of communicable dis- sion on Health Research (1990. p. 39) estimated, for exam- ease, will challenge healtlh systems to mount a broader range ple, that perhaps only a few pennies per death per year are of preventive interventions and to develop very low cost going into research onl such significant third-world killers as protocols for managing cases in increasing numbers. Several acute resplratory infections and tuberculosis; although rel- general conclusions follow: atively more is invested in tropical and parasitic disease research, the overall amount is quite limited. The commis- * As the increasing burden of noncommunicable disease sion estimates, overall, that only about 5 percent of the $30 is initially likely to affect the relatively more affluent and billion spent on health research in 1986 was oriented toward politically vocal older age groups, governments will need to developing countries, take great care to ensure completion of the unfinished Second, vaccine development efforts have the potential agenda for improving the health of children and the poor in both for providing cost-effective prevention for a much the face of resource demands placed (predominantly) by the broader range of conditions than is now possible and for relatively better off Almost certainly this equity objective reducing the cost and logistical complexity of currently avail- will be consistent with extending the investments in im- able vaccines. In relation to potential, research of this sort inunization and other interventions against infectious dis- currently receives very limited support-although the recent eases, which the chapters here have shown to offer the move toward creation of a Children's Vaccine Initiative sug- greatest gains in healthy life per dollar invested.8 A key gests that solutions to this problem mav soon emerge. input to completing the unfinished agenda will be invest- In summary the objectives of external assistance involve ment in research on vaccines-both to increase the range improving service delivery, improving the policy environ- of conditions to be addressed and, more important, to sim- ment, and supporting the generation of research findings that plify delivery logistics. underpin development of new interventions or more informed * Many of the risk factors for noncommunicable disease choice from among existing ones. Table 29-11 synopsizes these (smoking, sedentariness, increased motor vehicle use) tend, points and divides intervention concerning each objective for at least a tinie, to become more prevalent with increasing into two modalities: program implementation and capacity affluence; in this they differ from risk factors for most strengthening. Interventions oriented toward attainment of re- communicable diseases (with the exception of AIDS). The sults in the short term (and, often, this will be important) disadvantage of this is obvious; the advantage is that naturally emphasize the program implementation modality. In taxation-based preventive policies can actually generate the long term, however, strengthening capacity (usually capac- revenue for govemment while promoting health. More gen- 694 W. Henry Mosley, Jose Luis Bobadilla, and Dean T. Jamison Table 29-11. Directions for Intemnatiomal Aid Modality of assistance Objective Program implementation Capacity strengthening Service delivery Continue strong emphasis on most immunization Develop drug logistic capacity to support and family planning programs implementation priorities Enhance emphasis on: Offer pre- and in-service training of providers to * Measles immunization effectively manage priority procedures * Case management of acute respiratory infection * Control of vitamin A deficiency Develop capacity to deliver inexpensive * Tuberculosis chemotherapy rehabilitative services * Anthelmintic chemoprophylaxis * Control of sexually transmitted diseases Reduce emphasis on general institutional * Control of cancer pain development in favor of strengthening specific capacities Increase selectivity in delivery of ORT (oral rehydration therapy) and BCG (bacille Calmette- GLi&rin) immunization in low-risk environments Sharply reduce support for hospital facilities Policy improvement Implement full range of policies to limit use of Develop instruments for effecting sustainable tobacco increases in flow of resources to the health sector Implement policies to track and reduce use of Develop staff and institutional capacity for procedures of low cost-effectiveness formulating and implementing policies involving taxation, regulation, and communication, as well as Implement policies, including control of alcohol use, direct investment to reduce occupational and transport injuries Undertaking research Substantially increase aid resources for research Develop national and international capacity for conduct of essential national health research Finance and assist in the conduct of exemplary ENHR programs Develop and adequately finance international and national capacity for research on cardiovascular Increase epidemiologic operational research on: diseases in developing countries; also, perhaps, for * Cardiovascular disease other noncommunicable diseases and injuries * STDs (sexually transmitted diseases) * COPD (chronic obstructive pulmonary disease) Maintain and extend capacity for monitoring * Injury epidemiologic trends and efficacy of intervention in * Mental disorders well-documented populations (such as Matlab in Bangladesh) Assess intervention cost-effectiveness in different environments Source: Authors design. erally, increasing epidemiologic diversity will require a and rehabilitation-and to providing humane palliation for broader range of preventive measures; increasing use of the those whose lives could only be marginally extended (if at full range of govemment policy instruments (like taxes) can all) by affordable intervention. Some methods that are play an important role in implementing them. Of particular reasonably cost-effective have been identified, but signif- importance here is prompt national and intemational ac- icant efforts are required to develop and evaluate a more tion to control tobacco use. Acquisition of tobacco addic- comprehensive range of low-cost therapeutic interventions. tion by today's youth generates the dynamic for lung cancer, * Today's allocation of research resources to the health COPD, and cardiovascular disease epidemics in fifteen to sector in developing countries virtually ignores the problems thirty years. Taxes, prohibition on promotion, and other that will dominate the policy agenda in years to come. This effective interventions are available, and their prompt im- situation may have several roots: a sense that current re- plementation is high priority. search priorities should mirror operational ones; a sense that * To help preserve resources for the poor and to ensure the National Institutes of Health and their sister institutions broad access to reasonable treatment, great effort will need around the industrial world are doing what needs to be done to be devoted to implementing (or developing) low-cost about chronic disease; and, perhaps, a lack of appreciation ways of reaching the goals of secondary prevention, cure, for epidemiologic dynamics. Yet, as we have argued, case The Health Transition: Implicaions for Health Policy in Developing Countries 695 management of chronic disease will have to proceed in Latin America and the Caribbean environments drastically more cost-constrained than the ones for which institutions such as the National Institutes Antigua and Barbuda Jamaica of Health are working; relevant research and development Argentina Martinique efforts must be modified and evaluated for cost-effectiveness Bahamas Mexico in very different environments. Likewise, very little indeed Barbados Montserrat is known, for example, of the descriptive epidemiology of Belize Netherlands cardiovascular disease in the developing world, and no Bolivia Antilles available risk models are based on developing country data, Brazil Nicaragua which might include risk factors not observed in industrial Chile Panama countries. The list of examples could go on; the point is Colombia Paraguay simply that the analytic effort to address the emerging Costa Rica Peru health problems of developing countries in the 1990s and Cuba Puerto Rico beyond has barely begun. Dominica St. Kitts and Nevis * Just as manufacturers with older equipment expect Dominican Republic St. Lucia higher maintenance costs, older populations will generate, EcuLador St. Vincent and the for a variety of reasons, higher health maintenance costs for El Salvador Grenadines their country. National economic planners should expect to Grenada Suriname see, as populations age, expenditure on health steadily rising Guadeloupe Trinidad and Tobago as a percentage of the gross national product in the coming Guatemala Uruguay decades. Guyana Venezuela Haiti Virgin Islands (US) Honduras Other Latin America Appendix 29A: Regional Groupings of Countries Sub-Saharan Africa and Territories The following lists define the countries that are considered in Benln Mauritania the text. Botswana Mauritius Burkina Faso Mozambique Industrialized Market Economies Burundi Namibia Cameroon N iger Australia Japan Cape Verde Nigeria Austria Luxembourg Central African Republic Reunion Belgium Malta Chad Rwanda Canada Netherlands Comoros Sao Tome and Principe Channel Islands New Zealand Congo, People's Rep. of the Senegal Cyprus Norway Cote d'lvoire Seychelles Denmark Portugal Djibouti Sierra Leone Finland Spain Equatorial Guinea Somalia France Sweden Ethiopia South Africa Germany, former Switzerland Gabon Sudan Federal Republic of United Kingdom Gambia, The Swaziland Greece United States of America Ghana Tanzania Iceland Other Europe Guinea Togo Ireland Other North America Guinea-Bissau Uganda Italy Kenya Zaire Lesotho Zambia Liberia Zimbabwe Industrialized Transition Economies Madagascar Other West Africa Malawi Albania Hungary Bulgaria Poland Middle East and North Africa Czechoslovakia Romania Former German Former U.S.S.R. Afghanistan Bahrain Democratic Republic Yugoslavia Algeria Egypt, Arab Republic of 696 W. Henry Mosley Jose Lws Bobaditla, and Dean T. Jamison Middle East and North Africa (continued) our thinking of Richard Feachem and the effort he has led for the World Bank to review issues concerning the health of adults in the developing world. Last, Gaza Strip Qatar we acknowledge those who provided valuable comments on portions of this Gaza Strip Qatar chapter, which was given by Dean T. Jamison ab the Heath Clark Lecture for Iran, Islamic Republic of Saudi Arabia 1989-90 at the London School of Hygiene and Tropical Medicine. Iraq Syrian Arab Republic 1. Here and in much of what follows we draw heavily on estimates based on Israel Tunisia World Bank projections (Bulatao and Stephens 1989), which typically use Jordan Turkey combinations of model life tables rather than empirical estimates of mortality. Kuwat .United Arab Emirates Although we are aware of the shortcomings of this method, it provides the only globally complete projection model that is currently available. A more Lebanon West Bank epidemiologically based assessment of the distribution of death by cause in Libya Former Yemen, People's 1985, for the developing and the industrial countries, is presented in the Morocco Democratic Republic of chapter on cause of mortality (Lopez, chapter 2). Oman Former Yemen Arab Republic 2. Uemura (1989) has calculated excess mortality ratios for different coun- Pakistan Other North Africa tries and age groups at different points in time, using, as a reference, the lowest age- and sex-specific mortality rates so far observed in any country. His conclusions clearly show that the greatest gains to be made in developing Asia and the Pacific countries are in the younger age groups. (Greater gains are also possible, he shows, among females than among males at all ages, even though absolute Bangladesh Malaysia age-specific mortality rates are typically lower for females.) Bhutan Maldives 3. The Demographic and Health Surveys (DHS) is a nine-year project to assist govemment and private agencies in developing countries to conduct Brunei Mongolia national sample surveys on population and health. DHS is funded by the U.S. Cambodia Myanmar Agency for Intemational Development (USAID) and adminiscered by the China Nepal Institute for Resource Development. For more information about the DHS Fiji New Caledonia program (or copies of individual country reports) write to DHS, )RD/Macro Systems, 8850 Stanford Boulevard, Suite 4000, Columbia, Md. 21045, U.S.A. French Polynesia Pacific Islanids 4. Most of the implications of the health transition on the health system Guam Papua New Guinea described here refer to the health care subsystem. The possible effects on the Hong Kong Philippines other sectors of the economy are more difficult to anticipate, despite their India Singapore pctential importance. Particularly relevant is the effect of ill-health in the Indonesia Solomon Islands production of goods and services and the policy responses to the potential loss Kiribati Sri Lanka in productivity. 5. For a valuable general discussion of the economic role of govemment, its Korea, Democratic Thailand limits, and its comparative advantage, see Stiglitz 1989. A somewhat more People's Republic of Vanuatu mathematical treatment of these matters, with an emphasis on project evalu- Korea, Republic of Viet Nam ation methods, may be found in Starrett 1988. Birdsall (1989) provides an Lao People's Democratic Western Samoa extended discussion of the role of govemment in the health sector; she Republic Other ,.iArones,.emphasizes its past successes in many developing countries but calls for a Republic Other Micronesia redefinition of its role co leave more responsibility in routine areas to private Macao Other Polynesia actors and to achieve greater financial and administrative responsiveness in its own operations. Akin, Birdsall, and deFerranti (1987) discuss the financial aspects of these matters at greater length. Behrnan (1990) provides a clear overview of the central role of household decisionmaking (in many domains) Notes as determinants of health; this provides a context for assessing the role of government. We are deeply indebted to a number of our colleagues for comments and 6. We define "direct investment" broadly to include not only activities discussions concerning earlier drafts of parts of this material; they include directly administered by govemment but, also, services contracted for by Jacques Baudouy, Robert Black, John Briscoe, J. Richard Bumgamer, Donald government or natural monopolies (for example, tertiary facilities) that may Bundy, Guy Carrin, Lincoln Chen, E. Chigan, Andrew Creese, Joseph Davis, be partially privately owned or independently managed but whose policies are Nicholas Drager, Davidson Gwatkin, Jean-Pierre Habicht, Ann Hamilton, closely regulated in the public interest. Alaya Hammad, Jeffrey Hammer, Ralph Henderson, Kenneth Hill, Michel 7. It is worth stressing, however, that when price elasticities of demand are Jancloes, Jeffrey Koplan, Jean-Louis Lamborav, Joanne Leslie, Bernhard Liese, low, taxation ceases to be effective for changing behavior; for example, raising Judith McGuire, Richard Morrow, Mead Over, Thomas Pearson, Richard taxes on salt, even substantially, could be expected to have only a minimal Peto, Margaret Phillips, Nancy Pielemeier, Barry Popkin, Andr6 Prost, Wil- effect on consumption. liam Reinke, Ismail Sirageldin, Robert Steinglass, Eleuther Tarimo, Carl 8. World Bank (1989) and Feachem and others (1992) have assembled Taylor, Anne Tinker, Kenneth Warren, and David Werner. David Bell and evidence showing, convincingly, that the poor suffer more from chronic Anthony Measham have provided us particularly extensive and valuable diseases than do their well-offcounterparts. RelativeN, though, the poor suffer comments. The first section and the first part of the second section of this much more from infectious conditions; hence the desirability, from a dis- chapter are based on material we prepared with Donald A. Henderson ofJohns tributioinal perspective, of infectious disease control programs. Hopkins University (Mosley, Jamison, and Henderson 1990), and we would like to acknowledge both Dr. Henderson's direct contribution and his influ- ence on the chapter as a whole. An early version of those early parts of the chapcer was critically reviewed in August 1989 by the Board on Intemational References Health of the U.S. Institute of Medicine, and the comments of the chairman ofthat board, William Foege, as well as ofits other members, provided valuable Akin, John S.. Nancy J. Birdsall, and D. de Ferranti. 1987. Financing Health redirection. Likewise, we would like to acknowledge the strong influence on ServicesinDevelopingCountries:AnAgendaforRefon. AWorld Bank Policy The Health Transition: Implications for Health Policy in Developing Countries 697 Study. Policy and Research Division of the Population, Health and Nutri- Coleman, Patrick L., and R. C. Meyer. 1990. "The Enter-Educate Conference: tion Department. Washington, D.C.: World Bank. Entertainment for Social Change." In Proceedings. Baltimore: Johns Hop- Albanez, Teresa, Eduardo Bustelo, Giovanni Andrea Comia, and Eva Jesper- kins University Center for Communication Programs. sen. 1989. "Economic Decline and Child Survival: The Plight of Latin Comanor. William S. 1986. "The Political Economy of the Pharmaceutical America in the Eighties." Innocenti Occasional Paper 1, UNICEF Interna- Industry." Journal of Economic Literature 24(3):1178-1217. tional Child Development Centre, Florence, Italy, Commission on Health Research for Development. 1990. Health Research: Amler, Robert W., and H. Bruce Dull, eds. 1987. Closing the Gap: The Burden Essential Link to Equity in Developrent. Oxford: Oxford University Press. of Unnecessary Illness. New York: Oxford University Press. Comia, Giovanni Andrea, Richard jolly, and Frances Stewart, eds. 1987. Amler, Robert W. and D. L. Eddins. 1987. "Cross-sectional Analysis: Precurt Adjustment with a Hunan Face. Oxford: Clarendon Press. sors of Premature Death in the United States." In Robert W. Amler and H. Crimmins, Eileen M., Yasuhiko Saito, and Dominique Ingegneri. 1989. Bruce Dull, eds., Closing the Gap: The Burden of Unnecessary Illness. New "Changes in Life Expectancy and Disability-Free Life Expectancy in the York: Oxford University Press. United States." Population and Development Review 15(2):235-67. Babikir, Adam,Cole P. Dodge, and Ian Pett. 1989. "Monitoring Immunization Das Gupta, Monica. 1987. "Selective Discrimination against Female Children Coverage in Sudan." Health Policy and Planning 4(1):91-95. in Rural Punjab, India." Population and Development Review 13(1):77-100. Barcellos, T. M., and others. 1986. Segregacao urbana e mortalidade em P6rto Ernster, Virginia L. 1988. "Trends in Smoking, Cancer Risk, and Cigarette Alegre. Porto Alegre: Fundacao de Economia e Estatistica. Promotion." Cancer 62(8)Supplement: 1702-12. Barnum, Howard N., andJoseph Kutzin. 1993. Public Hospitals in Devel- Evans,John R.,KarenLashmanHall,andJeremyWarford. 1981 "HealthCare oping Countries: Resource Use, Cost, Financing. Washington, D.C.: World in the Developing World: Problems of Scarcity and Choice." New Englaid Bank. Journal of Medicine 305(19):1117-27. Behrman, J. R. 1990. "A Survey on Socioeconomic Development, Structural Feachem, Richard G. A., Dean T. Jamison, and E. R. Bos. 1991. "Changing Adjustment and Child Health and Mortality in Developing Countries." In Pattems of Disease and Mortality in Sub-Saharan Africa." In Richard G. Kenneth Hill, ed., Child Survival Prograns: Issues for the 1990s. Baltimore: A. Feachem and Dean T. Jamison, eds., Disease and Mortaltyv in Sub-Saharan Johns Hopkins University School of Hygiene and Public health, Institute Aica. New York: Oxford University Press. for International Programs. Feachem, Richard G. A., Tord Kjellstrom, Christopher J. L. Murray, Mead Bell, David E., and M. R. Reich, eds. 1988. Health, Nutrtion, and Economic Over, and M. A. Phillips. 1992. The Health of Adults in the Developing World. Crisis. Dover, Mass.: Auburn House for the Harvard School of Public New York: Oxford University Press for the World Bank. Health. Fenner, Frank, Donald A. Henderson, Isao Arita, Zdenek Jezek, and Ivan D. Berman, Peter, Carl Kendall, and Karahi Bhattacharyya. 1989. "The House- Ladnyi. 1988. Smallpox and Its Eradication. Geneva: World Health Organi- hold Production of Health: Putting People at the Center of Health Improve- zation. ment." In Ismail Sirageldin, W. Henry Mosley, Ruth Levine, Valerie Foege,WilliamH.,andDonaldA.Henderson.1986."ManagementPriorities Schwoebel and Kiyomi Horiuchi, eds., Towards More Efficacy in Chld in Primary Health Care." In J. A. Walsh and K198. MWaren, eds., Strategies Survival Strategies: Understanding the Social and Private Consrraints and Re- in Primar Health Care. Cni. A.iWalsh an K.iS. W ren s. sponsibiities. Baltimore: Johns Hopkins University School of Hygiene and for Priar Health Care. Chicago: University of Chicago Press. Public Health. Frenk, Julio, Jos6 Luis Bobadilla, Jaime Septilveda, and Malaquias Lopez Birdall Nacy j 199. T'hoght onGoo Heath nd ood ovenmet." Cervantes. 1989. "Health Transition in Middle-Income Countries: New BirdsalNs o d l or. Challenges for Health Care." Hea hPoicy and Planning 4(1):29-39. Bulatao, Rodolfo A., and Patience Stephens. (forthcoming). "Estimates and Fries, James F. 1989. "The Compression of Morbidity: Near or Far?" Milbank Projections of Mortality by Cause: A Global Overview, 1970-2015." Policy Quarterly 67(2 ):208-3 1. research working paper. Population Policy and Advisory Service of the Ghana Health Assessment Project Team. 1981. "A Quantitative Method of Director's Office of the Population and Human Resources Department. Assessing the Health Impact of Different Diseases in Less Developed World Bank. Washington, D.C. Countries." Internationaljournal of Epidemiology 10(1): 73-80. Bullock, M. B. 1980. An Amencan Transplant: The Rockefeller FourLdation and Gilluly, R. H., and S. H. Moore. 1986. "Radio-Spreading the Word on Family Peking Union Medical College. Berkeley: University of Califomia Press. Planning." Population Reports J-32. Johns Hopkins University Population Caldwell, John C., and Peter F. McDonald. 1981. "Influence of Maternal Information Program. Baltimore. Education on Infant and Child Mortality: Levels and Causes." In lnterna- Gray, Ronald H. 1987. "A Review of Methodological Approaches to Evaluat- tional Union for the Scientific Study of Population, Inernational Population ing Health Problems." IIP Occasional Paper 1. Report of the Workshop on Conference, Manila, 1981, vol. 2. Liege, Belgium. Health Impact Evaluation held at the Johns Hopkins University School of Chesnais, Jean-Claude. 1990. "Demographic Transition Pattems and Their Hygiene and Public Health, Institute for Intemational Programs, October Impact on the Age Structure." Population and Development Rerseuw 16(2): 9-10,1986, Baltimore. 327-36. Gunatilleke, Godfrey 1989. Government Policies and Nutrition in Sri Lanka: Church, C. A., and J. S. Geller. 1990. "Lights! Camera! Action!" Population Changes Dunng the Last Ten Years and Lessons Learned. Ithaca, N.Y.: Cornell Reports J-38. Johns Hopkins University Population Information Program. Food and Nutrition Policy Program. Baltimore. Hill, Kenneth, and Anne Pebley. 1990. "Child Mortality in the Developing Coale, Ansley J. 1991. "Excess Female Mortality and the Balance of the Sexes World." Populanon and Developmnent Review 15(4):657-81. in the Population: An Estimate of the Number of 'Missing Females."' Hobcraft, J. N., J. W. McDonald, and S. O. Rutstein. 1984. "Socio-economic Populatios and Development Revieu' 17(3):517-23. Factors in Infant and Child Mortality: A Cross-National Comparison." Coale, Ansley, and Paul R. Dimeny, with Barbara Vaughan. 1983. Regional Model Populanon Studies 38:193-224. Lfe Tables and Stable Populatwns. 2nd edition. New York: Academic Press. Hogerneil, Hans V., and Per Manell. 1989. "Computerized Drug Supply Cochrane, Susan. Joanne Leslie, and Donald J. O'Hara. 1982. "Parental SystemsforDevelopingCountries." HealthPolicyandPlanning4(2):177-81. Education and Child Health: Intracountry Evidence." Health Polcy and Hornik, R. C. 1988. Development Communication: Information, Agriculture, and Education 2(3/4):213-50. Nutmriion in the Third World. New York: Longman. Coleman, Patrick L. 1988. "Enter-Educate: New Word from Johns Hopkins." Hughes, Kenneth 1986. "Trends in Mortality from lschemic Heart Disease in JOICFP Review 15:28-31. Singapore, 1959 to 1983." Internationaljournal of Epidemiology 15(1 ):44-50. 698 W. Henry Mosley, Jose Luis Bobadilla, and Dean T. Jamison Ingle, Marcus, Noel Berge, and Marcia Hamilton. 1986. Microcomputers in Popkin, Bury M., Ge Keyou, Zhai Fengying, Xuguang Guo, Ma Haijiang, and Development: A Manager's Guide. West Hartford, Conn.: Kumarion Press. Namvar Zohoori. 1992. "The Nutrition Transition in China: A Cross- Jamison, Dean T.,John R. Evans, Timothy King, Ian Porter, Nicholas Prescott, sectional Analysis." Carolina Population Center. University of North Car- and Andre Prost. 1984. China: The Healfh Sector. A World Bank Country olina. EuropeanJournal of Cinical Nutrition (in press). Study. Washington, D.C.: World Bank. Preston, Samuel H. 1986. "The Relation between Actual and Intrinsic GTowth Kaplan, George A., Mary N. Haan, S. Leonard Syme, Meredith Minkler, and Rates." Population Studies 40(3):343-5 1. Marilyn Winkleby. 1987. "Socioeconomic Status and Health." In Rohert Quick,Jonathan D.,PatriciaForeman,DennisRoss-Degnan,andothers. 1988. W. Amler and H. Bruce Dull, eds., Closing the Gap: The Burden of Unneces- "Where Does the Tetracycline Go?: Health Center Prescribing and Child sary llness. New York: Oxford University Press. Survival in East Java and West Kalimantan, Indonesia." Unpublished report Kinsella, Kevin. 1988. Aging in the Third World. Intemational Population prepared by Drug Management Program of Management Sciences for Reports P-95, 79. Washington, D.C.: U.S. Department of Commerce, Health, Boston, Mass., with the Cffice of Population and Health, Jakarta, Bureau of the Census. and USAID. Kjellstrom,Tord,andLindaRosenstock. 1990. "TheRoleofEnvironmentand Rose, Geoffrey 1985. "Sick Individuals and Sick Populations." International Occupational Hazards in the Adult Health Transition." World Health Sta- Journal of Epidemiology 14(l):32-38. tistcs Quarterly 43(3): 188-96. Rutstein, S. 0. 1991. "Levels, Trends and Differentials in Infant and Child Lande, R. E., and J. S. Geller. 1991. "Paying for Family Planning." Population Mortality in the Less Developed Countries." In K. Hill, ed., Child Health Reports]-39. Johns Hopkins University Population Information Program. Priorities for the 1990s. Baltimore: Johns Hopkins University School of Baltimore. Hygiene and Public Health, Institute for International Programs. Last, John M. 1988. Dictionary of Epidemiology. 2d ed. New York: Oxford Sanders, David, and Rob Davies. 1988. "Economic Adjustment and Current University Press for the International Epidemiological Association. Trends in Child Survival: The Case of Zimbabwe." Health Policy and Lee, H. P., S. W. Duffy, N. E. Day, and K. Shanmugarathnam. 1988. "Recent Planning 3(3):195-204. Trends in Cancer Incidence among Singapore Chinese." InternationalJour- Stansfield, Sally 1990. "Potential Savings through Reduction of Inappropriate nol of Cancer 42(2):159-66. Use of Pharmaceuticals in the Treatment of ARI." In A. Gadomski, ed., Leslie, Joanne. 1991. "Women's Nutrition: The Key to Improving Family Acute loower Respiratory Infection and Child Survival in Developing Countries: Health in Developing Countries?" Health Polcy and Planning 6(1):1-19. Understanding the Current Status and Directions for the 1990s. Proceedings of Liskin, Laurie, and Ward Rinehart. 1985. "Minilaparotomy and Laparoscopy: a workshop held in Washington, D.C., August 2-3, 1989. Johns Hopkins Safe, Effective, and Widely Used." Population Reports C-9. Johns Hopkins University School of Hygiene and Public Health, Institute for International University Population Information Program. Baltimore. Programs. Baltimore. Mamdani, Masuma, and Godfrey Walker. 1986. "Essential Drugs in the Starrett, D. A. 1988. Founidation of Public Economnics. Cambridge: Cambridge Developing World." Health Policy and Planning 1(3): 187-201. University Press. Masironi, Robert, and Keith Rothwell. 1988. "Tendances et effets du Stlglitz, Joseph E., Mark Perlman, Douglass C. North, Dieter Bos, Chris tabagismedanslemonde."WorldHealthStatisticsQuarterly41(3/4):228-41. Freeman, A. H. E. M. Wellink, Ian MacGregor, jean-Jacques Laffont, Mosley, W. Henry. 1988. "Is There a Middle Way! Categorical Programs for and Arnold Heertje (ed.). 1989. The Economic Role of the State. Cam- PHC." Social Science and Medicine 26(9):907-8. bridge, Mass.: Basil Blackwell in association with Bank Insinger de . 1989. "Interactions of Technology with Household Production of Beauford NV. Health." In Ismail Sirageldin, W. Henry Mosley, Ruth Levine, Valerie Sullivan,J.M. 1991. "ThePaceofDeclineinUnder-Fn'eMortality:Evidence Schwoebel, Kiyomi Horiuchi, eds., Towsards Mcore Efficacy in Child Survival from the DHS Surveys." In iRD/Marco Intemational, Inc., Proceedings of the Strategies: Understanding the Social and Private Constraints and Responsibilities. Demograpnic and Health Surveys (D-HS) World Conference, Washington, D.C., Baltimore: Johns Hopkins University School of Hygiene and Public Health . 1991. Vol. 1, Columbia, Md. Murray, Christopher J. 1990. "Rational Approaches to Priority Setting in Inter- Susser, Murvyn. 1981. "Industrialization, Urbanization and Health: An Epi- national Health."JournalofTropicalMedicineandHygiene93(5):303-11. demiological View." In Intemational Union for the Scientific Study of Murray, Christopher ). L., David E. Bell, E. de Jonghe, Sarah Zaidi, and Population, International PopulationConference, Manila, 1981, vol. 2. Liege, Catherine Michaud. 1990. "A Study of Financial Resources Devoted to Belgium. Research on Health Problems of Developing Gountries."Journal of Tropical Taylor, Carl E., and Richard Jolly. 1988. "The Straw Men of Primary Health Medicine and Hygiene 93 (4):229-55. Care." Social Science and Medicine 26(9):971-77. Olshansky, S. Jay., and A. Brian Ault. 1986. "The Foutrh Stage of the Uemura, Kazuo. 1989. "Excess Mortality Ratio with Reference to the Lowest Epidemiologic Transition: The Age of Delayed Degenerative Diseases." Age-Sex-Specific Death Rates among Countries." World Health Stanstics MilbankQuarterly 64(3):355-91. Quarterly 42(1):26-41. Omran, Abdel R. 1971. "The Epidemiological Transition: A Theory of the United Nations, Department of Intemational Economic and Social Affairs. EpidemiologyofPopulationChange."TheMilbankMemorialFundQuarterly 1989. Prospects of World Urbanization, 1988. Population Studies 112 49(4):509-38. (sT/ESA/sER A/I 12). New York. PAHO/WHO (Pan-American Health Organization/World Health Organiza- USDHEW (U.S. Department ofHealth, Education, and Welfare), Public Health tion). 1965. Health Planning Problemns ofConceptan dMethod. Prepared at the Service. 1978. Ten Leading Causes of Death in the United States, 1975. Center for Development Studies (CENDEs) of the Central University of Atlanta: Centers for Disease Control. Venezuela, Caracas, in cooperation with the Pan American Sanitary Bu- ULSDHHS (U. S. Department of Health and Human Services). 1992. Smnoking reau. Washington, D.C. April. and Health in the Americas. DHHS Publication (cDc)92-8419. Atlanta: Piot, Peter, and King K. Holmes. 1989. "Sexually Transmitted Diseases." In Centers for Disease Control, Office of Smoking and Health. K. Warren and A. Mahmoud, eds., Tropical and Geographical Medicine. New Velez Gil, Aldolfo, Marco Tulio Galarza, Rodrigo Guerrero, Graciela Pardo York: McGraw-Hill. de Velez, Osler L. Peterson, and Bemard L. Bloom. 1983. "Surgeons and Popkin, Bury M., and E. Z. Bisgrove. 1988. "Urbanization and Nutrition in Operating Rooms: Under-Utilized Resources." Amencan Journal of Public Low-Income Countries." Food and Nutrition Bulletn 10(1):3-23. Health 73(12):1361-65. The Health Transition: Implicatons for Health Policy in Developing Countries 699 Verbrugge, Iois M. 1984. "Longer Life but Worsening Health?: Trends in 1990a. "China: Long-term Issues and Options in the Health Transi- Health and Mortality of Middle-aged and Older Persons." Milbank Menorial tion." Report 7965-CHA. Environment, Human Resources and Urban De- Fund Quarterly/Health and Society 62(3):475-519. velopment Division, Asia Country Dept. 111. World Bank. Washington, - . 1989. "Recent, Present, and Future Health of American Adults." In D.C. Lester Breslow, Jonathan E. Fielding, and Lester B. Lave, eds., Antiual --.1990b. "India-Tamil Nadu Integrated Nutrition Project Completion Review of Putblic Health, vol. 10. Palo Alto, Calif.: Annual Reviews. Report ." World Bank. Washington, D.C. Population and Human Resources Victora, C. G. 1986. "cric Anthropometric Software Package (CASP)." PC Division, India Country Department. software review. Health Policy and Planning l( 1):84-85. .1992. World Development Report 1992. New York: Oxford University Vilnius, Douglas, and Suzanne Dandoy. 1990. "A Priority Rating Systenm for Press. Public Health Programs." Public Health Reports 105(5):463-70. Yankauer, Alfred. 1983. "Lessons in Surgery for the Third World." Amencan White, K. L. 1985. "Health Surveys." World Health Stamtics Quanerly 38(1):2-14. Journal of Public Health 73(12):1359-60. WHO (World Health Organization). 1988. From Alma-Ata to the Year 200Q: Yu, J. J., M. E. Mattson, G. M. Boyd, M. D. Mueller, D. R. Shopland. T. F. Reflections at the Midpoint. Geneva. Pechacek, and J. W. Cullen. 1990. "A Comparison of Smoking Pattems in Wilson, R. G., B. E. Echols, H. H. BrTant, and A. Abrantes, eds. 1988. the People's Republic of China with the United States." Journal of the Management Information Systemrs and Microcomnputers in Prinmary; Health Care. Amenican Medical Association 264(12):1 575-79. Geneva: Aga Khan Foundation. Zaridze, D. G., and Richard Peto, eds. 1986. Tobacco: A Major Internaional World Bank. 1989. "Adult Health in Brazil: Adjustinig to New Challenges." Health Hazard. World Health Organization/IARC Scientific Publication 74. Report 7807-BR. Washington, D.C. New York: Oxford University Press. Appendixes Tobacco Production and Use Reducing Mortaity in Children under Five Priority Setingfor Health Senuice Efficiency Raionales for Choice in Public Healti Appendix A Control of Tobacco Production and Use Kenneth Stanley Tobacco was cultivated in the Americas more than 3,000 years Similarly, the adverse effects of tobacco on members of the ago and is believed to have originated there. By the arrival of family, coworkers, businesses and the environment have been Columbus in 1492, tobacco was being chewed, smoked, or investigated only in the last few years. snuffed in many areas of both North and South America. Tobacco cultivation was spread, primarily by the Spanish Health Effects and the Portuguese, to Europe, Africa, India, Turkey, Russia, China, and Japan by the early 1600s. By 1620 the Virginia The three leading causes of mortality for the productive age colony in North America was growing tobacco commercially group between fifteen and sixty-five years in both industrial for export. This lucrative trade helped to develop both the and developing countries are cardiovascular diseases, cancer, American colonies and the English merchant navy. . and accidents (WHO 1980). Chronic diseases are well recog- In the 1700s and the early 1800s, large quantities of tobacco nized as significant health problems in the industrial regions were being snuffed by the aristocracy of Europe and chewed by of the world. The prevalence of communicable diseases among the American pioneers as they pushed westward. By the middle children, however, often hides the fact that chronic diseases of the 1800s, however, the technology for making cigarettes are also becoming a serious problem in developing countries. and flue-curing tobacco had been developed, and the chewing Life expectancy in those countries has risen from 41.0 years in of tobacco was beginning to be seen as unhygienic. By World 1950-55 to 57.6 years today; it is projected to reach 70.4 years War I the mass production of cigarettes had begun, and smok- by 2020-25 (United Nations 1989). The correlation between ing rates among men in industrial countries began to rise life expectancy and mortality from cardiovascular diseases, dramatically. Cigarette smoking became popular among cancer, and infection is given in figure A-I. women in industrial countries starting about the time of World The association between tobacco use and ill health has been War 11. At this time, smoking rates also began to rise in men reviewed by many national and intemational committees and in developing countries. Filtered cigarettes became popular in organizations. Consistently, they conclude that tobacco use is the 1950s, and in the 1960s low-yield cigarettes entered the a significant cause of disability and premature death (RcP 1983; marketplace. Today, tobacco is cultivated commercially in WHO 1986; USDHHS 1989a). Worldwide, approximately 3 mil- more than 120 countries and is consumed in all countries of lion premature deaths per year are due to tobacco smoking (see the world. table A-i ). In Europe alone, there are more than 500,000 such Tobacco production and consumption influence various deaths each year; in the United States, the corresponding sectors of society in different ways, some negatively and some figure is 434,000, or one-sixth of all deaths. More than a quarter positively. As a result, it is important to consider the perspec- of all regular cigarette smokers die prematurely from smoking- tives of these various sectors, including the individual tobacco related diseases. user, the tobacco grower, the tobacco industry, the health The extent of mortality by disease that can be ascribed to community, and governments. My objective in this chapter is tobacco has been determined for the United States and is to review the influence of tobacco on each sector, to determine presented in table A-2. Lung cancer is the single largest con- the health and economic effect of tobacco, and to evaluate tributor, followed by ischemic heart disease. Lung cancer ac- strategies to control its use. counts for 26 percent of the mortality resulting from smoking. In the United States, about 1.2 million years of potential life The Adverse Effects of Tobacco before the age of sixty-five are lost each year, two-thirds among men and one-third among women (CDC 1991). Although the major diseases associated with tobacco have The rates of attributable mortality similar to those in table been known for more than thirty years, only recently have A-2 would be applicable for most industrial countries, in which many of the other health problems been firmly established. smoking has been a widespread habit for many years. Smoking 703 704 Kenneth Stanley Figure A-I. Relationship between Life Expectancy the early 1960s were infectious diseases, accidents, respiratory at Birth and Mortality from Cardiovascular diseases, digestive diseases, and neonatal deaths. But by the Diseases, Cancer, and Infections end of the 1970s, the most common causes of mortality in the area were cancer, cerebrovascular diseases, and heart diseases Percentage of all deaths (Gu and Chen 1982). This shift in health problems took place in less than twenty years and is marked by the emergence of 80 diseases caused by tobacco use. One of the common hindrances of effective action against 70 - CVD + cancer tobacco is the public's general lack of understanding of the relative importance of the various risks in daily living. Often, 60 - the local media provides continuous information about hazards from various factors with considerable sensationalism, in re- peated attempts to grab the public's attention. As a result a 50 - large portion of the public believes everything causes cancer Infections CVD so why worry only about cigarettes. In reality, however, to- 40 - bacco is the dominant public health hazard in industrial coun- tries. In the United States and many other industrial countries, 30 - smoking is responsible for more deaths than heroin, cocaine, alcohol, acquired immunodeficiency syndrome (AIDS), fires, homicide, suicide, and automobile accidents-combined. In 20 _ the United Kingdom, a report of the Royal College of Physi- cians expressed the extent of the problem by stating that 10 _ \ among 1,000 young male adults in England and Wales who \ 8 smoke cigarettes, on average about 1 will be murdered, 6 will 0 , , , , , , , , , i be killed on the roads, and 250 will be prematurely killed by 40 44 48 52 56 60 64 68 72 76 tobacco (RCP 1983). Life expectancy at birth (years) CANCER. In countries in which smoking has been a wide- spread habit, it is responsible for 80 to 90 percent of lung cancer Source: Based on analysis of United Nations statistics (Omran 1971), deaths and 40 percent of bladder cancer deaths. Tobacco is modified by Dodu. dah n 0preto lde acrdah.Tbcoi responsible for 30 percent of all cancer deaths, including some cancers of the oral cavity, larynx, esophagus, stomach, and has only recently become popular in many developing coun- cervix. tries, however, and the delay of twenty to twenty-five years An important feature in the relation between cigarette between the time one begins to smoke and the onset of many smoking and lung cancer is the strong correlation between the of the most important associated diseases such as lung cancer duration of regular cigarette smoking and subsequent lung means that current attributable rates calculated for developing cancer rates. A doubling of duration of regular tobacco use will countries will be somewhat less. Within ten to fifteen years, result in an increase in lung cancer incidence of approximately however, cigarette smoking will have been prevalent in many twenty-fold. This relationship holds particular relevance for developing countries for sufficient time to be the cause of a projecting the health problems of countries in which substan- mortality pattern closely approximating that currently seen in tial increases in tobacco smoking have occurred in the last industrial countries. The change has already occurred in some decade but the full health effects have not yet been felt. regions. For example, in Shanghai County, a rural and urban The concept that atmospheric pollution might be an im- area near the city of Shanghai, the leading causes of death in portant cause of lung cancer dates back to the 1930s, when it Table A- . Mortality Attributable to Smoking, by Region Region Deaths per year Year Source United Kingdom 110,700 1988 Health Education Authority 1991 United States 434,000 1988 CDC 1991 Europea 505,000 1985 WHO Regional Office for Europe 1988 Latin America 98,100 1985 USDHHS 1992 Industrial countries 1.7 million 1985 Peto and Lopez 1992 Worldwide 3.0 million 1990 WHO 1991 a. Exlcuding the former U.S.S.R. Source: See last column. Control of Tobacco Production and Use 705 Table A-2. Disease-Specific Mortality Attributable to Smoking, United States, 1988 Disease Males Ferniales Total Neoplasm Lip, oral cavity, pharynx 4,942 1,460 6,402 Esophagus 5,478 1,609 7,087 Pancreas 2,775 3,345 6,120 Larynx 2,401 589 2,990 Trachea, lung, bronchus 78,932 33,053 111,985 Cervix n.a. 1,246 1,246 Urinary bladder 2,951 963 3,914 Kidney, other urinary 2,729 363 3,092 Cardiovascular disease Hypertension 3,441 2,254 5,695 Ischemic heart disease Age 35-64 29,263 9,105 38,368 Older than 64 41,821 27,990 69,811 Other heart diseases 27,503 14,638 42,141 Cerebrovascular disease Age 35-64 5,121 4,504 9,625 Older than 64 11,554 5,134 16,688 Atherosclerosis 4,644 3,612 8,256 Aortic aneurysm 5,798 1,435 7,233 Other arterial disease 1,874 1,111 2,985 Respiratory disease Pneumonia, influenza 11,580 8,098 19,678 Bronchitis, emphysema 9,670 5,269 14,939 Chronic airways obstruction 29,838 16,884 46,722 Other respiratory diseases 828 690 1,518 Conditions in infants Short gestation, low birth weight 344 261 605 Respiratory distress syndrome 351 233 584 Other respiratory conditions 384 277 661 Sudden infant death syndrome 422 280 702 Bums 850 453 1,303 Passive smoking 1,330 2,495 3,825 Total 286,824 147,351 434,175 n.a. Not applicable. Source; COC 1991. was observed that lung cancer rates were higher in cities than CARDIOVASCULAR DISEASES. Approximately 25 percent of in towns. Subsequent investigations that have considered the ischemic heart disease deaths are due to smoking in countries effect of smoking habits, however, as well as national and in which smoking has been a common habit for many years intemational reviews, have led to the conclusion that no more (WHO 1979). The association with ischemic heart disease de- than 10 cases per 100,000 males each year could be ascribed to pends upon age, with the stronger effect for those at a younger atmospheric pollution in the high-risk populations and that age. As for lung cancer, the risk of death from ischemic heart the proportion of lung cancer attributable to smoking is of the disease decreases upon cessation of smoking. order of 80 to 90 percent. Smoking is also associated with atherosclerosis, hyperten- Oral cancer is a significant problem in South Asia, where sion, and cerebrovascular disease. In addition to mortality, the habit of chewing tobacco in the betel quid is common. Oral however, there is also significant morbidity associated with cancers almost always occur on the side of the mouth where tobacco; for example, amputation due to vascular disease in the tobacco quid is kept, and the risk of cancer rises dramati- the legs is common. cally for those who keep the tobacco quid in the mouth ovemight. Approximately 90 percent of oral cancers in this CHRONIC BRONCHITIS AND EMPHYSEMA. Soon after beginning part of the world can be attributed to tobacco chewing and to smoke, smokers develop a cough and produce more sputum smoking habits (WHO 1984). than nonsmokers; respiratory infections tend to increase, and 706 Kenneth Stanley lungfunctionbeginstobe impaired. Approximately 75 percent the family. In addition to the effects on the health of the of deaths from chronic bronchitis and emphysema are due to children of a smoking parent, and on that of nonsmoking adult smoking. In pure economic terms, bronchitis is probably the family members, the children are more likely to grow up to be most expensive of the smoking-related diseases because of the smokers also, with the resulting health problems for them- associated long-term morbidity. There is benefit in cessation selves, their spouses, and their children. Tobacco use among of smoking at any stage of bronchitis. children is one of the risk-taking activities which appears to be associated with an increased use of alcohol and other drugs. PREGNANCY, WOMEN, AND CHILDREN. Maternal smoking re- sults in slowing fetal growth because of reduction in the oxygen Effect on the Workplace supply reaching the baby through the placenta, to the extent that children born to smoking mothers weigh an average of Only within the last few years have the consequences of a 200 grams less than those born to abstaining mothers. Tobacco smoker in the workplace been realized. Studies in the United use causes a twofold increase in the risk of spontaneous abor- States (USDIiHS 1985) have revealed the following: tion and is associated with an increased risk of complications * Smokers take 50 percent more sick leave and are 50 during pregnancy and labor. The perinatal risk is increased by 35 percent for women who smoke more than twenty cigarettes percent more likely to be hospitalized; per day. It is estimated that more than 8,000 infant deaths each * Smokers are more than twice as likely to die during their year in industrial countries are caused by parental smoking. workig years (before age sixty-five); The effects of tobacco use by mothers in developing countries, * Smokers have twice as many on-the-job accidents; where birth weights are already low and perinatal risk high, * Smokers waste 2 to 6 percent of their working hours have not yet been determined. because of the smoking ritual; Smoking also increases the risk of cardiovascular disease for * Corporations incur increased cleaning, repair, and main- women who take contraceptive pills. Tobacco use is associated tenance costs because of smokers; and with increased rates of cervical cancer, and tobacco-related * Nonsmoking workers suffer significant irritation, dis- substances such as nicotine have been found in the cervical comfort, and health risks caused by smokers. fluid of smokers. Further, natural menopause occurs about two or three years earlier among smokers than among nonsmokers. The increased costs for life insurance (approximately 50 percent) and health insurance (30 percent) have been deter- PASSIVE SMOKING. The risk of lung cancer in nonsmokers mined by insurance companies, and programs developed to married to smokers is increased 25 to 35 percent as a result of return this money to the nonsmoking employees have served passive ("enforced") smoking, the breathing of other people's as inducements to promote some nonsmoking company poli- tobacco smoke. Children of parents who smoke have an in- cies in the United States. A West German branch of a U.S. creased incidence of bronchitis and pneumonia (NRC 1986; computer firm recently gave nonsmoking employees an extra USDHHS 1986). six days' vacation to compensate for cigarette breaks given year-round to smokers; as a result, 30 percent of the staff gave NICOTINE ADDICTION. All tobacco products contain nicotine, up smoking. It was estimated that in 1980 an average smoking a powerful drug that causes addiction, that is, the user's behav- employee costs an excess of $400 to $800 each year in 1983 ior is controlled to a considerable extent by the pharmacologic values (Kristein 1983). agent. Mechanisms of this addiction are similar to those of heroin and cocaine (USDHHS 1988). Effect on the Environment Effects on the Family Although only 0.3 percent of arable land worldwide is used to grow tobacco, most of this land could also he used to grow food A tobacco habit by one or more family members often drains and other crops. The reduction in food production associated a significant portion of the family income, typically in the with the growing of tobacco is likely to be associated with range of 1 to 5 percent of the income of a wage eamer in both increased prices for food locally and, hence, lower nutritional industrial and developing countries. Tobacco habits are more status in the general population. prevalent among the lower socioeconomic groups, and they The growing of tobacco requires large quantities of pesti- tend to be the hardest hit financially. The effect is likely to be cides and herbicides throughout most of its growing season. It greater in the poorest developing countries. It has been calcu- also depletes soil nutrients at a higher rate than most other lated for Bangladesh that the smoking of only five cigarettes crops and requires either fertile soils or the extensive use of per day would result in a monthly dietary deficiency of approx- commercial fertilizers. In tropical developing countries, which imately 8,000 calories in a poor household, seriously endanger- often have poor soils, the result is that either the farmer ing the survival of a large number of children (Cohen 1981). consumes considerable fertilizer (at a substantial cost to the Of course, smoking-related deaths and morbidity, such as farmer or the govemment) or periodically seeks out new crop- debilitating respiratory diseases, also mean a loss of income to land, often by deforestation. A significant problem also arises Control of Tobacco Production and Use 707 with the misuse of pesticides (purchased in larger-than-usual dients are added to the tobacco mixture. Cigars are made from quantities because of the increased cash profitability of tobacco air-cured and fermented tobaccos and vary considerably in as a crop) and possible contamination of village water supplies shape and size. Their smaller cousins, cheroots, are made from as a result of poor training and lack of education of the farmers, heavy-bodied tobaccos. a problem compounded by lack of health services in the area. The most common tobacco product smoked in India and Deforestation has been called the most serious environmen- neighboring countries is the bidi, made by rolling a small tal problem now facing developing countries. Approximately amount of ground tobacco in a tembumi leaf and tying it one-half of tobacco grown is flue-cured; in poor countries with a thread. In southeastern India, women practice reverse without coal, such as Brazil and most of Africa, this means smoking, in which the smoker turns a cheroot around and curing by the burning of wood. Farmers are taught the rule of keeps the lit end inside the mouth. Cloves are added to the thumb that one hectare of tobacco will need one hectare of tobacco mixture in Indonesia, to create local cigarettes called wood for curing. In many developing countries areas of tobacco kreteks. Many other areas of the world also produce local production are easily located by their lack of trees. The in- tobacco-smoking products, each with its own special charac- crease in erosion, deforestation, and prices of wood for other teristic and name. uses are among the results associated with the curing of to- bacco. In response to this problem in Africa, the British- SMOKELESS TOBACCO. Smokeless tobacco products, consist- American Tobacco Company (BAT) has initiated a replanting ing of tobacco leaf and a wide variety of flavoring and other plan, which, however, as yet has not produced a significant ingredients, are used either orally or nasally. In industrial reversal of the trend. countries, chewing tobacco is produced by shredding tobacco It has been estimated that 7 to 11 percent of fire losses in leaf, pressing the leaf into bricks (plugs), or by drying it out and the United States are associated with tobacco smoking, result- forming twists. Pieces are bit off and chewed or placed between ing in an annual cost of approximately one-third of a billion the cheek (or lip) and gum. Snuff, which may be sniffed or U.S. dollars (Kristein 1983). It is reasonable to suspect propor- placed in the mouth, has a much finer consistency than chew- tionally higher tobacco-smoking fire losses in developing countries. Table A-3. Worldwide Tobacco Leaf Production Tobacco Production and Consumption Production in Annual change 1990 (thousands between 1980 and Tobacco products are among the items manufactured most of metric tons) 1990' (percent) frequently by mankind. Approximately 5 trillion cigarettes are Country produced each year, or 1,000 cigarettes for each man, woman, China 3,019 10.3 and child on earth. United States 737 (.0) India 564 2.1 Tobacco Habits Brazil 444 1.1 Turkey 288 3.5 Worldwide, tobacco is consumed in a wide variety of ways, Italy 205 4.6 many in combination with other ingredients. Tobacco con- U.S.S.R. (former) 200 (3.5) many in ~~~~~~~~~~~~~~~~~~Indonesia 150 3.6 sumption can be divided into two broad categories, depending Zimbabw'e 140 3.0 on whether it is smoked or not. Greece 125 0 TOBACCO SMOKING. The most common form of tobacco use Region Africa 378 2.8 is the manufactured cigarette. This familiar product is made North and Central America 940 (1.5) from a blend of as many as 150 lots of tobacco, wrapped most South America 588 0.5 often in paper. The types of tobacco blended to produce the Asia b 4,660 6.0 cigarette vary, depending on the regional taste preference; Europe 667 (0.8) flue-cured tobaccos are popular in North America and most of Oceania 14 (3.0) Europe, whereas dark air-cured types are preferred in France Global and parts of North Africa and South America. Tar yields also Industrial countries 1,791 (1.7) vary, depending on the blend, lower levels generally being Developing countries 5,654 5.1 found in the industrial countries. Currently, there are about Developing countries 280 cigarette brands in the United States alone. (except China) 2,635 1.3 Pipe smoking was probably the earliest form of tobacco use Worldwide (except China) 7,446 3.0 and often has had social or ceremonial significance in the local culture. Water pipes of various types are in common use a. Calculated by author. Baseline values at 1980 are averages of 1979-81. Decreases are given in parentheses. throughout much of the Middle East, South Asia, China, and b. Excluding the former U.S.S.R. parts of Africa (IARC 1986). Often, molasses and other ingre- Source: FAO 1991. 708 Kenneth Stunley ing tobacco and is made from powdered or finely cut tobacco Table A-4. Worldwide Cigarette Consumption leaves. Moist snuff taken orally (dipped) has been used for Consumption in Annual change many years in Sweden and the United States, and it has 1990 (thousands between 1982 and recently become popular among adolescent males in those of millions) 1990a (percent) countries. Some tobacco companies have begun marketing it Country in small paper containers, like tea bags. China 1,641 7.2 For centuries, plant products have been chewed by eastem United States 547 (1.6) Mediterranean and South Asian population groups. When U.S.S.R. (former) 378 (1.5) tobacco was introduced, it was readily incorporated into many Japan 315 0.1 of these chewing habits (WHO 1988b). The most common oral Brazil 164 2.7 use of tobacco is the betel quid, widely used in South Asia and Germany 162 1.5 parts of Oceania. It consists of a leaf from the betel vine Indonesda 104 613 wrapped around sliced or shredded areca nut, tobacco, slaked Italy 96 (1.3) lime, and various flavorings. The large number of variations of France 96 1.1 oral use of tobacco, especially in South Asia, is remarkable. Region Africa 199 1.8 TOBACCO PRODUCTION. Tobacco is grown in more than 120 North and Central America 695 (1.6) countries worldwide, occupying a small portion (about 0.3 South America 270 2.0 percent ) of the world's arable land. This proportion, however, Asia b 2,734 4.6 is considerably larger in some countries, such as Malawi (3.8 Europe 923 0.2 percent), Greece (3.1 percent), Bulgaria (2.7 percent), and Oceania 43 0.2 Zimbabwe (2.6 percent). About 5 million hectares are under Global cultivation currently worldwide, with an average yield of about Industrial countries 2,299 (0.6) 1,500 kilograms of tobacco leaf per hectare (FAo 1991). Developing countries 2,943 4.5 Global tobacco leaf production is given in table A-3. China Developing countries is the world's leading producer of tobacco (40 percent), fol- (except China) 1,302 1.9 Worldwide 5,242 2.0 lowed by the United States (10 percent) and India (8 percent). Worldwide (except China) 3,601 0.2 The majority of the world's tobacco is grown in Asia; 76 percent is produced in the developing countries worldwide. Note: Consumption is defined as output plus imports minus exports. a. Calculated by author. Decreases are given in parentheses. The dominant trend in tobacco production is the 10.3 b.Excluding the former U.S.S.R. percent annual increase in China. Worldwide production of Source: USDA 1988, 1991. tobacco is increasing by 3.0 percent each year, but if China is excluded from the calculation, production is virtually stable. Production in the United States is decreasing at an annual rate percent), followed by the United States (10 percent), the of 1.0 percent per year. Commonwealth of Independent States (former U.S.S.R.; 7 The majority of tobacco leaf produced today is the flue- percent), andJapan (6 percent). Fifty-two percent of cigarettes cured type because of increasing preference for its use in are consumed in Asia. Worldwide, consumption is increasing cigarettes. At the current price of tobacco in most countries of about 2.0 percent per year, with the greatest rise occurring in $1.50 to $3.50 per kilogram, the value of the world's annual the developing countries. Cigarette consumption has been tobacco leaf production can be estimated at $10 billion to decreasing at an annual rate of about 1.6 percent in the United $20 billion. States and 2.5 percent in the United Kingdom since 1982. Approximately 85 percent of tobacco leaf grown worldwide Because of increasing health concems, the preference is used for cigarettes. There is considerable variation among worldwide has been moving toward cigarettes with filter tips. countries, however. Whereas virtually all tobacco is used for In China, the percentage of cigarettes with filter tips was 41 cigarettes in Japan, in the United States about 80 percent is percent in 1990, tripling the percentage of 1986. Filter-tipped used for cigarettes, 10 percent for cigars, and 10 percent for cigarettes account for more than 95 percent of the cigarettes other tobacco products. In India, about 30 percent is used for in Brazil, Germany, Japan, and the United States but for making the bidi, 20 percent for chewing, 15 percent for ciga- only 87 percent in Italy, 73 percent in Indonesia, 64 percent rettes, and the remainder for a wide variety of tobacco prod- in Poland, 60 percent in France, and 28 percent in the former ucts; about seven bidi are produced for each cigarette in India U.S.S.R. (USDA 1988). (USDA 1988). Health concems have also had an effect on the tar and nicotine levels of cigarettes. Median tar levels are less than 20 Tobacco Consumption milligrams per cigarette in Germany, Japan, and the United States, but high levels are found in China (26 milligrams per Cigarette consumption is shown by country and region in table cigarette) and Indonesia (36 milligrams per cigarette). Tar A-4. China is the world's leading consumer of cigarettes (31 levels in the United Kingdom and the United States have been Control of Tobacco Production and Use 709 falling at an annual rate of about 3 percent during the last the habit is more common among women than men, who twenty years (IARC 1986). Due to advances in technology and prefer to smoke. Smokeless tobacco habits often begin at the fact that less tobacco is needed in filtered cigarettes, the very young ages, and prevalence rates of 15 to 25 percent amount of tobacco per cigarette in the United States has for children ten years of age or younger have been reported been declining by about 1.5 percent per year during the last (IARC 1985). thirty years. More than 10 million people use smokeless tobacco in the The value of cigarette production worldwide is difficult to United States; annual sales amount to approximately $1 bil- determine because a large component of the price is taxes. lion. The situation is similar in Sweden, where more than 30 Taking an average price of approximately $1.00 for a pack of percent of the males age sixteen to thirty-five use snuff. Signif- twenty cigarettes as a crude benchmark leads to an estimate of icant use by children younger than six years of age has been the retail value of all manufactured cigarettes of $150 billion reported in some areas of the United States (Rouse 1989). to $250 billion-a more than tenfold increase over the price of the tobacco leaf alone. Tobacco Industry and Promotion Few countries have carried out national surveys of smoking prevalence, and rates can vary markedly within a country, Approximately 5 million hectares are under cultivation for especially between the urban and rural areas. Limited surveys tobacco, 80 percent of which are in the developing countries. have been conducted in nearly all countries, however, and can The tobacco manufacturing industry processes tobacco leaf be used to determine approximate national tobacco-use habits into cigarettes, cigars, chewing tobacco, and a wide variety of (see table A-5). Worldwide, about half of adult males and 10 other products, thereby increasing the value of the tobacco percent of adult females smoke. The difference in rates be- about tenfold. Few businesses are as profitable and as difficult tween the sexes is largest in the developing countries, partic- to enter as this industry. ularly in Asia. In a number of European countries, smoking rates among adolescent girls exceed those of the boys. Tobacco Growers Higher education levels tend to be associated with lower smoking rates worldwide (Chasov, Oganov, and Glasunov Whereas an average tobacco farm in the United States has 1984; Pierce 1989). For example, in China the smoking rate about 2 hectares planted in tobacco, in developing countries of male peasants was 81 percent, whereas that of white- and tobacco is often cultivated on smaller plots of 0.5 to 1.5 blue-collar workers was 42 to 58 percent (Tomson and Coulter hectares. There are about 4 million tobacco farms worldwide. 1987). Each hectare yields an average of 1,500 kilograms of tobacco, It is estimated that more than 200 million adults in South resulting in an annual global production of about 7.5 million Asia use smokeless tobacco. In Indonesia and parts of India, metric tons of tobacco leaf (FAo 1991 ). From estimates of labor use per hectare (USDA 1986), it can be determined that the average-size tobacco farm could be managed by a single full- Table A-5. Smoking Prevalence Rates in Adults, time farmer, who would have substantial time left over, if it 1985-90 were possible to spread the workload evenly over the year. Extra hands are typically used at planting and harvesting time, Most populous countries Males (percent) Females (percent) however. Therefore, although tobacco could provide full-time Country employment for something less than about 4 million farmers, China 61 7 in reality it provides part-time employment for a larger num- Indiaa 52 3 ber of farmers and laborers, very often women in developing U.S.S.R. (former) 65 11 countries. United States 32 27 It has been reported that 6 million people in India are Indonesia 61 5 employed in tobacco growing, and 35 million people are so Brazil 40 36 employed worldwide (Tobacco International 1974; FAO, Com- Japan b 66 14 mittee on Commodity Problems 1989). It is relatively easy to Pakisranb 44 6 Bangladesh 70 20 see, however, that these estimates are somewhat excessive, Nigeria 29 20 because comparison with the number of hectares under culti- vation (FAO 1991) shows that there would be more than Global fourteen farmers per hectare in India and about seven farmers Industrial countries 51 21 per hectare of tobacco being grown worldwide. Developing countries 54 8 Nearly always, tobacco is grown in rotation with other crops, Worldwide 52 1 0 such as maize, cotton, wheat, and soybeans. In most areas, Note: Regional and global estimates are based on population-weighted tobacco is a competitor with food for the arable land. In parts results of surveys for the most populous countries in each category. of Greece, Turkey, Malawi, and Zimbabwe, however, the soil a. Includes bidi and other forms of smoking. b. Includes chewers. is regarded as unsuitable for other crops, and the issue of Source: Author's compilation from World Health Organization surveys. competition with food crops does not arise. 710 Kenneth Stanley In many developing countries, either the govemment or the of about 2,500 rupees and net returns of 800 to 900 rupees (FAO tobacco industry provides considerable support to the tobacco 1982). In some developing countries, limits are set by the farmers, often in the form of technical assistance and training, government on prices for food crops in order to provide low- logistical support, and soft loans. Benefits of this support are priced food for urban centers; such limits reduce the incentive observed beyond the tobacco crop alone, because the sup- for farmers to grow food and increase the incentive to grow ported tobacco farmers also tend to produce superior yields tobacco. Nearly all the considerable resources used for growing of other crops grown in rotation or concurrently with the tobacco could easily be used for producing food instead. The tobacco. generation of hard currency and the fact that tobacco requires Over the years, improvement in agriculture technology has very little arable land notwithstanding, tobacco is grown in a led to a significant rise in tobacco yield per hectare of land. number of areas in developing countries where food is in short Tobacco production worldwide would be much greater than supply and could be grown. On the basis of the earlier estimates the present level if it were not for a network of governmental of approximately$ 10 billion to $20 billion as the value of world programs to limit the size of the tobacco harvest each year in annual tobacco leaf production and approximately 4 million order to keep the price high enough to provide a reasonable tobacco farmers worldwide (if they were full-time farmers), the profit for the farmer. A number of subsidies, incentives, and average tobacco farmer worldwide would receive a gross re- guaranteed price supports and other mechanisms are pro- tum of $2,500 to $5,000 and a net return of about $1,300 to vided by govemments to keep the tobacco-growing industry $2,500. healthy. Such mechanisms provide excess income to the farm- Although there has been considerable interest in determin- ers, giving them an incentive and the means to band together ing which crops can be substituted for tobacco that would to exert political influence to retain their preferential treat- provide a suitable economic altemative for the farmer, this is ment. Of course, this phenomenon is not restricted to tobacco a complicated issue. It depends on a number of factors, includ- farmers. ing soil types, climate, local dietary pattems, available man- Crop selection for farmers in a market economy is largely power, transportation system, crop-destroying pests, local and based on the maximization of net profits. For farmers in devel- external market prices, proximity to urban centers, processing oping countries, a number of other factors also come into play. plants, and trade centers; in addition there are the more Tobacco is a labor-intensive crop, and tobacco provides em- controllable factors of local govemment policies on price ployment for family and community members. Tobacco farm- supports, price limits, subsidies, production quotas and limits, ers can be assured of a relatively stable high price for their andagriculturalextensionservicesfortobaccoandothercrops. crops, often in hard currency. And, they do not face the usual Nevertheless, there appears to be sufficient reason to believe problem of needing rapid transport to avoid spoilage-in areas that a multinational effort, including governmental policy often bereft of even rudimentary services-that is encountered modifications, could produce a situation in a number of coun- with most food crops. The record-setting tobacco crop in tries in which it would be financially advantageous for farmers China in 1985 was largely due to an increase in the relative to grow food rather than tobacco. price for tobacco combined with govemment policy changes In the short term, however, the number of individuals giving the farmers greater freedom in planting decisions. addicted to tobacco use worldwide, many of whom would be Tobacco provides the farmers with gross retums per hectare willing to pay exorbitant prices, would indicate that the con- that are significantly higher than for all or most other crops, siderable industry made up of tobacco growing and tobacco depending on the soil type. Still, considerable costs are in- processing will continue to exist for many years. It will exist as curredinthegrowingoftobacco. In theUnited States, in 1985, either a legal or a black-market activity until society norms the cost of growing flue-cured tobacco amounted to 76 percent change sufficiently to produce a tobacco-free generation. of the value of the crop produced; 21 percent was for labor; 19 percent for machinery; 10 percent for curing fuel; 10 percent Tobacco Manufacturing Industry for the plant bed, fertilizer, and pesticides; and the remainder for marketing and inspection fees and other farm management Approximately 45 percent of world cigarette production is expenses (Grise and Clauson 1985). This excludes the cost of controlled by state industries in centrally planned economies, the land. In the United States, 14 percent of the tobacco farms and 14 percent is controlled by state-level tobacco monopolies; were operated by tenants. Although the relatively high costs the remaining 41 percent is dominated by a few intemational incurred in the growing of tobacco in the United States cannot conglomerates such as the British-American Tobacco Com- be extrapolated to the world, the need for labor, fertilizers, pany and Rothman's International, based in the United King- pesticides, and transport is virtually certain, and these ex- dom, and the Philip Morris Companies and the R. J. Reynolds penses can be estimated to be approximately half of the value Tobacco Company, based in the United States (USDA 1988). of the crop worldwide. The forces which have led to the situation in which cigarette Nevertheless, net receipts per hectare from tobacco often manufacturing is undertaken by a relatively limited number of exceed or are close to the gross receipts from most other crops. large enterprises include the highly automated technology For example, in India a hectare of tobacco produced a gross used for cigarette production, the need for sophisticated ad- return of approximately 8,000 rupees and a net return of 3,000 vertising and promotional techniques (in countries that per- rupees, whereas cotton and groundnuts produced gross returns mit advertising), and the high profitability, which provides Control of Tobacco Production and Use 711 funds that can be used to diversify and to deploy for political percent to the distributors, and the remainder to the manufac- advantage. turers. A net profit for the manufacturers of approximately 23 In centrally planned economies, excess income produced by percent, after advertising, materials and manufacturing tobacco is often used to offset shortfalls in other areas. There expenses, and taxes are removed, is the source of the consid- is little, if any, price competition among the conglomerates, erable pool of funds available for diversification, market ex- primarily because of the high taxes on tobacco products, which pansion, and other activities devised to maintain the position minimize the effect of any change in the manufacturer's price. of the tobacco conglomerates. And, although they compete vigorously for market shares Tobacco processing and manufacturing is the source of through advertising and promotional activities, they cooperate considerable employment-77,000 workers in the United in many other areas, including the sharing of manufacturing States, 44,000 in Brazil, and 40,000 in Cuba (USDHHS 1992). facilities. Ninety-two percent of the cigarettes produced worldwide Over the last few years, the conglomerates have diversified are consumed domestically (USDA 1988). Many of the tobacco- extensively-sometimes into related industries such as trans- growing developing countries, however, especially in Af- port and fuel, to control their costs, and sometimes into food, rica, supply tobacco leaf but are effectively excluded from clothing, and cosmetics. Although it is often said that this participating in the more lucrative industry of manufactur- diversification was undertaken because of projected future ing cigarettes. declines in tobacco consumption, there are significant eco- The leading exporter of cigarettes in 1990 was the United nomic reasons for it, related to the limited size of the tobacco States, with 27 percent of the world's total, followed by the market, surplus cash, and the need for continued growth to be Netherlands, Hong Kong, Bulgaria, Germany, and the United economically competitive. Kingdom. Hong Kong exports six times more cigarettes than The U.S. cigarette industry is depicted in figure A-2. Of the China. In the first half of the 1980s, Bulgaria was the world's retail price of cigarettes, about 30 percent goes for taxes, 25 leading exporter of cigarettes, mostly to the former U.S.S.R. Since 1989, however, the United States has been exporting more than twice as many cigarettes as any other country Figure A-2. Cgarette Busmess i the United (164 billion in 1990). The former U.S.S.R. imports the largest (SilltonsofUS do5lars) number of cigarettes (65 billion or 15 percent of the total world imports), followed by Japan and France. In 1990, China exported 10.4 billion cigarettes and imported 6.2 billion Federal $4.4 (USDA 1991). Consumer expeniture axes state $4.3 No large tobacco manufacturing company has yet admitted $30.2 $89 Local $0.2 publicly that tobacco use is harmful to health. The virtual certainty of overwhelming liability suits is probably the pri- mary reason for this position, even though it is indefensible Foreign sales before the battery of extensive evidence that tobacco is clearly Distritors $1.2 the cause of many significant and deadly health problems. The lack of effective public education programs in many de- veloping countries, combined with the efforts of many sectors Total farm within some countries which represent vested interests and _ Domestic tobacco $1.7 _ $3.1 which therefore do not want to hear about the detrimental effects of tobacco, give rise to the current situation in which -| Foreign tobacco $0.6 t tobacco habits are spreading rapidly throughout the develop- Tobacco inwo _ Nontobacco materials exports $1.5 ing world. IV r l r I $2.1 Other manufactu_ng Advertising and Promotion _Other manufacturing costs $1.4 Numerous advertising and promotional activities have been Manufacturers - $14.9 Advertising and undertaken by the tobacco companies. The industry claims Promotion $2.5 that the purpose of advertising is to maintain brand loyalty and to achieve brand switches among smokers, rather than to induce nonsmokers to start, or current smokers to increase consumption. Advertising is viewed by the health sector, Coo_t n$1.7 however, as one of the strongest inducements to smoking, especially to beginning it, and a ban on advertising is an Net income $3.4 important milestone in a national tobacco control program. The tobacco industry's arguments are weakened by the fact that they very strongly resist bans on advertising, much more Source: USDHHS 1989b. so than other restrictive measures, and that tobacco advertis- 712 Kenneth Stanley ing takes place even in countries where there is only one The Role of Governments in Tobacco Promotion tobacco company. For example, for many years BAT had a monopoly of the cigarette market in Kenva and still was the Governments, in virtually all cases, fall short of being homo- country's fourth largest commercial advertiser (Muller 1978). geneous decisionmaking and implementing bodies. For both It should be pointed out, however, that advertising is not market economies and planned economies, they consist of a needed to create a rise in tobacco consumption rates. Smoking number of competing factions. increased significantly after World War 11 in many countries, such as in eastern Europe, where there was no advertising MINISTRY OF HEALTHAND OTHER MINISTRIES. The key govern- of any products until recently. Often in these countries, to- mental factions involved with tobacco are the ministry of bacco consumption trends were linked to trends in disposable agriculture in close alliance with the ministry of finance, and income. the ministry of health, with, in general, a looser alliance to the In industrial countries tobacco industry advertising pro- ministry of education. The ministry of agriculture and the motes youth, fun, and adventure. In developing countries, it ministry of finance are nearly always two of the stronger stresses success and a high quality of life. Creation or mainte- ministries, with the former handling the crucial tasks of keep- nance of an image is the key strategy in modern advertising. ing the population fed and a large section of the population For example, the successful Marlboro cigarette brand advertis- employed, and the latter handling the allotment of public ing has even used a picture of only horses to promote its funds, keeping the economy running, and monitoring the product-the well-understood image being the free-spirited balance of payments. Conversely, although the ministry of life of the cowboy. Tobacco advertisements in the industrial health handles an important and politically sensitive function, countries in the 1960s and 1970s attempted to promote smok- it is generally thought of as one of the weaker ministries. ing among women by linking it to women's rights, proclaiming Further, both the ministry of health and the ministry of edu- in the 1980s, "You've come a long way, baby." cation are often viewed by the other ministries as bottomless In developing countries, tobacco is often advertised in a pits for absorbing public funds. manner that would be unacceptable in the industrial countries Ministries are influenced from without by a combination of in which the cigarettes are actually produced. Products have specific constituencies and broad public opinion. A large group names such as "Long Life," "New Paradise," and "Sportsman," of tobacco farmers with sufficient cash receipts to organize and the advertising is blatant and forceful. In counitries in themselves is a political force to be reckoned with. Lucrative which tobacco's health risks are not widely known, advertising taxes from the production, import, export, and sale of tobacco other than that of the tobacco industry's is limited and the provide a reliable source of government revenues, often pro- general desire to become "Westernized" is extensively pro- viding a significant portion of government income and thereby moted in movies, so even a small amount of advertising expen- capturing the alliance of the ministry of finance. In cigarette- diture can have a dramatic effect (RCP 1983). producing countries, however, the most important political A tobacco promotion technique that is becoming increas- force to be dealt with is usually the cigarette manufacturing ingly useful to the industry is the sponsorship of sports and industry, which produces for itself excess revenues that are cultural events. These activities give the tobacco industry considerably greater than those available to the farmers. much positive visibility, provide considerable leverage in some The specific constlituency for the ministry of health largely sectors, increase contacts with political decisionmakers, and consists of doctors, who are trained to diagnose and treat are much more cost-beneficial than direct advertising. disease rather than to deal with disease prevention. Only a In the United States, about $3.3 billion are spent each year small proportion of employees, if any, in a ministry of health on tobacco advertising and promotion-27 percent forpromo- is likely to have any training to deal with tobacco control tional allowances (paid to retailers and others to facilitate issues, such as legislation, public education, or childhood edu- tobacco sales), 41 percent for a variety of other promotional cation. Although it is this ministry which must take the lead activities, 14 percent for newspaper and magazine advertising, in the struggle against tobacco, a considerable portion of its 11 percent for outdoor and transit advertising, and 7 percent efforts in this regard involve working through other ministries for point-of-sale advertising (cLuc 1990). Tobacco advertising and organizations to achieve the desired results. in the United Kingdom amounts to more than 40 million Also, rather than disease prevention, the public most often English pounds per year. In Ghana, Malaysia, and Kenya, demands more hospitals, clinics, and medicines from the min- cigarette advertising accounts for 15 percent , 9 percent, and istry of health. And as always in prevention, the constituency 5 percent of all advertising, respectively (Wickstrom 1979). that benefits the most-those who have been prevented from The influence exerted on the media industry by this proportion acquiring the diseases and suffering the morbidity and prema- of advertising is, of course, considerable. Eight of twelve Amer- ture death-never know it. Only an epidemiologist can even ican large-circulation women's magazines did not feature a give an estimate of how many would have been in this group single article on the hazards of smoking for a period of more were it not for preventive efforts long ago. than twelve years, despite regularly featuring articles on health Further, the general public around the world, especially in issues (Whelan and others 1981). developing countries, has a difficult time understanding how Control of Tobacco Production and Use 713 a tobacco habit today will result in an increased chance (not impose sanctions on countries, such as Japan, South Korea, and a certainty) of an internal disease twenty or thirty years later. Taiwan (China), that have bans on or barriers to imported In some cultures it is not even possible to express this concept tobacco products. These countries have subsequently opened in the local language. their doors to U.S. tobacco to avoid possible trade sanctions. Cigarette advertising jumped from fortieth place to second TAXES Governments may tax tobacco in many ways, includ- place in total advertising time in Japan in two years, primarily ing taxes on the farmer, based on the amount of tobacco leaf as a result of American-style advertising campaigns. grown; taxes on the tobacco manufacturing companies, based In China, producing and marketing are controlled by the on the numbers of cigarettes or amount of other tobacco government through the China National Tobacco Company, products produced; taxes on import of tobacco leaf or manu- which uses quotas and allocations. China does not export factured products; and taxes at the time of retail sales. Taxes significant numbers of cigarettes, and virtually all imported are a significant factor in the price of tobacco products. For cigarettes come from Hong Kong. Companies who wish to example, they amount to more than 70 percent of the retail export cigarettes to China must purchase tobacco leaf grown price of cigarettes in Brazil, Canada, Denmark, Italy, and in China. Foreign tobacco companies were among the first to the United Kingdom; more than 50 percent in Mexico take advantage of the special economic zones and favored and Zimbabwe; and about 30 percent of the price in the investment conditions recently offered. In 1988, the R. J. United States. Reynolds Tobacco Company opened a $21 million cigarette Cigarettes provide a considerable portion of govemment factory in Xiamen. A German company is building a plant in income in many countries-between I and 2 percent of total Hong Kong for the primary purpose of producing cigarettes governmelt revenue in Italy, Japan, and the United States; for China. between 2 and 3 percent in Canada and Denmark; and be- Tobacco export and import are a considerable source of tween 3 and 4 percent in Greece and the United Kingdom. In currency transfer among countries. Tobacco leaf accounts for Argentina and Brazil, they provide 22.5 and 7.4 percent of 48 percent of the total commodity export earnings for Malawi government revenue, respectively (OECD 1985; USDHHS 1992). and 23 percent for Zimbabwe. A comparison of the total value Tobacco taxes yield more than $1 billion each year for the of tobacco imports with that of exports shows that the United Indian govenmment. States has the most significant positive currency flow (approx- From a govenmment perspective, taxes on tobacco have a imately $2 billion per year) from tobacco. Bulgaria, Greece, number of advantages. Not only are they a significant source the Netherlands, Turkey, Zimbabwe, and probably Brazil and of income for both the central and local govemments, thev are Malawi all have a currency flow from tobacco of more than relatively easy to collect-the tobacco manufacturing compa- $100 million annually. The former U.S.S.R. incurs the greatest nies simply transfer the funds into a government account; a currency loss (approximately $800 million), and China, Egypt, large collection agency is not needed. Further, governments France, Italy, Japan, and Spain have losses of more than $100 must try to raise tax revenue without too much public resis- millioni each. tance. Because tobacco consumption is generally greater in the lower socioeconomic classes, tobacco taxes raise significant Tobacco Control Strategies revenues trom this large public sector voluntarily, with virtu- ally no complaints. Still, tobacco taxes are basically transfers The aim of tobacco control programs is to establish nonuse of of funds. They do not increase national wealth but are a tobacco as normal social behavior, and the key to successfully convenient means of raising govenmment revenue. doing so is effective national action. The basic components of a tobacco control program are legislative measures, education SUBSIDIES. As mentioned earlier, govemments provide a and information, and national program organization. These wide range ofdirect and indirect supports to promote tobacco, components are described in the following sections and are primarily on behalf of the farmers. These include price sup- summarized in table A-6. Focusing on any single component, ports, incentives, production quotas, soft loans, import restric- such as public information alone, however, is unlikely to be tions, agriculture extension services, foreign-marketing successful. The optimal strategy is a comprehensive one in limitations, state trading, and state monopolies. Often a devel- which all important components are integrated; persistent oping country will invest in tobacco cultivation to provide a pressure should be maintained across the entire range of activ- source of employment and hard currency. In the United States ities and greater efforts made in specific areas as priorities the tobacco price support program cost only $66 million from dictate and as resources and opportunities make themselves 1933 to mid-1986; it is estimated that since then, however, it available. has cost approximately $1 billion (Warner 1988). The control of tobacco presents a different problem from most in public health. In this instance, the resistance to action EXPORTS AND IMPORTS.. Although significant strides have is not an insect vector or a shortage of trained health care been taken to control tobacco consumption in the United workers but rather is often a well-organized intemational in- States, section 301 of the U.S. Trade Act has been used to dustry withsubstantial monetary resources and an active media 714 Kenneth Stanley Table A-6. Effectiveness, Cost, and Resistance from Tobacco Industry for Components of a National Tobacco Control Program Resistance from Component Effectiveness Costa tobacco industry Legislative measures Increased taxation on tobacco products and other economic measures Very Inexpensive Strong Ban on tobacco advertising Very Inexpensive Strong Health wamings on tobacco products and advertisements Marginal Inexpensive Moderate Limiting the amount of harmful substances in tobacco products and specifying the amount on packages Marginal Inexpensive Little Protecting the rights of nonsmokers Moderate Inexpensive Moderate Protecting minors Moderate Inexpensive Little Education and information Informing leaders and key social groups Moderate Inexpensive Little Encouraging medical personnel and public figures to take leadership roles Very Inexpensive Little Informing the public about health risks Moderate Expensive Little Encouraging the public, especially children, never to adopt any tobacco habit Very Expensive Little Encouraging people who use tobacco to stop or decrease use Marginal Expensive Little Encouraging workers in high-risk industries and pregnant women to stop any tobacco habit Moderate Moderate Little National program organization Establishing a national agency to plan and coordinate the program Moderate Moderate Little a. For an agencv charged with planning and rinning a national tobacco control prograim. b. Such as &on public transportatisn and in restaurants and work sites. Source Author. campaign; in addition the industry provides considerable rev- some regulate to limit supply and others regulate to limit enue for governments and the media industry. Therefore, demand; and some provide a comprehensive range of controls. although some national-level strategies have been developed, Further, some laws are enacted on paper but never enforced implementation of these measures often meets with con- and hence are only of symbolic importance. The role of legis- siderable resistance. Continuous evaluation of the strategies lation in helping to establish nonsmoking as normal social of tobacco control and counterstrategies of the tobacco behavior, however, goes beyond its direct effect; legislation industry form the basis of the modem public health effort expresses public policy and sends a clear message to the popu- in this area. lation that tobacco use is harmful. The enactment of legisla- National tobacco control programs will, of course, differ tion represents a maturity of public concem about the health among countries, depending on a number of factors, including effects of tobacco and is a significant milestone in national the extent and type of current tobacco use, the extent of public health policy. As of 1986, sixty-four countries had current tobacco-associated health problems in relation to enacted legislation, whereas ten years earlier only nineteen other health problems, other pressing social problems, the had done so (Roemer 1986). Critics say that legislation can be extent of dependence on the tobacco industry, local cultural expensive or difficult to enforce. But experience has shown attitudes and public perception of the tobacco problem, and that if the legislation is not leading public opinion by too the commitment of the national leaders and physicians with great a distance and is accompanied by effective education respect to disease prevention. In most industrial countries, the programs, it can be implemented and will serve to change diseases associated with tobacco are highly prevalent, and as a the social environment and hasten the decline of tobacco result the public is in general agreement with control efforts. consumption. In many developing countries, however, cigarette smoking has The tobacco industry will vigorously oppose many aspects been common for only a few years, and the resulting health of legislation, particularly those measures that have been problems are just emerging. The public may therefore not yet shown to be the most effective-price increases and advertis- see the need for reduction in tobacco consumption. ing bans (see table A-6). The industry's opposition is often couched in the form of indirect attacks on the legislation that Legislative Measures appeal to people's fears that their right to freedom is being taken away or that "Big Brother" is looking over their shoulder. One of the best measures of national commitment to tobacco Success in achieving the enactment of legislation requires control is the extent of national legislation. Antitobacco laws extensive public information efforts and action by citizens to vary in rigor and scope: some are stringent, others exert mod- persuade their legislators of the necessity for legislation erate controls, and still others impose only weak restrictions; (Peachment 1984). As the WHO Expert Committee on Smok- Control of Tobacco Production and Use 715 ing Control Strategies in Developing Countries stated in 1983: decision to smoke fewer cigarettes; thus it will have an import- "It may be tempting to try introducing smoking control pro- ant role in reducing the number starting a tobacco habit. grams without a legislative component, in the hope that rela- The strong association between cigarette price and con- tively inoffensive activity of this nature will placate those sumption has also been observed in developing countries. For concerned with public health, while generating no real oppo- example, in India, cigarette sales declined by 15 percent after sition from cigarette manufacturers. This approach, however, the excise tax was more than doubled on the popular manu- is not likely to succeed. A genuine broadly defined education factured cigarette brands in 1986 (USDA 1987). program aimed at reducing smoking must be complemented by An increase in cigarette prices not only affects cigarette legislation and restrictive measures" (WHO 1983, p. 43). consumption; it also results in a switch to lower-priced brands Admittedly, it is difficult to demonstrate that a single legis- (often unfiltered), to hand-rolled cigarettes, and to other to- lative intervention will reduce consumption because so many bacco products, and if excessive it could lead to an increase in factors are involved in the use of tobacco. But studies have bootlegging. Tobacco duties were increased by 39 percent in shown a decline of smoking associated with controls on adver- the Federal Republic of Germany in 1982. By the next year, tising, introduction of rotating wamings, price increases, and the sale of name brands had dropped by 17 percent, but the airing of antismoking messages. Multifaceted legislative mea- sales of low-priced cigarettes and of tobacco for hand-rolled sures, in conjunction with other tobacco control measures, cigarettes had increased markedly, making up for 60 percent have resulted in substantial reduction in tobacco consump- of the decline of sales in name brand cigarettes (Ramstrom tion, for example in Finland and Norway (Roemer 1987). 1986). Although increases in cigarette prices are clearly one of Hong Kong, Ireland, Israel, and New Zealand have taken the most effective public health tools available to reduce the significant step of banning the importation and sale of cigarette consumption, only about half of the effect is a real smokeless tobacco products. Voluntary agreements between reduction in cigarette consumption, the other half being a the government and industry, such as those in Denmark and restructuring of the market. This problem can largely be solved the United Kingdom, have sought to control promotion of by marker-neutral simultaneous increases in the cost of all tobacco, but problems of interpretation and enforcement of tobacco products, with greater proportional increases in the the agreements have led health authorities to call for replace- least expensive, such as tobacco for hand-rolled cigarettes, ment of those agreements with legislation. which needs a proportional price increase in relation to ciga- rettes of more than three to one. PRICE POLICY In nearly all countries, the government plays The most frequent arguments against the raising of tobacco a significant role in setting the price of cigarettes, primarily prices are that it will lead to a decrease in governmental through taxes, and there is considerable variation in price tobacco tax revenues and that it will increase inflation. In among countries. For example, in northern and westem Euro- reality, however, an increase in tobacco taxes will cause a rise pean countries the retail price of twenty cigarettes varied from rather than a fall in tax revenues for a country (Wamer 1984; $4.17 in Norway and $3.60 in Denmark to $0.80 in France and Townsend 1987). The primary reason for this is that although $1.21 in Italy in 1987. The tax rate on cigarettes in European a price increase will result in a decrease in consumption, the countries varies from 35 to 87 percent of the retail price, decrease in consumption is proportionally smaller than the averaging about 53 percent (Roemer 1987). Tax rates are increase in tax revenues. It has been estimated that a 10 within this range for the majority of countries worldwide. percent increase in the tobacco tax rate will result in a 5 to 8 The most significant reductions in tobacco consumption percent rise in tobacco tax revenues (Godfrey and Maynard are apparently produced by a combination of regular price 1988; Jones and Posnett 1988). It is obvious that this relation- increases of tobacco products and an effective health edu- ship will not continue to hold if prices are raised to astronom- cation program. If either portion is missing, the effect is ical levels, but they can be raised considerably in all countries markedly reduced, and a decrease in tobacco tax rates can before a point of diminishing returns is reached. It must be easily negate the effect of other components of a tobacco pointed out also that by price increases, we mean increases control program. above the rate of inflation. If price increases do not keep The effect of raising taxes on tobacco products is measured pace with inflation, consumption will increase and tax reve- by the price elasticity of demand, the percentage of change in nues will fall. tobacco consumption associated with a 1 percent increase in An increase in tobacco prices could be inflationary, espe- price, adjusted for inflation. The price elasticity for cigarettes cially if the cost of tobacco items is linked to a cost-of-living in North America and western Europe is approximately -0.4; index. For this reason, a retail price index excluding the price that is, for every 10 percent increase in the price of cigarettes, of tobacco and alcohol products is now calculated by the consumption will fall 4 percent (Townsend and the Advisory Commission of European Commodities. An increase in taxes Committee 1987; USDHHS 1992). The fall in consumption is may even be deflationary, however, because taxes take money greater for teenagers (an elasticity of -1.4 [Lewit 19811), par- out of circulation. ticularly young males and those in the lower socioeconomic groups. Further, an increase in price will have a greater effect BAN ON ADVERTISING. Advertising is the strongest compo- on the decision to start or stop smoking than it will on the nent of the tobacco industry's promotional effort. In the 716 Kenneth Stanley United States, the tobacco industry puts more than 8 percent idea itself that these cigarettes may be less harmful leads to of the retail price of the cigarettes directly back into advertising the initiation of this habit by large numbers of youth. Fur- and promotion, an amount inexcessof$3 billionannually (FTc ther, by giving smokers support for their rationalizing be- 1988; cLc 1990). The magnitude of this financial commitment havior, such cigarettes weaken their will to quit the harmful is perhaps the best evidence that the payoff for the tobacco addictive habit. industry is dramatic. Tobacco advertising has an elasticity of Restrictions on tobacco use in public places, such as the approximately 0.09; that is, for a 10 percent increase in adver- banning of smoking or the setting aside of areas for smokers in tising expenditure, the tobacco industry can expect about a I public places, such as restaurants, public transport, and the percent rise in consumption (Townsend and the Advisory workplace, have been enacted in forty-eight countries (Roe- Committee 1987). mer 1987). These restrictions are designed to protect non- As mentioned earlier, the industry claims that the purpose smokers from the effects of passive smoking and to convey the of advertising is only to improve its share of the market, and message that smoking is not normal social behavior and can not to induce nonsmokers to start a tobacco habit. This view be harmful to nearby nonsmokers. Studies have shown an has been negated by studies, however, which show, for exam- increase in lung cancer in the nonsmoking wives of smokers ple, that brand loyalty for cigarettes is higher than for mnost that is three and one-half times greater than in the nonsmok- other consumer products (Tye, Warner, and Glantz 1987) and ing wives of nonsmokers. In fact, some passengers in the that the decision of teenagers to start smoking is largely a result nonsmoking sections of airplanes experience nicotine levels of the positive image promoted by the tobacco industry's comparable to those of individuals in the smoking section. In advertising. Recent advertisements by the R. J. Reynolds To- the United States this has led to the banning of smoking on bacco Company in Europe and the United States have featured all intemal commercial flights of six or fewer hours. the cartoon character "Joe Camel"; targeting of the youth in There has been little resistance from the tobacco industry this campaign was evidenced bya91 percent name recognition to legislation enacted to prevent youth from smoking, mostly rate in six-year-olds. by sales restrictions and prohibition in schools. The industry The first priority of legislation in this area should be a total is well aware that if one desires to encourage teenagers to do ban on tobacco advertising on television, radio, and other mass something, just make it illegal until they reach adulthood. media. Promotion of tobacco through the industry's sponsor- Nevertheless, such laws are important because they communi- ship of sports and cultural events and other indirect advertising cate to the youth that smoking is harmful. When the laws are should also be restricted. The tobacco industry often evades enforced and supported by strong education programs, the advertising restrictions by advertising nontobacco products combined effect can be a considerable reduction in the number such as clothing, shoes, and lighters, using advertisements that of young people who start smoking. are virtually indistinguishable from earlier tobacco advertise- In the past, laws conceming sales to minors have been ments. A total ban or at least some restriction on tobacco poorly enforced, but recent experience has shown that the advertising has been enacted in at least fifty-seven countries imposition of fines in a few well-publicized cases, together with (Roemer 1987). Of course, it should be mentioned that the required posting of notices that it is illegal to sell tobacco to former U.S.S.R., China, and some other countries already minors, can achieve compliance with the laws. Prohibiting or have complete bans on all commercial advertising; continuing restricting sales of tobacco in vending machines is another increases of tobacco consumption in those countries is proba- measure necessary to prevent sales to minors. bly related to increases in disposable income and availability Legislation can also be enacted to eliminate government of cigarettes. subsidies of the growing and manufacturing of tobacco. The U.S. and some other governments are against intemational OTHER TYPES OF LEGISLATION. The placing of health wamings legislation in the field of tobacco-apparently because of the on tobacco product packages is required in at least forty-three precedent it would set in further hindering free trade among countries worldwide (Townsend and the Advisory Committee countries. 1987). The use of strong rotating health warnings has largely solved the problem of the ineffectiveness of a single familiar Education and Public Information warning. Although epidemiological studies indicate that low-tar cig- A common misconception is that people will change their arettes are associated with a reduction in lung cancer rates of behavior if they are told how dangerous something is. The approximately 20 percent (Hammond and others 1976), there overwhelming majority of adults worldwide have been in- is no reduction in harmful effect with respect to cardiovascular formed of the health risks associated with tobacco, but this, by disease, respiratory function, pregnancy complications, and itself, has had little effect in slowing the spread of tobacco other diseases. Low-yield cigarettes, however, are used by the habits. Informing populations about the risks is, however, a industry to promote the erroneous concept that there is such necessary component of a comprehensive education program; a thing as a safe cigarette. Far from being safe, however, these mass media is effective in changing knowledge, attitudes, and cigarettes make it easier for youth and women to start smoking. beliefs. And although mass media can sometimes influence Smokers will change their habits to compensate (such as by behavior, individual contact is often necessary to change be- inhaling more deeply or smoking more frequently), and the havior significantly (Flay 1987). Control of Tobacco Production and Use 717 Some prefer to view the tobacco, alcohol, and drug-control advice led to tobacco habit cessation rates of 4 to 12 percent strategy efforts as a situation of supply versus demand, with in three study areas (Gupta and others 1986). education leading the effort to reduce demand. The use of this model in the tobacco field, however, is often supported by the SCHOOLCHILDREN. The most important component in the tobacco industry, who ingeneral know that their Achilles'heel control of tobacco is childhood education. Health education is legislation and that the industry can easily outspend public programs in schools, however, are generally poor worldwide education in advertising. Equal resources on both sides of a because health is often not a priority and teachers only rarely struggle between advertising and public information to influ- have training in health education. It is important that school- ence the public's perception of tobacco would be theoretically child programs begin at a young age, because by the age of interesting, but the tobacco industry would never agree to a twelve a child's attitudes and skills in health decisionmaking level playing field. The industry strongly resisted the ban on are largely formed. Further, health education should be com- tobacco advertising on television in the United States until prehensive, covering topics from personal hygiene to nutri- the health sector started running television commercials with tion, and should not focus only on a single topic such as Brooke Shields showing how socially unattractive smoking tobacco. was; the industry capitulated shortly thereafter. Over the years, certain strategies in childhood health edu- cation have been determined not to work. These include the KEY GROUPS. The first step in an education program is to appealtofear, inwhich individualsaretoldtheywillgetcancer inform key groups of the ill effects of tobacco and what should or heart disease if they smoke, and an emphasis on technical be done (see table A-6). One of the key groups is physicians, information, in which, for example, the aspects of tobacco who should be persuaded to take leadership roles. If the physi- production and cigarette manufacture are stressed. Often, this cians do not adopt healthy lifestyles, the public will not adopt latter strategy is counterproductive, leading to experimenta- them either. tion with tobacco. Moreover, the threat of disease and death in far-off years is rarely effective with young people. THE GENERAL PUBLIC. A cornerstone in a national education A number of comprehensive school-based education pro- program is informing the general public about the risks associ- grams have been developed, primarily during the last fifteen ated with tobacco. An appeal to fear, however, is ineffective years. Effective programs consist of two interlinked compo- as a long-term information strategy. To be effective, a program nents-health beliefs and skills development. The beliefs and must be run for a long time and should be characterized by opinions of the children concerning health should be openly simple messages on a common theme, affecting society's image discussed, with small group participation activities whenever of the tobacco user. Foran example of highly successful efforts, possible. The focus should be on susceptibility to problems it IS sufficient to observe a few of the tobacco industry's thought by children to be important; a child is often more advertising campaigns in industrial countries; these campaigns concerned about the smell of tobacco smoke or offending are well funded, generally involve a variety of medias and others than about the risk of heart disease or cancer. Perceived extensive visual images, and are of high professional quality. benefits and barriers to risk-reducing behaviors should also be The countering of these images, in the United States, for discussed, with the emphasis on nonuse of tobacco as normal example, has led to the creation of public information offices social behavior. which produce similarly high-quality public material, often Children also need to develop social resistance skills (to involving well-known and trusted public figures, that is aimed at resist peer pressure, poor adult models, advertisements, and establishing nonsmoking as preferable, normal social behavior. mass media), decisionmaking skills, and assertiveness. The Frequently, the public's perception of risk differs markedly setting of lifestyle goals by a child also often forms a basis for from the epidemiological reality. For example, risk-opinion resistance of peer pressure in the later childhood years. surveys indicate that the public in the United States views Most effective programs use either the existing teachers, nuclear power, handguns, and motor vehicles as greater risks older children (peer leaders), or a combination of both, rather than smoking (Upton 1982), whereas in reality smoking is far than specialized health education teachers. The "child-to- more dangerous: 30 percentofall casesofcancer in the United child" program (UNESCO 1988) was designed for use in devel- States are attributable to tobacco use (Doll and Peto 1981). oping countries and uses older children to teach the younger More than one-sixth of all deaths in males over the age of children, building on a linkage already existing in many of fifteen in India are attributable to tobacco (Gupta 1988). As these countries. This UNICEF-sponsored program is used in in Western countries, chronic diseases are the primary cause fifty-eight countries. of adult mortality in lndia, but whereas tobacco habits are Two of the most well known programs in industrial coun- prevalent, other high-risk habits resulting in chronic disease, tries are the "Growing Healthy" and "Know Your Body" pro- such as diets high in animal fat and low in fiber, are not grams. More than I million schoolchildren in the United common. States are studying "Growing Healthy." Both programs have One of the few large prospective controlled studies on the been shown to reduce the initiation of smoking by more than primary prevention of cancer was conducted in India. This 50 percent, as well as conferring other health benefits ("Results investigation of more than 36,000 tobacco chewers and smok- of the School Health Education Evaluation" 1985; Walter, ers showed that a combination of mass media and personal Vaughan, and Wynder 1989). 718 Kenneth Stanley SMOKING CESSATION. People continue smoking in the face of use. In the United Kingdom, smoking among males twenty overwhelming evidence of its detrimental effects because of years and older fell from 52 percent in 1974 to 35 percent in the social acceptance of smoking, the addiction, and, where 1986; for females, the corresponding decline was from 41 permitted, the constant pressure of advertising. The nicotine percent to 31 percent (Pierce 1989). Because of the lack of in cigarettes is one of the most addictive substances known. legislation and poor results from the voluntary agreements There is, however, a wide range in the level of addiction in a with the tobacco industry, the most significant component of smoking population. It has been estimated that about 95 the United Kingdom effort against tobacco has been health percent of the 37 million Americans who have stopped smok- education. ing have stopped on their own with no support groups or other The comprehensive Tobacco Act was passed in Finland in assistance. Still, stopping smoking often requires three, four, 1976. This legislation was one of the first and most successful or five attempts. Only a small portion of smokers participate national program actions taken against tobacco; one compo- in cessation clinics and in the associated research studies. nent of this act obligated the state to set aside 0.5 percent of In the United States, approximately 70 percent of all adults the tobacco excise tax to combat smoking. In 1975,40 percent see a physician at least once a year, but only about half of of adult males were daily smokers, and rates had been increas- smokershaveeverbeenadvisedbyaphysiciantostopsmoking. ing yearly. But by 1984, there had been a reduction to 33 Although physicians should play an important role in smoking percent (Leppo and Vertio 1986). cessation, the most effective cessation activities involve both The first report on smoking and health by the U.S. surgeon physicians and nonphysicians, and frequent contacts with the general in 1964 was an extensive review of the scientific smokers. Reliance on single methods, such as nicotine chewing literature, and the key political step against tobacco in the gum or counseling, is not as effective as combinations of United States. This report was communicated to the public methods, in which change of the social environment for the and served as the basis for formulating policy to control to- smoker is stressed. The average success rate of cessation pro- bacco consumption. The series of surgeon general's reports has grams at one year is about 5 to 10 percent (Kottke and others continued and now numbers twenty-two, and as a set it is the 1988). Routine minimal (30 to 40 seconds) advice to quit most comprehensive review and analysis of the association smoking, given by physicians and primary health care workers, between tobacco and health in the world today. In the United would produce significant effects worldwide simply as a result States, smoking prevalence among adults fell from 40 percent of the large number of contacts. Specialized cessation advice in 1965 to 29 percent in 1987 (USDHHS 1989a). for expectant mothers and workers in high-risk industries can Lung cancer mortality (or incidence) rates are perhaps the easily be incorporated into existing health counseling services. best marker of significant progress against smoking on the national level. A hard look at the lung cancer mortality trends, Tobacco Control Prograns however, compels one to conclude that the fight to control this disease worldwide is currently being lost. An effective national effort against tobacco normally requires In 1985, WHO reported its study of cancer mortality trends the establishment of a national agency or office to plan and covering the period 1960-80 in twenty-eight industrial coun- coordinate all aspects of the program. A budget for such a tries, representing 75 percent of the population of the indus- national agency in an industrial country of average size is in trial world. The most dramatic rise in age-adjusted mortality the rangeof$1 million to$10million. Frequently, thecreation was registered for lung cancer-76 percent for men and 135 of a national group to review the scientific literature and percent for women (wHO 1985). Mortality trends for males in recommend specific national actions to the public is the driv- selected countries are given in figure A-3. The mortality ing force behind the political will to take the necessary steps. reductions seen in Finland and the United Kingdom, where Even though it has been mainly countries in northern Europe comprehensive antismoking campaigns were first im- and North America that have developed national tobacco plemented, are the strongest evidence of effective national control programs, countries such as Chile and India have also programs. taken significant steps in formulating such programs. In many ways, the war against tobacco is analogous to the Economic Analysis and Conclusions war against drugs. Demand-side strategies can help to slow the growth, but supply-side strategies are also necessary in order to As the debate on the control of tobacco worldwide matures, it reduce consumption dramatically. It should be pointed out, is turning more to economic analyses. When the public listens however, that in the United States and most countries, sub- to this debate, it is faced with incomprehensibly large financial stantially more resources have gone to fighting drugs, which amounts on both sides of the issue. In this section the value of have claimed far fewer lives than tobacco. the retail market of cigarettes (VRM) is used as a yardstick The first national-level body to review the evidence against against which the costs and benefits can be compared. tobacco was the Royal College of Physicians in the United Kingdom, whose first report was published in 1962. This phy- The Economic Benefits of Tobacco Use sician-led group drew up recommendations for action and in 1971 established the organization Action on Smoking and The economic benefits of tobacco can be divided into the Health (ASH) to coordinate voluntary efforts against tobacco following categories: Control of Tobacco Production and Use 719 Figure A-3. Trends in Age-Specific Lung Cancer Mortality of Males Age 50-54, Selected Countries Age-specific death rates per 100,000 140 Poland 120- "..* 7 ,--- Italy 1 0 0 ........... ....1..0.......0....... United States 80 .';/ " _ France 80 v ,.. - ;/ / ~ - United Kingdom .- ,. / / / - Finland 60 ,- .' / ~ / /Australia Norway Chile _. _ = =~~~~~~ ~~_ ~Sweden 20 ==:= _ ----- Japan ,,,,,,,,,,,,-----''''' m~~~~~----------- 0~~~~~~~~~~~~~~~~~~~~~~~~ 0 I I I I Ii 1950-54 1955-59 1960-64 1965-69 1970-74 1975-79 1980-84 Year Source: WHO 1987. * Employment in the tobacco manufacturing industry billion) and a similar amount for tobacco exports. About 5 * Employtment for wholesalers, distributors, and retailers percent ($1.5 billion) of the VRM was spent by the manufactur- * Employment in the advertising and media industry ing industry for salaries to its workers and 8 percent for adver- * Income for tobacco farmers tising and promotion (see figure A-2; USDHHS 1989b). For Canada in 1979, about 50 percent of the VRM went for taxes, 7 percent to the farmers, 1 7 percent to the retailers and distrib- * Export of tobacco utors, and 25 percent to the manufacturing industry (Col- lishaw and Myers 1979). In Northern Ireland in 1984, 74 For the United States in 1985, about 30 percent of the value percent of the VRM was cigarette tax revenue. One-third of the of cigarette sales went for taxes, and 25 percent to the distrib- remainder went to retailers and two-thirds to the manufactur- utors. The farmers received about 6 percent of this VRM ($1.7 ing industry as employee earnings (31.6 million; Nelson 720 Kenneth StanLey 1986). In 1981-82, about 74 percent of the VRM in Egypt was percent per hectare from the "next-best" crop, if there were no tax revenues; 4 percent went to the retailers and 3 percent to shortage of arable land the decline in their profits would not the salary of employees in tobacco manufacturing (YE 21.4 be nearly as large. million; Omar 1987). Tax revenues from tobacco products should not be consid- The export of tobacco leaf is an important source of income ered an economic benefit from tobacco because they are for countries such as Brazil, Malawi, Turkey, and Zimbabwe. merely transfer paymnents-they do not affect the gross na- The majority ofcigarettesworldwide are produced withdomes- tional product or the standard of living. Taxes can be raised in tic tobacco, however, and the value of the tobacco leaf is other ways. As mentioned previously, cigarette consumption typically only 10 percent of the value of the processed tobacco is greater in the lower socioeconomic groups; perhaps tax products. The export of tobacco leaf is a source of hard cur- revenues could be raised in a more equitable manner. rency transfer from industrial countries to a few developing Thus, if tobacco were eliminated, the real loss in economic countries, but on a global scale this is only a small part of the benefit to society would be of the order of 5 to 10 percent of economic picture, less than 1 percent of the value of the the VRM for the industrial countries. In developing countries, worldwide retail cigarette market. even where tobacco-related employment is sometimes consid- erable, because of the "next-best" employment and crop phe- IF TOBAC(CO WERE ELIMINATEL) Consider the changes that noinenon, the economic loss associated with the elimination would take place if tobacco were eliminated worldwide. As of tobacco would still be only a small portion of the VRM, opposed to arguments of the tobacco industry, there would not probably never nearing 25 percent of the VRM in any country. be an absolIte loss equivalenit to the total value ofall generated It should he noted that the entire VRM came from disposable salaries and all indirect goods and services associated with the income paid out by tobacco users. If tobacco were eliminated, industry and its employees. Rather, the people employed in the nearly all this disposable income of tobacco users and their tobacco manufacturing industry would move to their "next- families would alternatively be used for the purchase of other best" employment opportunity, with possibly a few being goods and services-thereby supporting the economic devel- unemployed in the short term and a small number permanently opment of those sectors, providing employment and tax reve- unemployed. Instead of being involved in manufacturing a nues, although probably at a lower tax rate. Tobacco is a legal product that causes harm, they would disseminate to other product only because of history. If it were to try to enter businesses and the economy would simply adjust. The step the market today, it could not do so because of the built-in down in total income for this group would be no more than a safeguards against harmful products that now exist around few percent and probably less than 10 percent overall, even the world. when adding in the unemployed. As an example, for Canada, employment in the manufacturing industry has a value of 25 The Economic Costs of Tobacco Use percent of the VRM; the real decrease associated with the elimination of tobacco would be a 2.5 percent drop of the VRM. Virtually all analyses of health care expenditure attributable Similarly, for the people employed in tobacco advertising, to tobacco have been conducted in industrial countries, and distribution, sales, and other businesses related to the tobacco even from a single country the results of those analyses vary industry, individuals would either have to adjust their busi- considerably. Nevertheless, the broad conclusions are rela- nesses or seek "next-best" employment. tively consistent. For the newspapers and magazines that depend heavily on The costs of tobacco use are often categorized into one of tobacco advertising, its removal would be, from their perspec- three groups: tive, a virtual full loss of tobacco advertising revenues. But again, the resulting staff movement would be to the "next-best" * Direct health care costs-the costs of treating the dis- employer. It is conceivable that a small number of newspapers eases attributable to tobacco and magazines that depend heavily on tobacco advertising may * Indirect costs of lost productivity-lost income because fail. The cost of those failures, however, is a small price for of illness and premature death attributable to tobacco society to pay for the elimination oftobacco. A rough estimate * Nonmedical costs-including accidental fires and the wotild be a 10 percent drop from current benefit levels. loss of wood for the curing of tobacco If tobacco were eliminated, the farmers would shift to pro- ducing the "next-best" crop, often providing considerably less For the United States, direct health care costs associated income per hectare. The resources needed for growing a hect- with smoking were estimated in 1982 to be $16 billion (7 are of tobacco, however, such as labor and fertilizer, are con- percent of the national total health care costs and 73 percent siderablygreaterthanthoseneededforgrowingthe"next-best" of the VR\M) in 1980 (Rice and Hodgson 1983). The corre- crop. If land were not a limiting factor, and it often is not, then sponding estimate of indirect mortality and morbidity cost was the farmer could grow many hectares of another crop with the $26 billion ( 18 percent of the VRM). The Office of Technol- resources used to grow a single hectare of tobacco. Therefore, ogy Assessment of the U.S. Congress estimated the direct althougil farmers might expect a drop in net return of 50 to 70 health care costs to be $22 billion (70 percent of the VRM) and Control of Tobacco Production and Use 721 indirect costs of lost productivity to be $43 billion (140 percent I conclude that the dominant economic cost of tobacco use of the VRM) for 1985, but the office put a wide range on the in industrial countries is the indirect one of lost productivity, possible total-from $38 billion to $95 billion (U.S. Congress which is approximately two-thirds of the value of the retail 1985). In both analyses, the indirect costs alone were greater cigarette market, or larger. Male smokers are more than twice than the retail value of the cigarettes sold, or the VRM. Eco- as likely to die during their working years (before age sixty-five) nomic costs of tobacco have also been calculated separately for than nonsmokers (Mattson, Pollack, and Cullen 1987). The six states of the United States and for New York City; for these cost in developing countries is likely to reach that same level, areas either the total of direct and indirect costs or the indirect at a rate which depends on the twenty to twenty-five-year lag costs alone exceeded the VRM (Shultz 1986). time in health problems after the start of considerable tobacco For Canada in 1979, direct health care costs were estimated use among the population. to be $1.7 billion (Canadian), or 60 percent of the VRM; lost As mentioned previously, a number of govemments subsi- income due to premature mortality was estimated to be $3.3 dize tobacco growing or manufacture through a wide variety of billion (Canadian), or I 10 percent of the VRM; and fire damage measures. Elimination of these subsidies would free these gov- was estimated to be $85 million (Canadian), or 3 percent of ernment resources. the VRM (Collishaw and Myers 1979). For Northem Ireland in There are no reliable estimates of the value of the wood 1984, the cost to the individual smoker and family was esti- consumed for the curing of tobacco worldwide, but the value mated to be $271 million, or 137 percent of the VRNI; and the is almost certainly in excess of $1 billion. And the price of cost to the employer, $135 million, or 68 percent of the VRM recovery from desertification is probably also of considerable (Nelson 1986). For Egypt, direct health care costs associated magnitude-if, indeed, it is possible at all in those areas with with tobacco use were estimated in 1982 to be $151 million, insufficient rainfall. Further, if the wood were not used for or 17 percent of the VRM; and indirect costs of lost productivity, curing, much of it would probably enter the marketplace and approximately $78 million, or 9 percent of the VRM (Omar thereby reduce the price of fuel for the general public. 1987). Also, Egypt had to pay an amount equal to 16 percent Each year about 3 million premature deaths worldwide are of the vRM to import foreign cigarettes and the tobacco leaf and due to tobacco. This tobacco is grown on 5 million hectares of other materials to make domestic cigarettes. land. Hence, it can be estimated that each seven hectares of Although there is wide variation among countries, it is tobacco grown will result in approximately four deaths each possible to conclude that in industrial countries in which year: one death from lung cancer, one death from ischemic smoking has been common for many years, the total of the heart disease, one death from another cancer or cardiovascular direct health care costs and the indirect costs of lost produc- disease, and one death from a respiratory or other disease. tivity are significantly greater than the value of the retail cigarette market, and that either the direct cost or the indirect Conclusions cost, taken alone, is likely to be at least two-thirds of the value of this market. In developing countries, the costs of tobacco When all the economic costs and benefits of tobacco use are use are directly linked to the proportion of disease attributable summarized and compared. the single element that emerges as to tobacco, which in tum is directly associated with the length determining the conclusion is the simple fact that male smok- of time of significant tobacco consumption in the country. The ers are more than twice as likely to die during their working costs in these countries will continue to rise in the next twenty years (before age sixty-five) than nonsmokers. The energy and to twenty-five years and will Liltimately reach the same levels productivity of these people have been wasted. If tobacco were as in the industrial world. eliminated worldwide, virtually all other economic concems The previous cost analysis does not take into account that related to this event would either be of a much smaller order in the absence of tobacco, people will still die and thus incur of magnitude, or the system would simply adjust-individuals health costs, although years later. Although tobacco-associated would seek employment elsewhere, farmers in developing diseases tend to be more expensive to treat than other compet- countries would grow food rather than tobacco, and taxes ing causes of death at the same age, health costs for more would be raised by other means. elderly individuals would also be more expensive on average. Of course, the real reason for reducing tobacco consumption An analysis in Switzerland included a comparison of the health is disease and suffering, not economics. It is virtually impossible care costs of a smoking population with those of a hypothetical to put a value on life or suffering. Whatever amount we are matched nonsmoking population; the costs were virtually willing to pay to keep ourselves alive and healthy is the value identical (Leu and Schaub 1985). It should also be pointed out of health. On top of the economic loss to society due to that direct health care costs are resources that could be directed tobacco, one must consider the immeasurable suffering and loss to other uses and are therefore not real economic losses to deriving from the premature death of millions of individuals. society. The indirect costs of lost productivity, however, are The control of tobacco is one of the most important public real losses to society: contributions of energy and knowledge, health issues facing mankind, if not the most important. Future often in the years of peak productivity and income, have been generations will look back and wonder why it took so long for wasted. us to ban such an obvious hazard. 722 Kenneth Stanley References Lewit, E. M., D. Coate, and M. Grossman. 1981. "The Effects of Govemment Regulation on Teenage Smoking."Journal of Law and Economics 24:545-69. CDC (Centers for Disease Control). 1990. "Cigarette Advertising-United Mattson, M. E., E. C. Pollack, and J. W. Cullen. 1987. "What Are the Odds States, 1988." Morbidity and Mortality Weekly Report 39(16):261-65. That Smok ing Will Kill You ?" Arnerican Journal of Public Health 77:425-31. . 1991. "Smoking-Attributable Mortality and Years of Potential l1ife Muller, Mike. 1978. Tobacco and the Third World: Tomorrow's Epidemic., Lost-United States, 1988." Morbidity and Mortality WeeklN Report London: War on Want. 40(4):62-71 . Nelson, Hugh. 1986. The Economic Consequences of Smolang in Northern Chasov, E. I., R. G. Oganov, and 1. S. Glasunov. 1984. "Prevention of Diseases Ireland: A Cost-Benefit Analysis of Tobacco Production and Use in the Province. by Annual Screening of the Population" (in Russian). Sovietskaja Belfast:ActiononSmokingandHealth(NorthemIreland)andCommittee Sdravochranenie 10:3-6. of the Ulster Cancer Foundation. Cohen, Nicholas. 1981. "Smoking, Health, and Survival: Prospects in NRC (National Research Council). 1986. Environmental Tobacco Smoke: Mea- Bangladesh." Lancet 1:1090-93. suring Exposures and Assessing Health Effects. Washington, D.C.: National Collishaw, N. E., and G. Myers. 1984. "Dollar Estimates of the Consequences Academy Press. ofTobaccoUseinCanada, 1979."CanadianJournialofPublicHealth 75:192-99. OECD (Organisation for Economic Co-operation and Development). 1985. Doll, Richard, and Richard Peto. 1981. "The Causes of Cancer: Quantitative Measuring Health Care, 1960-1983. OECD Social Policy Studies 2. Paris. Estimates ofAvoidable Risks ofCancer in the United States Today."Journal Omar, Sherif 1987. The Economic Consequences ofSmoking in Egypt. CanceT ofthe National Cancer Institute 66:1191-1308. Institute, Cairo University, Cairo. FAO (Food and Agriculture Organization). 1982. "The Economic Significance Omran, A. R. 1971. "The Epidemiologic Transition: A Theory of the Epide- of Tobacco." ESC: Misc. 82/1. Rome. miology of Population Change." Milbank Memorial Fund Quarterly - 1988. Production Yearbook 1987. Rome. 4:509-38. - 1991. FAO Quarterly Bulletin of Stanstics 3(4), p. 61. Peachment, Allan. 1984. "Leaming frons Legislative Disasters: The Defeat of FAO (Food and Agriculture Organization), Committee on Commoditv Prob- the Westem Australian Govemment's Tobacco (Promotion and Sales) lems. 1989. "The Economic Significance of Tobacco." cc.P89/17. Rome. Bill." MedicalJournal of Australia 140(8):482-85. Flay. B. R. 1987. "Mass Media and Smoking Cessation: A Critical Review." Peto, Richard, A. D. Lopez, J. Boreham, M. Thun, and C. Heath. 1992. American Journial of Public Health 77:153-60. "Mortality from Tobacco in Developed Countries: Indirect Estimation from National Vital Statistics." Lancer 339:1268-78. FTC (Federal Trade Commission). 1988. "Report to Congress Pursuant to the Federal Cigarette Labeling and Advertising Act, 1986." Washington, D.C. Pierce, ]. P. 1989. "Intemational Comparisons of Trends in Cigarette Smoking Godfrey, C., and A. Maynard. 1988. "Economic Aspects of Tobacco Use and Prevalence." Americanlournal of Public Health 79:152-59. Taxation Policy." Bntish MedicalJournal 297:339-43. Ramstrom, L. M. 1986. "Worldwide Changes and Tends in Cigarette Brands Grise, V. N., and A. Clauson. 1985. "Costs of Producing and Selling Flue- and Consumption," In D. G. Zaridze and Richard Peto, eds., Tobacco: A Cured Tobacco." Tobacco Outlook and Situation Report. U.S. Department Major International Health Hazard. IARC Scientific Publications 74. Intema- of Agriculture, Economic Research, Washington, D.C. tional Agency for Research on Cancer, Lyons. Gu, X. Y., and M. L. Chen. 1982. "Vital Statistics, Health Services in Shanghai RCP (Royal College of Physicians). 1983. Health or Smoking? London: Pitman. County." Amencan Journal of Public Health 72:19-23. "Results of the School Health Education Evaluation." 1985. Journal of School Gupta, P. C. 1988. "Health Consequences of Tobacco Use in India." World Health 55:295-355. Smokingand Health 13:5-1 0. Rice, D. P., and T. A. Hodgson. 1983. "Economic Costs of Smoking: An Gupta, P. C., F. S. 11ehta, J. J. Pindborg, and others. 1986. "Intervention Study Analysis of Data for the United States." Paper presented at Allied Social for Primary Prevention of Oral Cancer among 36,000 Indian Tobacco Science Association Annual Meetings, San Francisco, DecembeT 28. Users." Lancet 1:1235-39. Roemer, Ruth. 1986. Recent Developments in Legislation to Combat the World Hammond, E. C., L. Garfinkel, H. Seidman, and E. A. Lew. 1976. "Tar and Smoking Epidemic. World Health Organization, Geneva. Nicotine Content of Cigarette Smoke in Relation to Death Rates." Envi- . 1987. Legislative Strategies for a Smoke-Free Europe. Smoke-Free Eu- ronmental Research 12:263-74. rope 2. World Health Organization, Copenhagen. Health Education Authority. 1991. The Smoking Epidemic-Counting the Cost Rouse, B. A. 1989. "Epidemiology of Smokeless Tobacco Use: A National in England. London. Study." Joumal of the National Cancer Institute 8:29-33. IARC (Intemational Agency for Research on Cancer). 1985. Tobacco Habits Shultz, J. M. 1986. Smoking-Attributable Mortalty, Morbidiry, and Economic Other Than Smoking: Betel-Quid and Areca-Nut Chewing, and Some Related Costs (SAMMEC). Center for Nonsmoking and Health, Minnesota Depart- Nitrosamines 1ARC Monographs on the Evaluation of the Carcinogenic Risk ment of Health, Minneapolis. of Chemicals to Humans 37. Lyons. Tobacco International. 1974. 186(4):717. 1988. Tobacco Smoking. IAR(' Monographs on the Evaluation of the Tomson, D., and A. Coulter. 1987. "The Bamboo Smoke Screen: Tobacco Carcinogenic Risk of Chemicals to Humans 38. Lyons. Smoking in China." Health Promotion 2:95-108. Jones, A., and]. Posnett. 1988. "The Revenue and Welfare Effects of Cigarette Townsend, Joy. 1987. "Economic and Health Consequences of Reduced Taxes." Applied Econiomics 20:1223-32. Smoking." In A. Williams, ed., Health and Economics. New York: Macmil- Kottke, T. E., R. N. Battista, G. H. DeFriese, and M. L. Brekke. 1988. lan. "Attributes of Successful Smoking Cessation Interventions in Medical Townsend,Joy, and the Advisory Committee on Health Education ofFinland. Practice-A Meta-analysis of 39 Controlled Trials." JAMA 259:2882-89. 1987. Tobacco Price and the Smohing Epidemic. Smoke-Free Europe 9. World Kristein, M. NI. 1983. "How Much Can Business Expect to Profit from Health Organization, Copenhagen. Smoking Cessation?" PreventiveMedicine 12:358-81. Tye, J. B., K. E. Wamer, and S. A. Glantz. 1987. "Tobacco Advertising and Leppo, K., and H. Vertio. 1986. "Smoking Control in Finland: A Case Study Consumption: Evidence of a Causal Relationship." Joumal of Public Health in Policy Formulation and Implementation." Health Promoton 1:5-16. Policy 8:492-508. Leu, R. E., and T. Schaub. 1985. Smokingand Health Care Costs: Plus or Minus? UNESCO (United Nations Educational, Scientific, and Cultural Organization). Proceedings of the 5th World Conference on Smoking and Health, Witinipeg, 1988. Child-to-Child: Another Path to Learning. UNESCO Institute for Educa- Canada, 1983. Vol. 1. Canadian Council on Smoking and Health, Ottawa. tion Monograph 13. Hamburg. Control of Tobacco Production and Use 723 United Nations. 1989. World Populauon Prospects 1988. Population Studies U.S. Office ofTechnology Assessment Health Program, Office ofTechnology 106. New York. Assessment. 1985. "Smoking-Related Deaths and Financial Costs." Upton, A. C. 1982. "The Biological EffectsofLow-Level[lonizingRadiation." Walter, H.J., R. D. Vaughan, and E. L. Wynder. 1989. "Primary Prevention Scientific American 246(2):29-37. of Cancer among Children: Changes in Cigarette Smoking and Diet after USDA (U.S. Department of Agriculture). 1986. Economic Indicators of the Farm Six Years of lntervention."Journalof the NationalCancer Institute 81:995-99. Sector, Production and Efficiency Statistcs, 1984. ECIFS 4-4, ERS. Washing- Warner, Kenneth. 1984. "Cigarette Taxation: Doing Good by Doing Well." ton, D.C. Journal of Public Health Polcy 5:312-19. .1987. World Tobacco Situation. Circular ET-4-87. Foreign Agricultural - . 1988. "The Tobacco Subsidy: Does It Matter?"Journal of the National Service, Washington, D.C. Cancer Institute 80:81-83. .1988. WorldTobaccoSituation. Circular F-8-88. ForeignAgricultural Whelan, E. M., M. J. Sheridan, K. A. Meister, and B. A. Mosher. 1981. Service, Washington, D.C. "Analysis of Coverage of Tobacco Hazards in Women's Magazines."Journal .1991. WorldTobaccoSituation. CircularFt-8-91. Foreign Agricultural of Public Health Policy 2:28-35. Service, Washington, D.C. WHO (World Health Organization). 1979. Controlling the Smoking Epidemic. USDHHS (U.S. Department of Health and Human Services). 1985. A Decision Technical Report 636. Geneva. Maker's Guide to Reducing Smokingat the Worksite. Office of Disease Preven- . 1980. "Sixth Report on the World Health Situation, 1973-1977. Part tion and Health Promotion, and Office on Smoking and Health, Washing- 1: Global Analysis." Geneva. ton, D.C. - 1983. Smoking Control Strategies in Developing Countries. Technical 1986. The Health Consequences of Smoming: Involuntarv Smoking. Report 695. Geneva. Report of the Surgeon General. DHHS publication (CLDC)87-8398. Office on . 1984. "Control of Oral Cancer in Developing Countries." Bulletin of Snioking and Health, Washington, D.C. the World Health Organization 62:817-30. .1988. The Health Consequences of Smoking: Nicotine Addiction. Report . 1985. "Cancer in Developed Countries: Assessing the Trends." World of the Surgeon General. DHIIS publication (cDc)88-8406. Office on Smok- Health Organization Chronicle 39:109-11. ing and Health, Washington, D.C. - 1986. Tobacco or Health. EB77/1986/RECIl. Geneva. . 1989a. Reducing the Health Consequences of Smoking: 25 Years of . 1987. 1987 World Health Statistics Annual. Geneva. Progress. Report of the Surgeon General. DHHS publication (cDc)89-841 1, Office on Smoking and Health, Washington, D.C. 1988a. "A 5 Year Action Plan: Smoke Free Europe." World Health .1989b. Smoking, Tobacco, and Health: A Fact Book. Office on Smoking Organization, Regional Office for Europe, Copenhagen. and Health, Washington, D.C. . 1988b. Smokeless Tobacco Control. Technical Report 733. Geneva. .1992. Smoktng and Health in the Americas. A 1992 report of the . 1991. "Tobacco-Attributable Mortality: Global Estimates and Pro- Surgeon General, in collaboration with the Pan-American Health Organi- jections." Tobacco Alert 4-5. zation. DHHS publication (cDxc)92-8419. Office on Smoking and Health, Wickstrom, B. 0. 1979. Cigarette Marketingin the Third World:AStudYofFour Washington, D.C. Countries. University of Gothenburg, Gothenburg, Sweden. Appendix B Reducing Mortality in Children under Five: A Continuing Priority Carl E. Taylor and Vulmiri Ramalingaswami In recent years, several international agencies have given ness to the basic needs of children is greatly magnified by highest priority to supporting programs for child survival and group response to effective messages in the public media. development in developing countries. This emphasis is justi- Communities seem more willing to correct social problems fied and should continue at least until the year 2000. Child when children are involved than to do so for adults. Political mortality is still the world's largest public health problem in leaders know their image among constituents can be im- numbers of individuals dying and years of life lost. The devel- proved by showing concern for the needs of childreni. In the opment of children determines the quality of future popula- past ten years, UNICEF's Child Survival and Development tions. Another reason children deserve priority is because child Revolution (CSDR) has stimulated unprecedented levels of health interventions tend to be the most cost-effective health intemational support, public awareness, and concem. area activities in all parts of the world. In this appendix we will * In more rapidly developing countries attention is shift- focus on selected issues relevant to the implementation of ing from straightforward concern about child survival to practical programs. child development. This includes efforts not only to im- Some specific justifications for maintaining priority atten- prove health and nutrition but also to promote intellectual tion on the problems of children are the following: development and learning. Obviously, such child develop- * Children under five years of age make up about 15 ment activities must give intersectoral emphasis to educa- percent of the population in most developing countries, and tion and strengthening the capacity and role of mothers. women in the reproductive age group make up about 20 The goal is not only that all children should be healthy but percent. In poor countries, young children have higher that they should be educated. mortality rates than any other age group-30 to 40 percent. * Child health interventions have a proven record dem- Morbidity and mnalnutrition are also high. Children in unhy- onstraring that their implementation is feasible in national gienic environments have to cope with the synergistic prob- programs. It has been repeatedly shown that a cluster of lems of numerous infections, to which they have to develop low-cost interventions can dramatically improve survival immunity at the same time that they are adapting nutrition- (UNICEF 1988). The methods can be implemented at home ally to rapid growth on limited diets. The United Nations or in peripheral health facilities. They can often be applied Children's Fund (UNICEF) annual State of the World's Chil- best through community participation, and they have great- dren reports have clearly demonstrated the many dimen- est long-term sustainability when they become incorporated sions of the massive need. Estimates and projections of into local cultural patterns. Education about a definable, mortality in the present collection provide detailed infor- locally relevant group of child care practices can produce mation on the relative importance of the principal diseases. continuing change in health habits that strengthe . l gen- It is worth noting here that the World Health Organization eral sense of self-reliance. (WHO) in conjunction with UNICEF has assigned highest * Becauseof nationalexpectations that internationalsup- priority to inimunizable diseases, the pneumonia-diarrhea portforcsm rwill continue, developingcountrieshave made complex, malnutrition, perinatal problems, and conditions long-term commitments. If international donors were to associated with maternal health. reduce support for child health and shift their attention to o Both nationally and internationally no appeal for funds adult problems, a serious loss of credibility would occur as and support can match the donor generosity stimulated hy mortality increased. Many countries have already become children in need. The pathos of a simple picture of a suffer- skeptical of international donor support because of its ing child stirs an eagemess to help. Individual responsive- Lunreliability." Continuity should be maintained until the 725 726 Carl E. Taylor and Vulmiri Ramalingaswaami promised potential has been fulfilled. For instance, very some Asian countries. It may be possible to break the cycle of large amounts of international funds are making rapid im- transmission and prevent the long-term sequelae of liver can- provements in child immunization levels. If these funds are cer and cirrhosis. reduced after the global targets are achieved, there is no way that costs can be covered by national budgets. It has been Diarrheal Diseases calculated that about half of the poor African and Asian countries about whom data were available will not be able The second priority intervention with which considerable to finance present types of immunization programs even progress has been made in the Child Survival and Develop- using the most optimistic projections of economic growth ment Revolution is in the use of oral rehydration therapy (ORT) up to the year 2000 (Rosenthal 1989). Any slackening of for watery diarrhea. In the 1980s 4 million deaths of children intemational funding will raise the dismaying prospect of from diarrhea were estimated to occur each year before current future epidemics in unimmunized populations, a situation programs were implemented, and that number can probably be reminiscent of the collapse of malaria eradication efforts in halved by the use of ORT. Dehydration can be prevented by the 1960s. Altemative and less expensive means of running early home treatment through the use of simple local adapta- immunization programs are available as a routine part of tions of traditional preparations, particularly those that are primary health care. Instead of expensive campaigns, it has cereal-based (Taylor and Greenough 1989). As with immu- been shown that periodic "pulse" activities are helpful in nizations, the most effective ORT programs are where primary increasing coverage. health care infrastructure provides backstopping and support. * A global ethic of concem for children seems to be Packets of oral rehydration salts, intravenous fluids, antibiot- developing. It is fragile, but long-term commitments should ics, and other appropriate treatment should be used by health follow passage of the United Nations Convention on the facilities for severe cases of dysentery and chronic diarrhea. It Rights of the Child. Promoting equity and disparity reduc- is especially necessary to increase efforts to stop the massive tion are difficult national and intemational goals, but these overuse of ineffective medicines. A growing priority is recog- concepts are more likely to be implemented in programs for nition of the close linkage between diarrhea and malnutrition. children than in most other activities. All children deserve Diarrhea is one of the most important causes of childhood everything that their society and international resources can malnutrition, and nutrient loss can be partially compensated provide, because children represent the future in every for by appropriate feeding during and after illness. country of the world. Diarrhea programs are also beginning to pay increasing * Calculations of the cost-effectiveness of child health attention to the promotion of preventive strategies. The measures almost automatically look good in comparison World Health Organization's Diarrhoeal Diseases Control with health care for adults. This is usually most apparent Program has sponsored useful analyses comparing the potential when calculations are based on years of life saved. effect of more than twenty interventions (Feachem 1986). Measles, which can be prevented by immunization, frequently causes diarrhea. When it becomes available, rotavirus immu- Some Priority Interventions nization will probably be highly cost-effective. Promotion of breastfeeding during the first four to six months of life, im- Recent experience has demonstrated that the following inter- proved personal and domestic hygiene through hand washing ventions deserve priority consideration in practical programs. and cleanliness, and better preparation and preservation of food are practical and low-cost methods of diarrheal disease Immunizations prevention. Sanitary disposal of excreta and improving the availability of water are of continuing importance. Particular Rapid progress is being made in worldwide efforts to immunize attention needs to be paid to the safe disposal of stools of children against the six diseases covered by WHO's Expanded children because they have the highest infection rates. Programme on Immunization (EPI): measles, poliomyelitis, diphtheria, tetanus, pertussis, and tuberculosis. The best pros- Acute Respiratory Infections pects for sustainable control are in situations where good primary health care infrastructure has been built up. Immu- Ranking with diarrhea in importance as a cause of death of nizations as part of primary health care should continue to have children are pneumonia and other infections of the lower high priority until herd immunity is acceptably high and respiratory tract. Although little has been done yet in large- immunization services can be integrated into continuing pre- scale application of new methods of control, it is possible now ventive services. As these services are stabilized, other types of to affirm with some assurance that approximately half of the 4 immunizations can be added according to local priorities. For million deaths of children from acute respiratory infection instance, as indicated in chapter 15 of this collection, frequent (ARI) each year can be prevented with available methods of transmission of the hepatitis B virus to babies from mothers casemanagement. lmmunizationswillpreventofsomeofthese who are type B carriers and the availability of an effective deaths. In addition to the vaccines for measles and pertussis, hepatitis B vaccine make this immunization a high priority in new pneumococcus and hemophilus vaccines will probably Reducing Morralirt in Children under Five: A Continuing Priority 727 become available in a few years, but they may not be affordable tions. In many situations it has become evident that simply in public programs for some time. Priority attention to the providing food is not sufficient to reduce malnutrition and that implementation of national programs for improved case man- interventions to control infections are also needed. Most stud- agement of lower respiratory tract infections in children is ies have focused on single nutritional interventions. Food overdue (Gadomski 1990). Methods have been developed to availability, food affordability, and food use in the family are train village health workers to make early diagnosis of pneu- all key issues that need to be addressed. It is necessary now to monia in children with cough by counting respirations and develop understanding of how to integrate services adapted to observing difficult breathing. Antibiotic treatment can then causal pattems in specific situations. be started promptly by the health worker using a simple proto- col. As with diarrheal diseases, this strategy requires an effec- Matermal and Perinatal Health Problems tive primary health care infrastructure to provide technical backstopping and professional support. A significant effort is The fact that 99 percent of pregnancy-related deaths in the needed to stop the overuse of antibiotics for upper respiratory world occur in developing countries is clear evidence that tract infections. Priorities for prevention have not yet been many of them could be prevented (World Bank 1987). For adequately defined. Improved matemal nutrition will increase both mothers and children, pregnancy and delivery are periods resistance in babies of low birth weight who have extremely of considerable risk due to the following types of problems: high neonatal pneumonia mortality rates. The severity of hemorrhage, sepsis, eclampsia, obstructed labor, and compli- respiratory infections in children seems to be influenced par- cations of abortion. Perinatal mortality among the poor is ticularly by parents who smoke and by smoke pollution in largely determined by delivery care and the maturity of the homes and urban areas. General factors such as crowding fetus, which is indicated best by the birth weight. It is estimated and poor housing also seem to be important in transmis- that 2.5 million deaths occur each year from perinatal causes; sion. More evidence is needed on practices such as swaddling, the long-term developmental defects from conditions such as which is common in some countries (for example, China and hypoxia during labor and birth trauma are more difficult to Turkey) and may interfere with respiratory function in new- estimate (WHO/UNICEF 1986). Control efforts are concentrated born babies. mainly on general preventive measures and improving case management through the use of high-risk monitoring (Back- Malaria ett, Davies, and Petros-Barzovian 1984). Preventive measures are focused mostly on improving matemal health, particular Globally, malaria control has long been given high priority emphasis beinggiven to nutrition of the mother. Deficiencies because malaria is an important killer of children and a cause in specific nutrients, such as iodine, iron, and folate, can of chronic disability and debility at all ages. Despite the col- seriously interfere with fetal development. Mothers who were lapse of the worldwide program of malaria eradication, control themselves malnourished in childhood tend to have small of malaria is one of the great success stories of international pelvises and stature and are more likely to experience compli- health. The massive reduction in mortality and prevalence in cations at the time of delivery. Sexually transmitted diseases Asia and the Americas has saved millions of lives, although and a variety of organisms in the genital tract can produce control has proved difficult in Africa and parts of Southeast serious sequelae when they infect babies (Bang and others Asia, where resistant strains of mosquitoes and parasites con- 1989). tinue to spread. Much of the improvement has resulted from Improved case management of pregnancy depends on a general environmental and socioeconomic change rather than health care system which can identify high-risk conditions specific health interventions. Malaria remains one of the great- early and arrange for the pregnant mother to be referred for est challenges both for the development of technology and for whatever special care is needed. This seems to work best when a delivery system (Breman and Campbell 1988). pregnancy care is part of primary health care and a high-risk surveillance system is adapted to the particular conditions and Protein-Energy Malnutrition resources available locally (Lettenmaier and others 1988). Initial screening by personnel who are trusted by mothers The high prevalence of protein-energy malnutrition is indi- should be readily available to foster the goal of achieving cated by the UNICEF estimate that 29 percent ( 165 million) of complete coverage of all pregnant women. The referral pro- the world's children are malnourished according to standard cess, then, will depend on local networks of health personnel weight-for-age criteria. Of these, 98 million children are in and facilities and also on communications and transport. South Asia. The prevalence of maternal malnutrition is most evident in the high proportion of babies with low birth weight. Breastfeeding It is estimated that about 16 percent of all babies bom each year worldwide, or 20 million, weigh less than 2,500 grams at Good lactation is the most natural of health interventions. birth. Control of protein-energy malnutrition continues to Cessation of breastfeeding is, however, part of modemization deserve high priority because of its effect on child development trends in many areas of the world. Abundant evidence shows and because of synergistic two-way interactions with infec- that in the poorest communities the effect of declining 728 Carl E. Taylor and Vulmiri Rarnalingaswani breastfeeding rates is disastrous with regard to child mortality Some people have criticized UNICEF in recent years because and nutrition. Its protective value is shown by the fact that its child-survival strategy is said to be based on a "selective" breastfed babies are relatively healthy for the first six months disease-oriented analysis of priorities. The fact is, however, of life and also by the good start they get for subsequent that UNICEF's promotion of the Child Survival and Develop- development. Breastfeeding is also the world's most ubiquitous ment Revolution had its origin in the Alma Ata principles of method of family planning. getting complete coverage for equity in primary health care. To this was added focused efforts to accelerate action in the Birth Spacing home and community (Taylor and Jolly 1988). An important distinction is that UNICEF's GOBI (growth monitoring, oral rehy- Many studies have shown that family planning is an effective dration, breastfeeding, immunizations) priorities are interven- means of improving child health (IPPF 1988). It is usually tions and not specific diseases. They were selected largely associated with social and economic conditions in the home because they were simple and low cost, had high effectiveness, associated with being a "wanted child." Intemational programs and were suitable for mass implementation because they used for family planning, unfortunately, have often been dissociated methods which stressed social mobilization and self-reliance. from health services. For eventual sustainability the two should They resulted in high expectations among political leaders be naturally linked, especially because this is the way mothers who understood that these programs would give them oppor- think about their problems. In a two-way interaction, family tunities to be involved in activities which had potential for planning directly contributes to better child health, and child public recognition at low cost (UNICEF 1988). The expectations survival increases motivation to practice family planning. UNICEF created in persuading national leaders to support the Child Siurvival and Development Revolution are now coming Defining Local Priorities due as the people expect to benefit from the significant improve- ments that were promised. Great effort is being expended in It has become increasingly evident that only a few interven- trying to meet deadlines for ambitious targets that were set on tions, such as immunizations and commlLunity environmental the basis of limited analysis. The overall effort will have control measures, are sufficiently widespread around the world benefits even if particular interventions do not avert the to justify their being given global priority. Once those pro- number of deaths that was originally projected. The point. grams have been introduced, the rernaining priority problems however, is that without the ambitious targets, much less will vary greatly, depending on local conditions and available would have been achieved. In the 1990s there should be more resources. Examples range from infections such as malaria, emphasis on building sustainable infrastructures. hemorrhagic dengue. and Japanese B encephalitis to localized Practical programs for control of the main communicable prevalence of kwashiorkor and iodine deficiency. diseases of children require a combination of research and The systematic setting of priorities in health care on the application. A common assumption of research workers has basis of cost-effectiveness criteria will almost automatically been that all they need to do is to make new findings and give greater emphasis to prevention than to curative care. This technology available, and then implementation will follow is especially true for the communicable diseases which contrib- spontaneously. This is not necessarily true, as experience with ute most to childhood mortality, morbidity, and developmen- smallpox has shown. Even though an effective vaccine was tal deficit. available, much persuasion and field research had to be done Most decisions about priorities require balancing concerns for a very long period of time before countries were willing to about the technological effectiveness of specific procedures make serious efforts to control the disease. The chapters in this and the feasibility and cost of using them. A health problem collection that are concerned with the high-priority diseases that has high priority because of high rates of mortality and of childhood show the complexity of determinants of effective morbidity may be given low priority because no effective control even with the simplest interventions. Mass programs control measures are available. Our greatest limitations seem have had some clear successes, but the limitations of looking to be in community-based delivery systems and in the process just for simple solutions have also become evident. As initial of adapting procedures to local conditions. successes have been achieved, greater effort is needed to iden- Decisions about priorities concerning what will actually be tify means of promoting sustainability and cost-effectiveness done are always influenced more by administrative and polit- adapted to the community level. This requires new methods ical issues than by epidemiological information about diseases. for adapting implementation to the particular cultural, socio- The interventions focused on in this collection are selected economic, and administrative traditions of local groups and mainly because of potential efficacy for particular diseases, but service sectors in what has been called country-specific health there is also a need for more general types of information. Even research (Commission on Health Research for Development though it may be appropriate to start a priority-setting process 1990). by considering the cost-effectiveness of disease-specific inter- ventions, it then becomes necessary to consider how those Determinants of Successful Child Survival Programs interventions can be integrated in a total health system and what other resources are available within local socioeconomic The ultimate need is that all special child survival programs constraints. should help to build a primary health care infrastructure. If any Red uczng Mortalitv in Children under Five. A Contiiuinig Prioritvy 729 intervention or technology is introduced as part of a special enforce mass implementation using legal or centralized reg- program, it should eventually be integrated into the local ulatory controls. To make these methods acceptable to the infrastructure except in the rare instances in which eradication public, the evidence for effectiveness and safety must he so seems feasible. The long-standing confrontation between pro- indisputable that the people accept the principle that social ponents of vertical and horizontal programs should be viewed good should take precedence over individual choice. Social from the perspective that a balance is needed (Taylor and Jolly mobilization can be applied through community pressure to 1988). Setting priorities implies giving focused emphasis for a enlist the compliance of individuals and groups. period of time to particular activities. These will not have * Case management interventions (ORT, ARI, monitoring of sustainability. however, unless they are built into continuing high-risk pregnancies and growth). In a two-phase process, services based on community priorities. initiative is taken first by the health system in setting up To achieve sustainability it is essential from the beginning screening or surveillance procedures to identify specific to have programs that are based on sociocuLirural appropriate- health problems early, or individuals and groups who are at ness and sensitivity. Community relationships should be highrisk.Thesecondphaserequiresanappropriateresponse developed in ways that ensure acceptability, accessibility, ad- by the individual and family to apply preventive or correc- ministrative feasibility, and continuing financing. tive interventions. A strong educational process is necessary Any international initiative should not impose outside pri- to stimulate awareness and motivation to act. Equally im- orities but be responsive to continuing dialogue between portant is clear definition of what support is necessary from national and local expressions of effective demand. It is coun- health services and reliable logistics to make appropriate terproductive to push ahead in global initiatives without ad- drugs, equipment. and supplies accessible and available. justing to local perceptions of community priorities. These responses can range from implementation of inter- The most profound and long-term changes in child survival ventions that can be readily applied in the home to knowing and development will occur as a result of behavioral changes when to go to a health facility. Examples include knowing applied in family patterns of child care. Simplified procedures the right amount of salt to add to dilute rice porridge for should he established to introduce new home methods in daily simple home-based Or, having readily available co-trim- routines such as hand washing, oral rehydration, or improving oxazole for childhood pneumonia that has been diagnosed weaning methods. In order for these routines to be accepted as by trained community health workers according to the WHO parental behavioral norms, there will need to be strong social protocol for ARI, having easy access to weighing scales for support for individual families. For instance, the goal should growth monitoring, and testing blood pressure as part of be for every case of watery diarrhea to he managed by mothers routine prenatal examination. with home-based simple fluids to prevent dehydration. * Primarv previention in the home Some of the most positive New strategies are needed to combine interventions into and lasting changes in promoting better health and nutri- rationalized packages of services as part of primary health care. tion result from basic behavioral changes in routine child Such packages should have great cost-effectiveness because of care. These depend on family initiative and represent one shared costs for programs which have multiple benefits. Little of the ultimate goals in improving child survival and devel- has been done to define entry points at which one type of opment. Examples include personal hygiene and home san- intervention can facilitate the introduction of other interven- itation, preventing the exposure of children to home air tions. Problems in measurement methodology need to be pollution and passive smoking, diets for long-term healthy resolved to develop means of calculating cost-effectiveness in development, and appropriate stimulation for intellectual cases in which input, output, and outcome variables overlap. development of babies. Sorely needed is quantitative analysis of integrated services seen in the successful national experiences in China and Sri Lanka. It may mean that we should set up new multipurpose Competing Concerns studies on integrated services (wlno 1986), which would likely have to be done in relatively small but representative popula- Two competing concerns must be balanced. The first is the tions exemplified in the district strategy advocated by WHO. need in almost all programs to increase community participa- tion through decentralization. This requires local setting of Types of Child Survival Interventions priorities and dialogue between community leaders and health system workers to ensure technical quality of services, logistics, training, and supervision while taking into account local de- A simple categorization may contribute to one's understanding sires and priorities. Community participation can be a strong of how to focus activities. Three strategic models can help in force in promoting sustainability. deciding about how particular child survival interventions can The second concer is to ensure equity by setting up at- be most effectively implemented. rangements so that services reach those in greatest need. By * Preventive interventions organized by public or government targeting high-risk groups for special attention, a program services (EPI, community uwater supplies, regulations to control should be able to reach those who have the highest prevalence epidemics). Some interventions may be considered so im- of disease. It is only through improving their health that a portant that public health authorities take the initiative to significant effect on morbidity and mortality can be achieved. 730 Carl E. Taylor and Vulmiri Ramalingaswami It is, however, especially hard to reach those in greatest need Program development can be greatly facilitated by setting up because they tend to be suspicious and poor and they live in a leaming process to find the best way of organizing locally places with difficult transport and communication. Their ac- appropriate services. Problems which arise in day-to-day activ- cess to care is limited by their own time constraints and ities can be brought to the experimental area for study in the well-established social and economic barriers. They are typi- field. These field studies should progressively advance knowl- cally bypassed because they do not know where or how to get edge so that there is incremental learning of what works under care. Getting care is not only inconvenient; local arrange- local circumstances. Regional linkages between such experi- ments and the arrogance of health workers often violate local mental areas and field research teams can be coordinated in a cultural patterns. national network to build capacity and to provide mutual The conflict between the above two concems arises because support. of the ease with which local leaders can manipulate the process of community participation and priority setting. Local leaders responsible for community participation usually make sure that References benefits go first to their family and friends. Outside involve- ment in priority setting may help to ensure equity in coverage. Backett, E. M., A. M. Davies, and A. Petros-Barzovian. 1984. The Risk If health service systems set up standard measurement methods Approach in Health Care, with Special References to Maternal and Child Heakh as part of community-based surveillance, it should be possible including Family Planning. Public Health Papers 76. WHO, Geneva. to determine who in the community is in. greatest need. Then Bang , R. A., M. Baitule, S. Sarmukaddam, A. T. Bang, Y. Choudhary, and 0. to determine who in the community Is in.greatest need. AThen Tale. 1989. "High Prevalence of Gynaecological Diseases in Rural Indian dialogue between community elite and responsible health Women." Lancet 1:85-88. workers can allocate resources on the basis of data about Breman, J G.., and C. C. Campbell. 1988. "Combatting Severe Malaria in relative need. Having to meet clear coverage targets set by African Children." Bulletin of the World Health Organization 6:611-20. national programs requires health workers and local leaders Commission on Health Research for Development. 1990. Health Research: to make sure that special services get to all, especially the Essential Link to Equity and Deveiopment. Oxford: Oxford University Press. population pockets which have the greatest need. Based on Feachem, R. G. 1986. "Preventing Diarrhea: What Are the Policy Options?" systematic monitoring of the priority diseases in an area, the Health Policy and Planning 1: 109-17. selection of an appropriate mix of priority interventions should Gadomski. Anne,ed.1990. AcuteLowerRespiratoryInfectionandChildSurvival in Developing Countries. Baltimore: Johns Hopkins University, Institute for become a responsibility shared among the health service work- Intemational Programs. ers, the community participants, and representatives of other IPPF (International Planned Parenthood Federation). 1988. Better Health for sectors. Women and Children through Family Planning: Report of the International Conference, Nairobi, Kenya, October 1987. London: IPPF. Networks for Health Services Development Lettenmaier, C.. L. Liskin, C. A. Church, andJ. A. Harris. 1988. Mothers'Lives Matter: Maternal Health in the Community. Population Reports 16(2). Johns As primary health care infrastructure develops, one of the most Hopkins University, Population Information Program, Baltimore. important skills to be incorporated in local capacity building Rosenthal, Gerald. 1989. The Economic Burden of Sustainable EPI Implications isothe ab ilt c m t be ,r a in loa , p i buiding for Donor Polcy. Arlington, Va.: John Snow, REACH (Resources for Child Is the ability in decentralized units to work with communities Health). in setting local priorities. An effective means for developing Taylor, C. E., and W. B. Greenough. 1989. "Control of Diarrheal Diseases." this capacity is to organize a network of linkages between Annual Review of Public Health 10:221-46. academic centers and local health services (WHO 1986). This Taylor, C. E., and R. Jolly. 1988. "The Straw Men of Primary Health Care." gives institutions with capability in operations research and Social Science and Medicine 26:971-77. planning an opportunity to work out practical solutions for UNICEF (United Nations Children's Fund). 1988. The State of the World's local health problems in experimental areas. Solutions that Children 1988. New York: Oxford University Press. have been adapted to local conditions can then be generally World Bank. 1987. Preventing the Tragedy of Maternal Deaths. Report of the implemented in a systematic extension process. This strategy Intemational Safe Motherhood Conference, Nairobi, Kenya, February has the potential of promoting both community participation 1987. Washington, D.C. has potetiabl pistromotion bnot comn tys part wic ipai WHO (World Health Organization). 1986. National Heakh Development Net- and equitable distribution. Information systems with rapid works in Support of Prinmary Health Care. Geneva. feedback to local implementation are a critical component. V'WHO/UNICEF (World Health Organization/United Nations Children's Fund). Few preventive programs can be completely standardized 1986. Maternal Care for the Reduction of Perinatal and Neonatal Mortality. because they all need to be adapted to local circumstances. Geneva: WHO. Appendix C Priority Setting for Health Service Efficiency: The Role of Measurement of Burden of Illness Gavin Mooney and Andrew Creese The need to set priorities arises from the fact that not all plague epidemics and natural disasters were made in the sev- illness can be eradicated nor all needs met. This failure to be enteenth century by the English physician William Petty able to meet all needs arises not principally because of the (1699). Theepidemiological and economic tallyofdiseases on limitations of technology-the technology is currently avail- a national or global basis has been documented more recently able to-eliminate many of the most important diseases, such as in an empirical work by Walsh and Warren (1979). Accounts poliomyelitis and measles-but because of the scarcity of of the costs of individual health problems, such as road traffic resources. Policymakers in the health sector have to manage accidents or, more recently, acquired immunodeficiency syn- resources in ways that maximize health outcomes,whether this drome (AIDS), are regularly published in journals concerned means redeploying resources, allocating limited new resources, with health policy. (See, for example, Henke and Behrens or cutting back on the use of existing resources. They must also 1986; also see the subsequent debate on the cost of illness: get the most out of whatever they have available, which is Shiell, Gerard, and Donaldson 1987; Behrens and Henke likely to mean changing the mix of resource allocations. 1988; and Hodgson 1989.) The motivating factors, sometimes Priorities are about change. Setting priorities to achieve best implicit, sometimes explicit, behind such analyses appear to be possible value for the resources available should be based on an assessment or reassessment of priorities. Measuring the considerations of both benefits and costs. Using scarce burden of illness is thus seen as an ingredient in the rational resources in any way means, by definition, giving up the setting of priorities. opportunity to use them in some other way; providing benefits The reason for attempting to measure the burden of illness here means forgoing them elsewhere. Priority setting means is thus to allow a better (that is, more efficient) use of scarce developing analyses and procedures to ensure that the policies resources in reducing the effect of illness on a population, a that get priority (that is, those which get a higher call on extra group of individuals, or even single individuals. In some in- resources) are the ones that provide the greatest benefits per stances, such as when an epidemic or an important new disease additional dollar spent. If the dollars could have been better manifests itself-and AIDS is a classic case-awareness of the spent elsewhere, then they should have been spent elsewhere. burden of illness in itself forces a reassessment of expenditure In this appendix we illustrate first how, conceptually, infor- priorities. But even in this instance there is a need to assess the mation on the burden of illness can contribute to the process benefits and costs of different policy reactions and also to of priority setting. We then identify some of the practical review the priority status of the new problem with the same problems entailed in deriving appropriate information on both criteria used to measure illness problems that have been more the costs (briefly) and the outcomes of health interventions. long standing. We also consider here the usefulness and limitations of the One cannot examine issues of efficiency, however, without dollar cost per disability-adjusted life-year (DALY) gained. Finally, looking at both inputs and outputs-costs and benefits. The using the cost and outcome information summarized for indi- burden of illness, however measured, is not a particularly useful vidual chapters, we give examples of how a cost-effectiveness concept if it is assessed separately from the question of the strategy may be used in setting health priorities at sector, policies and resources associated with addressing that burden, project or program, and clinical levels of the health system. that is, questions of how effective and how costly different forms of treatment, care, or prevention are in dealing with the The Burden of Illness and Priority Setting illnesses being considered. Looking solely at the disease or illness side of the equation, The notion of illness as a social and economic burden is very and not simultaneously at the resource or input side, does not old. Quantitative estimates of society's losses from bubonic permit one to say anything conclusive about the assessment or 731 732 Gavin Mvoaex and Andrew Creese evaluation of priorities. Conceptually, it is necessary to give that if there is a need to choose between spending w on program some consideration to the relationship between the burden of ( and spending the same amount on program D, then for the illness and the effect that different treatment, care, or preven- sake of efficiency (what is called allocative efficiency) the tion interventions have Uponl it. investment should he in the program in which the benefit is The need to consider the disease and resource sides of the larger. The question then is, where can an increase in resources efficiency equation together is not, of course, an argument be deployed to decrease the burden of illness to the greatest against measuring the burden of illness; it is only an argument extent? It should be noted that the question in the other against the belief that the epidemiology of illness in itself is a direction is also relevant: where can a cut in resources be hasis for priority setting and against the idea that, in general made so that the increase in the burden of illness is minimized? (there are exceptions), measuring the total burden of an illness Here the relevant techniques are cost-utility and cost-benefit is a valuable thing to do. Thus if, for a particular country the analyses. burden of morbidity and premature mortality from childhood The ideal with cost-benefit analysis is to operate with the infections was greater than the burden of adult respiratory three key rules for allocative efficiency: disease, this in itself would tell us nothing about the relative Ifforaparticularprogramcostsaregreaterthanbenefits, resource allocation priority of these two problems. Assuming for the momnent that adult and infant lives are weighted of then that program should not be implemented. equal importance, conisideratiotns of the cost and effectiveness * lfbenefits are greater than costs, proceed with the policy. of available technology for altering the course of the disease But further, and ideally, these rules should be applied at the must still be introduced before the overall cost-effectiveness of margin. interventions can be ranked. Furthermore, in most instances * In other words, a policy should be pursued up to that it will not be the total costs and the total benefits (in the point where the marginal benefit equals the marginal cost- context of the burden of illness, the latter will normally be but not beyond that point. estimated as a reduction1 in that burden) that are relevant-or Such rules are made in recognition of the scarcity of resources indeed the average costs and benefits. (we cannot do everything) and the importance of efficiency The prime concern is with assessing change. If resource (we accept that society should attempt to provide as much inputs in one program are increased, to wlhat extent is the benefit as possible with what resources are available; for good burden of illness in that program reduced? If inptuts are in- reviews of economic appraisal, see Mills 1985 and Drummond creased again, how much more is the burdeni of illness reduced? adtd others 1986). Conversely, if resource allocations to a program are decreased, These issues can be summarized in the following five points: to what extent will the burden of illness increase? Economic thinking of this sort has clearly establislhed that what is rele- * Measuring the burden of illness is an important ingredi- vant in the setting of priorities are the marginal benefits and ent in rational priority setting. the marginal costs and consequently the marginal effect on the * Rationally set priorities are obtained by a process of burden of illness of an increase or decrease in the resources weighing costs and benefits, and benefits are obtained deployed in that program. largely through a reduction in the burden of illness. Thus the prime objective of efforts to estimate the burden * Priorities are set on the margin: it is the costs and benefits of illness is Lest seen in the context of attempting to estimate of change that matter. Accepting this leads to such ques- the reduction in the burden of illness through the application tions as, if resources are increased, where can they be used of some treatment or preventive regime which inevitably to reduce the burden of illness most? involves the use of scarce resources. Such efforts are a means a Priorities are not a function of total costs or total bene- toward allowing the quantification of the effectiveness of a fits, which means consequently that rational priority setting particular policy on a particular disease and help in answering has no interest in the total burden of an illness unless it is the question, For illness x, does treatment A do more good practical both technologically and economically to elimi- (reduce more the burden of illness x) than treatment s? nate that illness-and such instances occur very seldom. Thereafter the issues of operational efficien-cy can be ad- * There are two relevant forms of efficiency in priority dressed: how best, with regard to the cost per unlt of output, setting: operational efficiency when the priority questions can the burden of particular illness x be reduced? Here the relate to how; allocative efficiency when they relate to relevant techniques are cost-effectiveness analysis and cost- whether and how much. utility analysis, the latter having the advantage over the former of being able to consider more than one type of output (for example, both mortality and morbidity reductions). This is Measuring the Effectiveness of Interventions discussed in more detail later. A compariso n of the burden ot illness across different dis- Our main concerns in this appendix pertain to measurements eases, if such is possible, leads into still more interesting ques- of the burden of illness. Because of the importance attached to tions of the relative efficiency of using resources to deal with the costs of reducing the burden of illness, however, we briefly the effects of different illnesses. The clear implication here is look first at some issues of cost measurement. Pnorirs Setting for Health Service Efficiency: The Role of Measurement of Burden of Illness 733 Information on Costs health status and illness are multidimensional, involving phys- ical pain, physical impairment, mental disability, mortality, Several key principles are associated with all costing. First, and so on; second, health status is a value-laden concept; and what we seek to measure are the so-called opportunity costs, third, the appropriateness of one particular measure is likely to that is, the benefits forgone in the best alternative use of vary, depending on why it is being used. Infant mortality could resources. Where markets work well, market prices can often he a reasonable basis for comparing the effect of child immu- be used in estimating costs. In the health care sector and in nization programs across different countries. It would not be a developing countries generally, however, the frequent market suitable measure for the effectiveness of an antismoking cam- failings or distortions mean that "slhadow pricing" is required. paign among schoolchildren. Second, the relevant cost is always the cost of the change There is also a hierarchy of measurement which has to be being considered, and this can normally be defined as the noted. If all that is of interest is to answer the question "Is x marginal cost. If, for example, it is expected that a hospital will more effective than Y?" then an "ordinal" ranking is all that is have to deal with an extra hLIndred births next year, the required (that is, we can rank the relevant change in the relevant cost relates to the extra use of resources tfor staff, burden of illness as greater or less). If we want to go further and equipment, anid other resources for these births. (It should be say that a quantified amount more is obtained, then "cardinal" noted that this cost may have no similarity to the existing scaling is necessary. average cost per birth in the hospital.) In most contexts, cardinal scaling is necessary in priority- Third, the cost should normally include all resource use, no setting exercises because it is not enough to be able to say matteron whom it falls. Thus itis not just health service or public that x is more effective than Y-especially if x is also more sector costs that are relevant hut also costs falling on private expensive than ); we need to know how much more, the issue agencies, the patients themselves, their relatives, and so forth. of cardinality. Fourth, payments for sickness benefits, pensions, and the like are not costs as such but rather transfers from one grouip HEALTH CARE ACTIVITIES The most basic methods used in in the community (normally the working population) to an- measuring health care outputs are activity measures, such as orher (here, ill people and the elderly). These redistributions numbers of cases treated, numbers of consultations, and pro- of resources are not costs from society's point of view. They are porrion of population vaccinated, which do not directly mea- called "transfer payments." sure health at all. Of course, it is reasonable to assume that the At a more practical level, one of the great difficulties in more patients who are treated the greater will be the benefit making estimates of costs according to the above principles is with regard to reduction in the burden of illness. But that the paucity of existing data. What are often available from assumption requires various other assumptions about the effec- accounting data are average costs-and yet it is not these that tiveness of intervention, which it would he preferable to mea- are required. Because they are available, however, there is a sure more directly. For example, to couch the effectiveness of great temptation to use them. We would counsel against this a family planning campaign in terms of the proportion of and suggest that crude margillal cost estimates are better than women reached in the campaign may be a poor measure of the precise average costs. effect it has on family planniing per se or, more explicitly still, If rhe use of average costs is ro be rejected and marginal costs on rhe number of unwanted pregnancies conceived. calculated, how is this best done7 The answer is, quite simply, to ask the appropriate people for their estimiiates. Thus, in extending HEALTH INDEXES. The simplest methods which incorporate care to take account of an extra hundred births next year in a some assessment of health status involve using estimates of particular hospital (as in the example above), the starting point mortality or life-years lost. It is clearly the case, however, that is to ask the hospital manager or obstetrician what facilities and these estimates then ignore morbidity and any other aspects of resources will be needed to cope. An estimate can be made of the the burden of illness. Of course, there may be some situations extra time of doctors, of nurses, of auxiliarv staff, of equipment, of in which it is possible to justify such ignoring-for example. in food, and the like-and then each of these resources costed. That certain instances in which mortality and morbidity are highly then gives the relevant cost figure. correlated. Generally, however, such measures are of rather The lack of adequate, readily available, marginal cost data limited value. In the field of clinical or individual health status in health care (and not just in developing countries) is perhaps measurement, several different types of index exist. Such just as big a problem as the lack of good outcome measures. It indexes are important, because the objects of such measure- is normally easier, however, to overcome the problems on the ment are those on which population-based health status mea- cost side and get a sufficiently accurate estimate of the relevant sures should be based. For assessments of levels of physical and marginal costs. social functioning of individuals, see, for example, the Duke- UNC Health Profile (Parkerson 1981); the Sickness Impact Information on the Effectiveness of Health Interventions Profile (Bergner and others 1981); the Index of Well-being (Kaplan and BLish 1982). For a general review see Hall and It is clear that the measurement of the burden of illness is Masters 1986. For population-based measures, the review difficult. This is true for three reasons: first, the effects on method of Walsh and Warren (1979) is worth noting. Still, 734 Gavin Mooney and Andrew Creese the authors made no attempt to aggregate the morbidity and deals with totals but also that it deals with averages. We have mortality components of health and simply presented ordinal already considered the need to concentrate on the margin, and rankings of the main diseases, first by their morbidity and in some instances marginal benefits or marginal costs may turn second by their mortality. out to be closely approximated by average benefits or average More recent attempts to combine both types of information costs, respectively. But they may not. in a single aggregate have counted both avoided disability and Thus, and for example, in estimating the value of a death avoided mortality as the number of days of a "normal" life prevented, the relevant formula in the calculations of the gained. This has provided a common yardstick with which Ghana Health Assessment Project Team, when including an morbidity or spells of temporary incapacity can be arithmeti- allowance for length of survival, considers only the average age cally combined. at onset and the average age at death. It follows that the value The Ghana Health Assessment Project Team's calculations of a death prevented is then always calculated on the basis of of "healthy days of life" are of particular interest in the context the average years of life extended by the program. Yet if there of attempting to use burden-of-illness data in the setting of is considerable variability about these averages, then priorities priorities, particularly with respect to the need to be cautious may be wrongly set. Thus, for example, if we consider the in using such data. In this study, an index was developed for priority to be attached to screening women over the age of measuring days of healthy life lost to selected diseases which twenty for breast cancer and use only average figures, the involves the assumption that days spent being dead, being average increase in life expectancy is small. But if we then look permanently disabled, and being temporarily disabled are at specific age groups or risk groups, the position will be better equally valued. That seems a difficult assumption with which for some, worse for others. Thus using average figures is likely to agree, but provided the sensitivity (see the section on to lead to a misallocation of scarce resources. We have in this uncertainty below) of such assumptions is tested, then such example two lessons to be leamed: the danger of using total apparently gross assumptions may be defensible. The point is burden-of-illness data; and the danger of using average burden- that they ought to be tested-by, for example, determining of-illness data. what difference it makes if the weight attached to being A more recent study, in which program cost information is disabled is 0.5 compared to a weight for death of I. juxtaposed with an aggregate measure of effectiveness, is that Certainly such a method is valuable, provided its limitations by Prost and Prescott (1984) on onchocerciasis. The authors are recognized and provided it is not used to rank priorities in estimate the cost-effectiveness of prevention measures for terms of the total burden of illness. Bamum (1987), for exam- onchocerciasis using the altemative measures of effectiveness ple, makes the very relevant point that weightings should be reproduced in table C- 1. applied to estimates of lost healthy days, first, to reflect the Given the emphasis here on added benefits and on the time dimension (that is, the discounting of losses in the future) sensitivity of the results to different measures of effectiveness, and, second, to reflect productivity loss. Even here, however, this type of empirical work is potentially very useful. As Prost there are problems because the implication of the productivity and Prescott themselves state, however, "the relative cost- loss measure is that anyone older than fifty-seven years has a effectiveness of onchocerciasis control is very sensitive to the zero value. (Here we have a variant of the human capital choice of effectiveness measure" (1984, p.801). It is thus clear method of estimation without any attempt being made to that there remain problems in improving such measurement of avoid the problems of zero-weighting retired people.) burden of illness to allow relevant measures of effectiveness to Barnum states: "The results [provided by this approach] be designed. illustrate that weighting and discounting, and their interac- We accept that this is difficult, but it is what is required for tion, potentially ... affect the priorities and strategies that rational priority setting, and no amount of concem about lack evolve from an epidemiological analysis of the health sector" of data or about the problems involved in such development (1987, p. 838). This interpretation, however, gives the impres- will make the basic requirement change. It is, in our judgment, sion that the commentator is assuming that the total burden much better to attempt to adopt this methodology in some of different illnesses in itself provides some basis for setting form or other even if we get no closer than a crude approxima- priorities. As we have argued above, it does not, except in some tion than to adopt what are clearly wrong or inappropriate very restricted circumstances. (It may be that such total mea- measures. sures will have more relevance in setting research priorities in A related approach, hitherto used only in industrial coun- situations in which the relative size of a problem is the only tries, entails the adjustment of the quantity of additional days basis for setting priorities because we know nothing about of life by a factor designed to capture (and make comparable) either the different costs of research in different areas or the the dimension of quality. In comparing renal dialysis with different probabilities of success and therefore have to assume kidney transplantation, for example, as options for patients that neither varies with the illness. Such assumptions may be with end-stage renal failure, it is clear that a simple comparison an approximation of reality in setting research priorities; that of the dollar cost per case would fail to capture the superior is unlikely to be true, except infrequently, in the case of health quality of outcome of successful transplantation over dialysis. care policy priorities.) This factor is apparent to all-clinicians, patients and their A further problem revealed by the method used by the families, and potential patients, that is, the public. The out- Ghana Health Assessment Project Team is not only that it come resulting from transplantation is clearly "better" than Priority Settingffor Health Service Efficiency: The Role of Measurement of Burden of Illness 735 Table C 1. Cost-Effectiveness of Onchocerciasis tackling the question of how to measure burden of illness using Control health status or DALY measures. All are concemed with at- (U.S. dollars) tempting to quantify different health states-such as uncon- Ulnit of measure Cost sciousness, severe physical impairment, moderate pain-on a scale stretching from perfect health (given a weight, say, of 1) Per year of healthy life added 20 to death (weighted, say, as 0). Thus a year of life with, for Per productive year of healthy life added 20 example, significant physical impairment and considerable Per disability-adjusted year of healthy life added 150 pample thought tbeoly 8mpercent as godsayer Per discounted productive year of healthy life added 150 pain might be thought to be only 80 percent as good as a year - _______- ______-_______-______-_______-______- -_____- of perfect life. In such a case, the DALY for this health state Source: Prost and Prescott 1984. would be 0.8. Various instruments are available to assist in the attempt to measure DALYs. These include the following: that from dialysis. Although both options prevent premature * Theratngscalenormallyconsistsofalineonapagewith death, the difference in the quality of survivors' lives necessi- agscale from,y sonendtsoIa the othege th tate adustmnt o th nuber f yars aind torefect his a scale from, say, 0 at one end to 1 at the other, the end tates adjustment to the number of years gained to reflect this, points being defined as death and perfect health, respec- so that the outcomes are comparable. tively. Other health states are then placed at different points This approach, it has to be emphasized, tells us only about t he healtht ar the place at eren pits the relative burden of one disease as compared with others. Its o healie a or in th m b e d primary use is in attempting to rank for the purposes of priority setting the costs per disability-adjusted life-year (DALY) gained * The standard gamble involves a choice of the certainty of on the margin of different programs. Thus if one program has a health state Y as opposed to the probability of a health state an extra cost of$10,000 per DALY gained and another program x (where x would normally be preferred to Y). If x were has an extracostof$100,000, it wouldbe rational, if there were perfect health (weight as 1) and the probability which made no other considerations, to invest in the first program, because the valuer indifferent in this choice were 75 percent, then the number of DALYs gained would be greater. the DALY for Y would be 0.75. The fundamental problem with DALYs as with all such mea- * The weights of the time trade-off are determined by offer- sures of health status is in getting the appropriate weights for ing choices of different lengths of life in different health mortality and for all the possible forms of morbidity. Questions states and attempting to get "indifference" across different here relate to whom to ask to do the valuing; how one life is choices. to be compared with another-normally assumed to be the The actual use of DALY data linked to costs is provided in same; how to allow for uncertainty; and many other issues, table C-2, based on work by Williams (1985). Essentially what including whether the only output of health services is im- this means is that given an additional amount of money, say, proved health status. Because this last point is a concem with 14,000, (approximately $25,000) to spend on the listed pro- all the methods of measuring the burden of illness in this grams, spending it on pacemakers would give twenty DALYS, appendix, it will be consideted later in a more general context. whereas spending it on hospital hemodialysis would give only (For a critique of DALYs see Loomes and McKenzie 1989). one DALY. Two contributions to the field of development of DALYS are It is perhaps superfluous to add that the development of particularly noteworthy. We will discuss, first, some of the DALYS can be difficult. However it is done, attaching weights work of the "father of DALYs," George Torrance, from Canada to different morbidity states in relation to death, so that, and, second, the work of Alan Williams, from England (see ideallv, mortality and all forms of morbidity can be placed on Torrance 1985; Williams 1985). a single index, involves value judgments. It is also thecase that One of Torrance's key contributions to the field is with a .. respect to methodolog and in various papers hehasprowe know of no "fully fledged" DALY applications in developing respect to methodology and in various papers he has provided countries. much guidance for researchers in how to measure health states in practice. Thus he gives the main steps in developing health SOME COMMON PROCEDURAL POINTS. Whatever methodology status measures: is adopted for assessing the burden of illness, there are five * Identify the relevant health states for which preferences issues that need to be handled with care. are required. * Determination of the purpose * Describe the health states. * Discounting over time * Select the subjects whose preferences will be measured. * Other outputs * Determine the type of preferences required (ordinal, * Uncertainty cardinal). * Determine the measurement instrument to be used. Although determining the purpose may seem an obvious Although we cannot discuss all these steps in detail, it is point, it is worth stressing that how the burden of illness is best worth noting, regarding the last, that there are various ways of measured or valued is a function of why it is being measured or 736 Gavin Mooney and Andrew Creese Table C-2. Costs and Consequences of Selected has called the caring "externality." Information is also an Medical Procedures output. For example, informing patients about their state of (pounds sterling) health even if it is not changed or indeed cannot be changed Presenit value of may provide benefit to the patient. Being able to pass difficult extra cost per decisions to the doctor may also sometimes be of benefit to Procedure DALY gained some patients. It is difficult to say what weight will be attached to these Hipreplacemaerintation7for5heart0block other outputs. It is clear, however, that their importance is Hip replacement 750 Valve replacements for aortic stenosis 950 likely to vary both across different diseases and across different CABG for severe angina with left main disease 1,040 patients. Thus, although it is appropriate in assessing priorities CABG for moderate angina with three-vessel disease 2,400 to concentrate on the output side on reductions in the burden Kidney transplantation (cadaver) 3,000 of disease, these other forms of output may sometimes alter the Heart transplantation 5,000 priority rankings or weightings. Home hemodialvsis 11.000 Benefits in the future may be uncertain, and in such cases CABG for nmild angina with two-vessel disease 12,600 an adjustment should be made to reflect their expected value. Hospital hemodialysis 14,000 For example, the reduction of infant mortality may lead to Note. CABG (Coronary artery bypass gratting). greater benefits as a result of a health education campaign Soutce Drummond 1987; Williams 1985. concerned with hygiene to reduce childhood diarrhea. Often, however, it will not be possible to state precisely what all the potential effects of an intervention for the treatment or pre- valued and in what circumstances. Calculations concerning vention of some disease will be. the burden of illness will almost always be used as an estimate Where uncertainty exists, sensitivity analysis should be used of some output measures, and output measures have to be or in handling it; that is, a range of values should be put in for a ought to be related to the purpose or objective of the exercise. particular parameter to see what the effect of the different It is also clear that if the wrong measure is used it is quite likely values is-how sensitive the result is to the change in values. that a distorted answer will be obtained. (For example, if breast Where the result does change, it may be necessary to devote cancer treatment programs are related solely to percentage of some effort to trying to reduce that particular uncertainty. survival over, say, five years, then all aspects of quality of Although we accept that equity is an important goal in most life-pain, dignity, losing a breast, atid so on-will be ignored health care systems, we are focusing in this appendix on and given a zero weight. Yet it seems clear that women suffer- efficiency. Still, it is important to recognize that equity and ing from breast cancer will value more than just survival.) efficiency goals can sometimes conflict. Such a conflict may For calculations of the burden of illness (and also the mean that minimizing the burden of illness is not the goal or resource costs of interventions), the value attached at different at least that such a goal is constrained by concems for equity. points in time is not constant. As Barnum states: "Neither the For example, although it may in some instances be efficient tO individual nor the community is indifferent as to when the concentrate highly specialized facilities in the cities, this is etfects of disease occur.... A healthy day of life in the presenit unlikely to provide an equitable system with regard to geo- has a greater intrinsic value to the individual than a day in the graphical access. future (1987, p. 834)." The way to handle this phenomenon is It is also the case that if equity is concerned with access or through "discounting" future benefits and costs at some posi- use rather than with health per se, then factors other than tive rate of discount; such discounting results in a weighting purely burden of disease have to be taken into account. In over time which gives more weight to current effects, less to other words, if a society values the fact that individuals have those of the near future, and still less to those in the distant equal access to health care irrespective of whether they then future. This means. for example, that preventive programs may use it to obtain effective care, then such a set of values cannot seem to do rather badly as a result of discounting. This is be directly contained within burden-of-illness calculations. because they often involve costs now (which are therefore not Certainly in many-but admittedly not all-equity mea- discounted) aind benefits in the future (which are discounted). sures there will be some need to assess the relative burden of What rate of discounting to use is problematical, and it is disease across different groups in society. Such cases present normal procedure to tise a range of rates, usuIlly between about the few occasions in which the burden of disease itself, as 3 and 10 percent. opposed to its reduction, is the relevant policy measure with Although it can generally be agreed that the decrease of the which to operate. Whether that is the relevant measure of burden of disease on the sufferer is the prime output of- any equity to use is something that cannot be resolved in this health care system, other outputs are present and relevant for appendix (but see for more discussion Mooney 1992). Other the setting of priorities. For example, if infectious diseases are tactors such as access may become relevant. cured in some people, others who would otherwise have be- come infected will benefit. Again, nonsuifferers may benefit METHODS FOR PUTTING MONETARY VALUES ON OUTCOMES. We knowing that others'suffering is reduced-what Culyer(1976) have seen above that health status measures are more widely Pnonty Setting f(rn Health Service Efficiency : The Role of Measurement of Burden of Illness 737 usable if they aggregate the relevant components (mortality the ones that should be allowed to count. Also, the question and morbidity) in a single numeraire, such as healthy days or posed as a probability does seem appropriate. (For example, to DIALYS. In an analogous fashion, the usefulness of health out- ask an individual what he or she is prepared to pay to avoid come data for priority setting is substantially increased if an certain death is almost certainly an unanswerable question.) acceptable monetary yardstick can be found, to allow direct Whether it is possible to obtain valid answers to such ques- comparisons between the value of inputs used in improving tions, however, remains unclear. It is possible to study the health and the value of these improvements. When this is behavior of individuals in risk situations and elicit their im- possible, not only can cardinal comparisons be made between plied values (for example, in their willingness to pay for safety competing claims on resources, but the more fundamental devices on their cars), but many of these situations are so far cost-benefit questions can be asked and answered. removed from the sorts of choices relevant to health care There are three principal methods for putting monetary valuations that the values emerging may not be very useful. values on health outcomes: Additionally the studies that have been conducted in which this strategy wvas used yield a very wide range of values-but * The "human capital" method ones which are normally much higher than those based on the * The "willingness-to-pay" (for risk reduction) method human capital method. * The "implied values" method Despite the practical problems of the willingness-to-pay method, it has considerable theoretical advantages in that the The oldest and simplest of these methods in practice is the valuation basis of individuals' willingness to pay for reductions human capital one. In this method it is assumed that the in the risk of illness and death seems more defensible than that objective function that we are trying to maximize through in the human capital method. Of course, if there are equity improved health is gross national income in that the measure objections to the method on grounds that priorities in health of value is an individual's output, normally assumed to be equal care should not be based on individuals' ability to pav on to the gross labor costs of employment or in some instances which willingness to pay is inevitably based), then its aplolica- simply the earnings of the individual. Thus ifa person is unable tion has to be handled with care. The use of the meihod to to work because of illness, we would, using this method, date has been very restricted and has related more to willing- estimate the burden of that illness as being the work output ness to pay to reduce the risk of dying than to reduce the risk lost, which is equated with the gross labor costs of employing of having a nonfatal illness and injury. the individual over the relevant time period. If a person dies The third method cf evaluating the burden of illness, the as a result of illness, the burden is equated with the present implied values method, is somewhat similar to that of willing- value of the gross costs of employment over what would oth- ness of individuals to pay to reduce risk, except that now it is erwise have been his or her expected working life span. a question of determining what the implied willingness of There are some clear problems with this method. Unless health care and other health inducing organizations is to pay adjustments are made, it means that no weight is attached to for various health outputs or reductions in the burden of illness. retired people, housewives, children (as children), and others The basis of the method is simple: if a decision is made, at the not gainfully employed. Also it will give different values to margin, to spend $1 million to save a life, then by implication high earners and low earners, which may well be deemed an the value of that life must be at least $1 million, otherwise the inequitable basis on which to set health priorities. Further, investment would not be made. If a decision is made not to gross labor costs are at best an approximation of the value of spend $2 million to save a life, the value of the life is then by an employee's output. It is also assumed that there is no value implication less than $2 million. to health beyond the capacity it provides to produce output In this process ofestimating the implied values of life, ideally relevant to the gross national product, a somewhat restricted one would wish that for similar outputs the willingness of the view of the goal of health services. health care system to pay at the margin of each program would The willingness-to-pay method, most often applied to the be the same (the condition for an efficient solution). What saving of life or, more precisely, the reduction in risk of death, limited information exists, however, suggests that there is a adopts a different value stance. Here the nature of the social very wide range of values for like outputs. That does not mean welfare function-that is, what it is that is to be maximized that attempts to make the values explicit should not be pur- from a societal perspective-is based on individuals' values sued. The point is that the aim might first have to be to sort with respect to their willingness to pay for reductions in risk of out the inefficiencies implicit in the fact that there is a range death (or injury or illness). Thus it is assumed that it is of values rather than in the short run to use the values per se legitimate to ask potential victims or potential sufferers how in the assessmentof the burden of illness. Even then, however, much they are prepared to pay for a reduction in, say, the risk the use of a mean value in the short run would be a possible ofdeath from perhaps 3 in 10,000 to 2 in I0,000. If the response strategy. on average to such a question were $5, then the value of a One of the clear advantages of this implied values method "statistical" life would be $50,000 (that is, 10,000 x $5). is that it does not involve any change in the value system, This strategy has some advantage in that the question is put because the implied values would simply reflect those of the to the potential victim, whose values, it can well be argued, are existing system. The method is also relatively easy to apply. 738 Gavin Moonev and AnLdreu! Creese It must be obvious from what has been said that none of the encourage classification, rather than enumeration. The inter- methods outlined is ideal in both principle and practice. The ventions for which cost per DALY gained have been compared human capital method is simple but tends to treat people like are a tiny and nonrepresentative fraction of those available. machines, where their only value is as workers. It may be Indeed,theirbestuseliesmoreintheillustrationofthemethod argued, however, that estimates made on this basis can provide of cost-effectiveness in priority setting than for any realistic at least minimum values of life and sickness avoided. The debate on priorities at a global level. For a full review of willingness-to-pay method is, theoretically, to be preferred but priorities, more information is needed and on a more local has not yet been widely applied even within the mortality field, basis. where it is most frequently found. It also requires substantial Although numerically insignificant, however, the interven- investment in data. The implied values method at least pro- tions evaluated in the chapters of this collection do have an vides a basis for improving technical efficiency and is relatively epidemiologic significance beyond their mere number. They simple to apply. (For a fuller discussion of valuing life, see include interventions of known effectiveness against some of Mishan 1981 and Linnerooth 1982). the main sources of mortality. Many of these interventions might thus be expected to be prominent among health priori- Cost-Effectiveness Comparisons for Priority Setting ties even if the total number of cost- and outcome-documented additional health interventions were dramatically expanded. At the present stage of development of methods of priority Even for those interventions which are considered, there setting it is suggested that a simple, sensible way to proceed is remains some unevenness in the relevant types of cost and to identify the marginal costs of similar outputs across different health outcome data presented. In two particular areas this programs and adjust the allocation of funds to try to get such shortage of information may be a critical limitation. First, the marginal costs closer to equality. In other words, if (a) some sensitivity of the estimated costs per DALY gained is not, in all form of DALY measure can be devised and (b) the cost per DALY cases, subject to appraisal. Point estimates, or even "greater gained can be identified on the margin of each existing program, than" estimates, are of limited value when there are important we can then attempt to reallocate resources from programs margins of uncertainty surrounding them. As indicated above, in which the marginal cost per DALY gained is high to those in sensitivity analysis is important in narrowing down the areas which it is low. in which further information is required and in avoiding over- The reviews of contemporary empirical experience of cost dogmatic priority ranking where the state of available knowl- and outcome relationships contained in chapters of this col- edge should indicate caution. Second, the data presented are, lection constitutes an important piece of stock taking. Epide- in all cases, estimates of the average cost per DALY gained. As miologic, technologic, and economic characteristics of the emphasized above, such information may lead to inappropriate main diseases and the principal current interventions are pre- resource allocation decisions. If studies are conducted to estab- sented in a broadly similar format, which allows estimated lish the relationship between average costs and marginal costs, average costs per average number of days of healthy life gained then no problem arises. But there are few, if any, such studies to be compared (see figures 1-7 and 1-9). for the interventions reviewed in this compilation. By compar- What conclusions is it possible to draw from these reviews? ing costs and output for health interventions operating at Of equal importance, what conclusions is it not possible to differing scale, we can identify the effects of output variation derive from these data? In the first place, the very existence of on total and marginal costs. Once again, in too little of the such a quantity of information on such a range of interventions available empirical work have output variations in relation to is clearly to be welcomed. Too many studies have argued for costs been assessed. greater priority in funding for one specific disease or interven- So, in the absence of empirical information about the rela- tion, in the absence of any explicit comparisons. Such studies tionship between average and marginal cost, what analytical are the antithesis of an economic way of dealing with the use can be made of the available data? One route to follow is situation, in which the necessity for making trade-offs between to proceed on the assumption that marginal costs are close to activities, in the face of overall resource limitations, is taken average costs. This is a very special and potentially dangerous as a starting point. A galaxy of alternative patterns of resource assumption. The most casual observation of health care facil- use exists, even in the poorest country-in the target groups ities in developing countries suggests that chronic overuse (for (for example, adults or children), in the intervention strategy example, multiple occupants of hospital beds, "floor patients," (preventive or case management), and in the type of disease long lines at hospital clinics) coexists with equally chronic or health problem. underuse (for example, less than twenty consultations per The scope of these reviews, however, is still very modest month at a health post with a staff of four health workers, and when compared with the huge quantity of health-related ac- infant immunization rates of under 20 percent). A bold sim- tions coexisting in any country at a given moment, or even in plifying assumption of equality between marginal and average a single small general hospital. The range of available health cost thus seems more sanguine than intuitive. interventions, differing in input mix, location of treatment, Without the benefit of either a simplifying assumption or type of patient, type of illness, timing of intervention (primary some empirical basis to speculate about the relationship, at or secondary preventive, curative, or caring), is so large as to current output levels, between marginal and average cost, Pnonty Setting for Health Service Efficiency: The Role of Measurement of Burden of lUness 739 restraint should be used in applying such data to a review of more or fewer overall resources available, that changes the priorities. This is a disappointing conclusion. If marginal and position-but not the principle. Again if other outputs are average costs per DALY were roughly equivalent, if the data deemed relevant (for example, reassurance or information), as incorporated allowances for uncertainty, and if these interven- we believe they should be, then benefits other than reduced tions were taken as in some sense representative of technolog- burden of disease must be taken into account. ical options in health care, then the data in figures 1-7 and 1-9 The link between priority setting and efficiency is crucial in could be interpreted as revealing the following: the context not only of the burden of disease per se but of the * Globally, interventions aimed at children should receive debate about priority setting more generally. Let us restate a Glbaly, mervntlos amed t clldrn souldreclve clearly what our views are on this matter: higher priority, whether for case management or preven- tion, than those aimed at adults. * The need to set priorities arises from the fact that not * Although the average cost per DALY for preventive inter- all illness can be eradicated nor all needs met; this is not just ventions targeted on children is approximately half of that of a statement about technology but about the scarcity of case management interventions, the ten most cost-effective resources. activities (at $20 per DALY or less) are a mixture of both * Priorities are about change. Decisionmakers and policy- preventive and curative actions. makers have to try to redeploy resources, allocate some new * For adults the overall mean cost for preventive interven- (but limited) resources, and cut back on the use of existing tions is still lower than for case management, although the resources in such a way as to get the most out of whatever differences are now much less. resources they have. That means changing deployment. * The ten most cost-effective interventions for adults en- * Priorities should be based on both benefits and costs. tail a mix of preventive and curative actions. Using scarce resources in one way means, by definition, * Some service set providing integrated cure and preven- giving up the opportunity to use them in some other way; tion, rather than discrete vertical programs, would appear to providing benefits here means forgoing them there. Priority be the most appropriate delivery mechanism. setting means trying to ensure that those policies that get * The optimal mix of interventions will change as demo- priority (that is, what gets a higher call on resources) are graphic and epidemiologic profiles differ or shift, and thus it those providing greatest benefits per dollar spent. If the needs to be kept under continuous review. dollars could have been better spent elsewhere, then they should have been spent elsewhere. These tempting conclusions are not strictly possible. The data fitted into the cost-effectiveness apparatus are simply not good These three statements are central to priority setting. They enough-in quantity and in quality-to warrant such conclu- are very neatly summed up by Shiell, Gerard, and Donaldson sions. This does not mean that the exercise is worthless. If we (1987) in their critique of studies on the cost of illness: "the can reach such conclusions, for a project, country, or region, total 'costs of illness' can only indicate the benefits of treat- they are clearly of consequence. That we cannot yet do this- ment options if an intervention is capable oftotally eradicating although we may be close-gives urgency to the need to or entirely preventing the disease in question. This is only accelerate and improve the collection of relevant data. likely to be possible in the case of a very few infectious diseases. Additionally the emphasis on looking at marginal change The most pertinent questions facing policymakers usually will normally mean that collecting even crude data on mar- relate to scale; that is, by how much should an existing program ginal costs at a local level will be better than adopting national be expanded or contracted. The answer to this question or international average cost data. The message is clear. Better requires a marginal analysis which compares the expected to have approximate estimates of local marginal costs than change in benefits with the costs of the intervention which precise, more generalized, average costs. brings that change about." From this appendix a number of important conclusions Concluding Comments emerge on priority setting in the context of the burden of disease. First, the emphasis of efforts on priority setting ought Priority setting is about choice. It is about arranging things in to be firmly "on the margin": what can be bought with a few such a way that those policies and programs that are considered dollars more? what shifting of resources from one program to most worthwhile stand a better chance of being implemented another on the margin can provide the maximum reduction than others that are considered less worthwhile. In other possible of the existing burden of disease? if cuts have to be made, words, not all needs can be met because resources are scarce. where should this happen to minimize any increase in the burden Disease cannot be eliminated; it can only be reduced. So of disease? Second, developing some common measure of mar- priority should be given to those areas in which the burden of ginal changes in the burden of disease across different diseases illness can be reduced most per dollar spent. Indeed, we should is the key to progress in this area. Third, efforts to measure the continue to set priorities according to incremental or decre- total burden of any disease ought to be resisted because total mental changes until it is agreed that no further movement can burden is not the basis for setting priorities. Fourth, averages reduce the burden of disease even more. Clearly, if there are are likewise to be resisted except where it can be shown that 740 Gavin Mooney and Andrew Creese they are reasonable approximations for marginals. Fifth, care Hall. J., and G. Masters. 1986. "Measuring Outcomes of Health Services: must be exercised to ensure that all relevant factors are ac- A Review of Some Available Measures." Community Health Studies counted for-other nonhealth outputs, equity considerations IQ(2):147-55. Henke, K.-D., and C. S. Behrens. 1986. "The Economic Cost of Illness in uncertainty, discounting. Finally, whatever measures are the Federal Republic of Germany in the Year 1980." Health Policy adopted, sensitivity analysis should be applied to determine 6:119-43. how robust the results are to different assumptions. Hodgson, T. A. 1989. "Cost o f Illness Studies: No Aid to) Decision Making? Comments on the Second Opinion by Shiell et al." Health Policy 11:57-60. Notes Kaplan, R. M., and J. W. Bush. 1982. "Health Related Quality of Life Measurement for Evaluation Research and Policy Analysis." Health Psychol- Comments from Howard Bamum (World Bank), David Evans (World ogy 1:61468. Health Organization), Karen Gerard (University of Sydnev), Richard Morrow Linnerooth, Joanne. 1982. "Murdering Statistical Lives ' In M. W. (World Health Organization), David Parker (United Nations Children's Jones-Lee, ed., The Value of Life and Safety. Amsterdam: North Holland. Fund), Gerald Rosenthal (REACH project), and Carl Stevens (Reed College, Loomes, G., and L. McKenzie. 1989. "The Use of QALYS in Health Care Portland, Oregon) were particularly influential in shaping our thinking about Decision-Making." Social Science anid Medicine 28(4):299-308. this appendix. Remaining errors of fact and interpretation are our own. We Mills, Anne. 1985. "Survey and Examples of Economic Evaluation of Health are also indebted to) Anne Haastmup for her secretarial assistance in preparing Programs in Developing Counitries." World Health Statistical Quarterly and revising the manuscript. 38(4):402-31. M ishan, E. J. 1981. Cost Benefit Analysis. London: George Allen and Unwin. Mooney, G. H. 1992. Economics, Medicine, and Health Care. Brighton, Engl.: References Wheatsheaf. Parkerson, G. R., S. H. Gehlback, E. H. Wagner, S. A. James, N. E. Barnum, Howard. 1987. "Evaluating Healthy Days of Life Gained from Health Clapp, and L. H. Muhlbaier. 1981. "The DLike-UNC Health Profile: Projects." Social Science and Medicine 24(10):833-41. An Adult Health Status Instrument for Primary Care." Medical Care Behrens, C., and K.-D. Henke. 1988. "Cost of Illness Studies: No Aid to 19:806-28. Decision Making? Reply to Shiell et al." Health Polic' 10:1 37-41. Petty, Sir W. 1699. Political.AnthmericoraDiscourseConcemringthe Extentand Bergner, Marilyn, R. A. Babhitt, W. B. Carter, and B. S. Gilson. 1981. "The Value of Lands, People, Buildings, etc. London: Robert Clavel. Sickness Impact Profile: Development and Final Revision." Medical Care Prost, A., and N. Prescott. 1984. "Cost-Effectiveness of Blindness Prevention 19:787-805. by the Onchocerciasis Control Program in Upper Volta." Bulletin of the Culyer, A. J. 1976. Need and the National Health Service. London: Martin World Health Organization 62:795-802. Robertson. Shiell, A., K. Gerard, and C. Donaldson. 1987. "Cost of Illness Studies: An Dnummond. M. F. 1987. "Economic Evaluation and the Rational Diffusion Aid to Decision-Making"' Health Policy 8:317-23. and Use of Health Technology." Health Policy 7:309-24. Torrance, G. W. 1985. "Measurement of Health State Utilities for Econonsic Drummond. M. F.. A. Ludbrook, K. Lowson, and A. Steele. 1986. Studies Appraisal-A Review."Journalof Health Econiomics 5:1-30. in Economic Appraisal in Health Care. Oxford: Oxford Medical Publications. Walsh, J.. and K. Warren. 1979. "An Interim Strategy for Disease Control in Ghana Health Assessment Project Team. 1981. "A Quantitative Method oif Developing Countries.." Neuw EnglandJournal of Medicine 301:967-73. Assessing the Health Impact of Different Diseases in Less Developed Williams, Alan. 1985. "Economics of Coronary Artery Bypass Grafting." Countries." IiterntionalJourLal of Epidemiology 10:7 3-80. Bntish MedicalJounial 291:326-29. Appendix D Rationales for Choice in Public Health: The Role of Epidemiology Andre Prost and Michel Jancloes The decisionmaking process in public health has attracted certain periods of time. Epidemiology is the instrument of much attention in recent years. The share of the health and choice for measuring effectiveness. Background data collected social sectors in public and private expenditure has increased before the intervention constitute a resource which must be to the point that most systems no longer seem to be affordable elaborated into appropriate information in order to document either in industrial or developing countries. The importance precisely the changes that have occurred and to demonstrate of the economic factor has meant increasing challenge to the trends. Epidemiological techniques can also be used to assess rationale for making decisions only on technical grounds, a unforeseen benefits, provided they are accessible through process which is deeply rooted in a sector managed by a strong health and demographic indicators. technical constituenicy and which is backed by emotional This role can be extended to the forecasting of the outcome moves in public opinion expressed as "Health at any cost" or at the planning stage. Usually, several implementation strate- "Nothing is ioo expensive for the sake of saving life." New gies can fit the design of a project and be considered adequate processes are introduced, sometimes reluctantly, in the man- to meet the stated objectives. It is necessary to select the agement of the health sector: determinatiotn of priorities; ad- operational strategy that will maximize the outcome in rela- vocacy shifting from effectiveness of technology to that of use; tion to the input. Cost-effectiveness analyses of this type rely preference given to mass benefits over the satisfaction of heavily on the accuracy of the epidemiological situation and individual demand; and so on. trend analysis. Public opinion on health matters makes sectoral choices In addition, epidemiological assessment takes into account more than a policy issue: it makes them a political problem. the wide range of relationships and interdependencies within Final decisions result from the combination of technical judg- health systems and between health and other sectors. The ments, economic feasibility (rarely economic conseqLiences), comparative study of the importance of and interactions be- pressure from lobbies, social and psychological implications, tween health determinants may bring unforeseen side effects and predominantly circumstantial opportunity. Short-term to light. It also shows the effect on the health sector of considerations tend to prevail over long-term implications. It decisions made in other sectors, such as taxation and financial is therefore of the utmost importance to identify criteria for measures, agricultural and industrial reforns, and, more gen- decisionmaking which could introduce some rationality in the erally, public policies. decisionmaking process and possibly increase the chances of Because epidemiology has become the key factor for eval- reaching a consensus among all actors: providers, users, payers, uation and for effectiveness analyses in health, there has and policymakers. been an increasing tendency to use it at earlier stages of the Attempts to use an economic rationale as the main, or even process, namely at the stages of planning, policy setting, and the sole argument, have been opposed by both health providers dec is ionmaking. The technical method used in this disc ipl- and users. They cannot accept the impositioni, for financial ime together with its ability to rank priorities and to identify reasons, of any liimitation on the degree of sophistication of the selective programs would seem to reflect the concerns of all technology on one side, or on the benefits ihey may enjoy on the actors and make conclusions readily acceptable. Our the other. Thus, no consensus can be reached using cost- aim in this appendix is to analyze the advantages and the benefit or cost-effectiveness analyses. difficulties of this approach as well as the challenges that For the last ten years, epidemiology has been promoted as epidemiologists face in making a relevant use of their tech- an alternative tool for decisionmaking in health. Epidemiolo- nique. We argue that, although epidemiology is an essential gists aim at describing health and disease phenomena in pop- tool for policy setting, it cannot be the ultimate rationale ulation groups, their determinants, and changing pattems over for decisioninaking in health. 741 742 Andre Prost and Michel Jancloes Current Methods practical purposes is one single number, tends to express avail- able information with a degree of precision which far from The principal conceptual problem lies in the definition of reflects the confidence limits of the assumptions. The devel- health indicators which represent the variety of diseases, ag- opment of computerized systems of data collection and infor- gregate the differences among population subgroups, and ac- mation storage increases the illusion of exactness. It can be count for changes over a period of time. A single measure of misleading, especially for nontechnical people who are unfa- health status has never been unequivocally accepted; it may miliar with the critical assessment of the validity of results. even never have existed. Because of this difficulty, analyses For example, suppose an estimated incidence of severe gas- have usually been limited to the comparison of two diseases, troenteritisof 130casesper 1,000populationperyear: acritical or to the comparative effectiveness of control strategies for a assessment should consider whether this figure is based on the single disease. The lack of a standard measure of health status records of outpatient visits in health clinics or on a survey of a considerably hampers the use of epidemiology as an instrument population sample, whether the result has been adjusted to for planning and policy setting. account for differences in the age and sex distribution between Two methods have been designed in recent years to inter- the survey population and the whole population of the coun- pret conceptually the role of epidemiology in public health. try, what recall period was used in the survey, what criteria One method uses a quantitative compounded indicator-the were used in the definition of "severe cases," how the seasonal number of disability-adjusted life-days-to assess the effect of variations have been taken into account, and so on. The disease.I Thus, results are applied to the process of health confidence limits represent the range in which the majority of resource allocation and management (Ghana Health Assess- results from different sources will be included. The value of the ment Project Team 1981; Morrow, Smith, and Nimo 1982; assumptions and their confidence limits vary between diseases Romeder and McWhinnie 1977). The ranking of diseases is because of varying levels of precision in the data base and established by using the average estimated number of healthy because of differences in the complexity of the epidemiological days lost in a lifetime as a result of disease episodes. The loss pattern of diseases. attributable to every single disease is the sum of healthy days Striking differences in the ranking of diseases by order of lost through acute illness episodes (temporary disability), importance can occur, depending on the choice of the upper through chronic conditions and sequelae (partial or complete or the lower limit of the range of use in the calculations. Such disablement), and through premature death (considered ranking also depends on the magnitude of the multiplier effect equivalent to complete disablement). The loss to the commu- introduced by the various mathematical formula aiming at the nity is derived by multiplying the average loss attributable to calculation of a synthetic indicator (see note 1). every single disease by the annual incidence of each disease in the community. Validity of Indicators Another method is the "measurement iterative loop" devel- oped at McMaster University, Canada (Tugwell and others The ranking of diseases for the determination of priorities 1984). It is a framework intended to guide informed necessitates the use of a single indicator to allow for compari- decisionmaking in health. The seven successive steps reflect sons. Thus, ranking of diseases can be established by using, for a logical progression from the assessment of the burden of example, the death toll, or the incidence in the community, illnesses through hypothesis generation about the causes of or the degree and duration of the resulting disablement. The disease, and about the efficiency and the effectiveness of selection of the indicator represents a value judgment which prevention and treatment procedures, to evaluating the effect may reduce the freedom of the decisionmaker and somehow in a community. It is an approach of a programmatic nature, preempt the decision. Using mortality figures could mean, for based on a rational scheme of planning, including an in-depth example, that the social and economic cost of disablement is analysis of sector needs and constraints. Epidemiology is used obliterated. Using national averages does not allow for the in the determination of priorities for action, which result not identification of population groups especially deprived or par- only from the assessment of sectoral needs but also from ticularly at risk. an assessment of effect, and thereof from cost-effectiveness The use of a compounded indicator is an attempt to reduce choices. these biases. The method of disability-adjusted life-days (DALDS) Both methods are attractive. They represent valuable con- combines the effects of mortality, morbidity, and disablement; tributions to the theory of the decisionmaking process in it uses life expectancy as the reference period. There is no health. Unfortunately, their application to real situations does significant bias in the analysis when relatively similar health not meet the high expectations raised at the conceptual stage. conditions affecting the same age groups are to be compared. We shall attempt to determine reasons for failure. For example, the first mention of this indicator in the literature (Dempsey 1947) was applied to a comparison of mortality due Validation of the Results to tuberculosis, heart disease, and cancer; all these diseases are chronic and are prevalent mainly in adults. Quantitative epidemiology uses mathematical tools: averages, More recently, the proposal to use the DALDs lost indicator percentages, ratios, and so on. The resulting figure, which for as a general method for assessing the effect of diseases (Ghana Rationales for Choice in Public Health: The Role of Epidemiology 743 Health Assessment Project Team 1981) has resulted in more diseases and they do not cover the broad spectrum of health complex combinations. It aggregates not only the effects of disorders. In many developing countries, in the absence of any mortality, morbidity, and disablement, but also it combines the reliable data base, epidemiologists use assumptions derived effect of acute and chronic conditions, in all age groups from from scattered surveys, from incomplete reporting systems, or, birth to death. The authors did not discuss the implicit value even worse, from hospital statistics. judgments of this method. For example, the assumption that To improve the quality of data, health planners direct "the younger the death, the greater the loss to the community" considerable effort toward the collection of health statistics. derives directly from computing the difference between the life An example is given by the comprehensive epidemiological expectancy and the age at death. It implies that the death of a survey conducted by the health services of Mali with support child is a greater loss than the death of a young adult in the from the World Bank (Duflo and others 1986). The objective productive period of his or her life simply because life expec- of this survey was to help design a regional health development tancy for a child is much greater than it is for an adult. It implies project. The survey was conducted in a random sample of for the same reason that maternal mortality is of less conse- villages during a period of one month. Specific morbidity and quence to the community than the simultaneous death of the mortality rates were determined and were used to estimate the newbom. Also, by definition, the measure assumes that one number of days lost as a result of the diseases observed. year of complete disability is equivalent to one year of prema- This methodologically sound survey provided an accurate ture death. It could be argued that meeting the needs of a picture of the disease situation, with a reasonable degree of disabled person places a heavier burden on the community. precision. It emphasized the relative importance of neglected Finally, mixing together the effects of mortality and morbidity pathologies, such as eye diseases, cardiovascular disease, and results in minimizing the social and economic cost of common hemoglobinopathy. It did not, however, change the prelimi- though nonfatal diseases, which may represent up to 80 per- nary ranking of the main diseases (malaria, gastroenteritis, cent of the workload of outpatient clinics and a large share of measles, malnutrition, pregnancy complications, respiratory drug expenditure. infections) which had been established on the basis of poor It is obvious that the aggregation of morbidity and mortality quality data available from health providers in the area. It can into a single measure necessarily involves making value judg- be argued that data collected at a high cost during the survey ments about the relative weights that should be assigned to have not yielded any better information for the project design. each component. The assumption that additional years of life The additional precision in the assessment of the burden of are equally valuable, regardless of the age at which they diseases made justification of the project more difficult to accrue, conflicts with the common notion that adult mor- challenge. More precise data are ofa greater value as a baseline tality is more serious than child death. Weighting proce- for future evaluation of project benefits. They were not used at dures may alleviate the difficulty. It can be reflected through the planning stage. the assignment of a zero weight to years of life added before Important in the establishment of a data base is that the age fifteen, and a weight of one to those added beyond age system be conceived in relation to the needs of the users or fifteen. Any other weight could be proposed and discussed. potential users. Too often, data which are critically needed are This method of weighting for age preference can be com- missing, or they are impossible to retrieve from bulk informa- bined with the relative weighting of disability and death, tion. Too many epidemiologists, nationally and internation- and with weighting for time preference (that is, assigning ally, perceive their function as that of collecting the greatest lower weights to benefits which occur in a distant future). possible amount of information inorder tocombine allpossible Previous studies have shown that assessments of the effect of a variables. In fact, the role of epidemiologists is to tailor the disease and of the effectiveness of a health intervention are collection of data, using a problem-solving approach, in accor- very sensitive to the choice of different weights (Prost and dance with hypotheses generated at a preliminary stage. Prescott 1984; Barnum 1987). Introducing productivity weights (that is, allocating different weights according to the Comprehensive in Contrast to Selective Care status of the patient as a producer for the community) has even greater policy implications. Thus, there is no straightforward The ranking of diseases, based on whatever epidemiological weighting procedure which could lead to noncontroversial indicator is selected, singles out a list of diseases or individual measurements. On the contrary, the selection of weights health conditions as the target for control, either because they results from value judgments and therefore carries the risk of represent a public health scourge, or because of their socioeco- further distorting the objectivity of the method. nomic effect. Six diseases in Ghana (Ghana Health Assess- ment Project Team 1981) and eight in Mali (Duflo and others Quality of the Data 1986) account for 50 percent of the total number of disability- adjusted life-days lost to the community every year. It seems Good quality data bases do not exist in most countries. There essential, at first glance, to concentrate all efforts on combating is no better consensus on the incidence of home accidents in these diseases, or the mnost important of them, because larger Europe than on the number of diarrheal episodes in African benefits will accrue to the community. Thus, the search for children. When available, data are often limited to specific maximum efficiency leads to the development of disease- 744 Andre Prost and Michel Jancloes oriented programs, using specially designed control methods system, which places greater importance on individual (case finding, case management, evaluation), selective logisti- conditions and on adult morbidity, for example. Whereas cal support, and targeted retraining of staff. epidemiological surveys might conclude that diarrhea and One application of this concept is the Selective Primary measles are the priority diseases, a sociological survey might Health Care Strategy proposed by Walsh and Warren (1979). reveal, for example, that hernia, hemorrhoids, blindness, The ranking of priority diseases is based on the assessment of and complications of delivery are the priority concerns of their effect in the community and of the effectiveness of the population. available control methods using the implicit value judgment Moreover, the epidemiological method emphasizes the im- that reducing infant mortality to improve life expectancy is the portance of the determinants ofdiseases, leading to preventive objective of efficient health services. Therefore, selective pri- rather than to curative actions. The failure to appreciate the nmary health care has focused in most cases on diarrhea and primacy of prevention is now shared by the general public and diseases preventable by vaccine, and activities have been by a majority of the health profession (Terris 1980). Policies almost exclusively concentrated on oral rehydration and im- based on an epidemiological rationale are generally opposed munization campaigns. by both providers and users of health services. Especially at This strategy is both conceptually and practically mislead- times of economic stringency, programs of health promotion ing. First, it relates cost figures to disease control effectiveness and disease prevention are easier targets for short-term savings and not to health benefits. It does not consider that the than is specialized curative care. This attitude coincides with allocation of resources to one activity can have various types the expectations of the consumers and with the dominant of benefits. For example, in an experiment in Zaire, the villages position of the health professionals in the curative technical in which successful treatment for intestinal worms hadl been structures. carried out have shown improvements in immunization com- In almost all cases, an ethical conflict arises between epide- pliance, tuberculosis screening, and health education. There miological and sociological methods. Should planners ignore was also a decrease in the average number of patient visits to it, the community will develop sideline channels to meet the the health facilities (Jancloes 1989). In such a program, para- demanid (private practice, traditional healers, uncontrolled site control is used as a catalyst to trigger the compliance of the sale of drugs, and so on). The result is the lack of users' people with health services, and thus to progress toward the commitment to the successful implementation of the program real objective of improving the health of every family member. and the absence of a rational use of resources despite intensive Second, on the practical side, it is almost impossible, at the planning efforts. peripheral level, to focus on a limited number of diseases. In the case of the Mali project mentioned above, decision- Health services are multivalent by nature. The definition of makers took the demand aspect into account at the initial tasks results from the people's demand for care and from a stage. They considered that ensuring the effectiveness of the comprehensive public health strategy which combines the referral level in the treatment of adult diseases was essential to provision of curative care, prevention, hygiene education, and the credibility of the program. In a second phase, they took interaction with other sectors that influence health. Patients into account the epidemiologically determined needs, with have been reluctant to use the facilities available in pilot village interventions aimed at reducing child morbidity and projects set up to test the feasibility of the selective primary mortality. health care strategy, mainly because they realized that these facilities could not cover the broad spectrum of their com- The Decisionmaking Process plaints and that they would have to visit another health post for complementary treatment. Decisionmakers act on their own judgment as to whether they Thus, the determination of "priority diseases" is not only themselves or the society at large could derive more benefits misleading with regard to allocative efficiency but it ignores from the proposed strategy than from competing health inter- the multisectorality of the health determinants. It ignores ventions (and sometimes nonhealth interventions). The ben- observations that some of the most significant progress in efits are both those assessed by project evaluation and those health has derived from nonmedical interventions (for exam- perceived subjectively. ple, decrease in infant mortality with rising education levels, Decisionmaking is a complex process which involves a historical decrease in tuberculosis incidence before any effi- number of determinants: political, sociological, psychological, cient control method has been available, and so on). cultural, economic, technical, sometimes religious, and so on. Opportunity, feasibility, short-term rather than long-term The Demand for Health Services considerations, legal and administrative settings, financing systems, and institutional framework are also essential. The perception of health needs by people differ, often strik- Decisionmakers' approval of programs is often lacking because ingly, from the assessment of needs by epidemiologists. they do not give indirect benefits the same weight as techni- Whereas the latter determine risk groups and priority diseases cians do. In periods of economic stringency, the practical on the basis of various technical criteria (life expectancy, problem is not to determine sectoral priorities but to find mortality, and the like), communities use a different value politically realistic ways of moving toward greater economic Rationales for Choice in Public Health: The Role of Epideemiology 745 efficiency in the very short run, considering the role and the tool which can lead to unequivocal and unchallengeable power of the actors involved (van der Werff 1986). choices. Experience has proved that the technical rationale, as pro- At the beginning of this appendix, we stated that it was of vided by, among other things, epidemiological analyses, is the utmost importance to identify criteria for decisionmaking relatively low in the hierarchy of factors that influence which could form the basis for a consensus among all actors in decisionmaking. The failure of the economic rationality to the health sector: providers, users, payers, and policymakers. become the instrument of choice for decisionmaking gives Epidemiology, as a science, is universally praised. Its impli- little chance of success to the epidemiological rationality to cations for behavioral changes in users of health services and fulfill this function. Had it been the case, tobacco would have for the setting of public policies are not readil' acceptable already been banned from the face of the earth. unless an intensive educational effort is undertaken. Thus, it The comparison of epidemiology with economnics as a tool is evident that any attempt from the payers and from the for decisionmaking can be elaborated further in the context of deciders to impose policy decisions on the basis of an epidemi- development projects. External assistance sources use effi- ological rationality will be rejected. Similar attempts from the ciency and effectiveness to demonstrate to theirconstituencies health professionals can be opposed by their political and that a high rate of return is obtained. Thus epidemiology is used financial partners as a way of preempting the decision for to quantify health retums and to maximize these retums technical reasons. This conflict can be detrimental, because through appropriate choices at the planning stage. On the the validity of epidemiology as an analytical tool is at risk to other side of the partnership, national authorities and deciders be denied, for reasons of policy implications and not of genuine responsible for implementation are sensitive to preferences criticism. derived from the value system of the communities. They are in the midst of the ethical conflict described above. Attempts to impose epidemiology as an indisputable tool for decisionmak- Notes ing, in view of the neutral character of the scientific analysis, are perceived as a limitation to the freedom of judgment, and The authorswish to thank David Parker. SeniiorAdviser, Health Financing as a technique to impose targets and objectives which meet the of the United Nations Children's Fund for his extensive review of this chapter. concerns of donors rather than the needs of beneficiaries. I. The average numberofdisability-adjusted life-davs lost to the community by each patient with a disease (L) can be calculated as follows (L = Days lost due to [premature death] + [disability before death] + [chronic disabilityl + Concluding Remarks [acute illnessl): Epidemiological information should be collected as early as L = (C/lQO x [E(A ) - (Ad-A,)I x 365.25) + (C/IQQ x (Ai-AA) x D,dIOO) x 365.'5) possible in public health programs because the quality of any + (Q/IQ0 x E(A,) x DI100 x 365.25) future evaluation depends on the accuracy of baseline data. + (lX)l-C-QI/ lOG xt), The lack of such data may hamper the assessment of effective- in which A,, = average age at onset of the disease:; Ad = average age at death ness and thus be detrimenital to the continuation of activities. attributable to the disease; E(A,,) = life expectancy (in vears) at age A,; C = There is no alternative to epidemiological methods when case-fatalitv rate (expressed as a percentage); D,d = percenitage of disablement evaluating effects on public health, between onset of the disease and each death attributable to it; Q = percentage Epidemiological techniques provide snapshots of the situa- of permanentlv disabled among patients whon have recovered; D = percentage tion as well as inidications on trends. They may even allow for bf disablement of those permanently disabled; t = average period of temporary disablement during acute episodes; I = annual incidence of the disease (new a ranking of diseases, using whatever indicator is relevant to cases per 1000 population). the stated objective and provided that associated ethical issues As a result, the total loss to the community attributable to cases of the have been properly explored and accounted for. The use of disease OccuLrring in any single year (R) is the total number of days R = L x I epidemiology for choices in hiealth policy implies a double leap (per groLIp of 1000 population) (Ghana Health Assessment Project Team forward: a leap from the ranking of diseases to the setting of 198 priority objectives for action, and a leap from technical prior- ities to allocating resources on a selective basis. In both cases, epidemiology alone cannot substantiate the References move. The tools used in this discipline are not relevant, and results are often misleading. At the planning stage, objectives Barnum, Howard 1987. EvaluatngHealthy Days2oLif4GainedfromHealth are determined on the basis of all the factors involved in the decisionmaking process. The results of epidemiological analy- Dempsey, Michael. 1947. "Decline in Tuberculosis: The Death R-te Fails to Tell the Entire Storv." American Reve fTbruoi 6176 ses are to be considered among other factors. The importance of each of these other factors and their interactions should others. 1986. "Estimation de limpact des principales maladiesen zone rurale malienne." Revue d'Epidemiologie et de Santepiublique balance the importance of epidemiologically assessed needs. 34:405-18. The choice of the epidemiological indicator influences largely Glhana Health Assessment Project Team. 1981. "A Quantitative Method of the outcome of the results. Thus, these results should not be Assessing the Health Impact of Different Diseases in Less Developed used to preempt the decision. Epidemiologv is not the neutral COLontries. Internanonallournml of Epidemiology 10:73-80. 746 Andre Prost and Michel Jancloes Jancloes, Michel. 1989. "The Case for Control: Forging a Partnership with Terris, Milton. 1980. "Epidemiology as a Guide to Health Policy." Annual Decision Makers." In D. W. T. Crompton and others, eds., Ascariasis and Its Retiew of Public Health 1:323-44. Prevention and Control. London: Taylor and Francis. Tugwell, P., K. J. Bennett, D. Sackett, and B. Haynes. 1984. "Relative Risks, Morrow, R. H., P. G. Smith, and K. P. Nimo. 1982. "Assessing the Impact of Benefits, and Costs of Intervention." In K. S. Watren and A. A. F. Mah- Disease." World Health Forum 3:331-35. moud, eds., Tropical and Geographical Medicine. New York: McGraw-Hill. Prost, A.. and N. Prescott. 1984. "Cost Effectiveness of Blindness Prevention van der Werff, Albert. 1986. "Planning and Management for Health in Periods by the Onchocerciasis Control Programme in Upper Volta." Bullean of the of Economic Stringency and Instability: A Contingency Approach." Inter- World Health Organizaton 62:795-802. nationalJournal of Health Planning and Managetnent 1:227-40. Romeder, J. M., and J. R. McWhinnie. 1977. "Potential Years of Life Lost Walsh, J. A., and K. S. Warren. 1979. "Selective Primary Health Care: An between Ages I and 70: An Indicator of Premature Mortality for Health Interim Strategy for Disease Control in Developing Countries." Neu Eng- Planning." InternanonalJournal of Epidemiology 6:143-51. land Journal of Medicine 301:967-74. 1%) 4? I  1' j.I PThz \7c.r)! Y. :Ys