Report No. 9825-TA Tanzania AIDS Assessment and Planning Study June 1992 Population and Human Resources Division M i' O1F I CHE 001 Y Southern Africa Department African Regional Office I,,rt. 1' . : it,u: 5 'I'A Ty: ( ;to) ) Tj . 1 1 Il JTh A,S'.-2MENT & l ANN r NI: ;'i .. Authloc: MURF1HY. JFAINNEPTE Ext- 34425 Roor J1 1 081 At)ep. A1 tl FOR OFFICIAL USE ONLY Document of the World Bank This document has a restricted distribution and may be used by recipients only in the performance of their official duties. Its contents may not otherwise be disclosed without World Bank authorization. CURRENCY EQUIVALENTS Currency Unit = Tanzania Shilling (TSh) US$1.00 = TSh 193 (February 1991) ABBREVIATIONS AND ACRONYMS AIDS Acquired Inmunodeficiency Syndrome AIDSTECH Technical Intervention for the Containient of AIDS AMR Adult Mortality Rate ARI Acute Respiratory Infection AZT Azidothymidine, former name for zidovudine, a drug BCG Vaccine to protect from TB CCM Tanzania National Ruling Party (Chama Cha Mapinduzi) CDD Control of Diarrheal Diseases program DAC District AIDS Coordinator DANIDA Danish International Development Agency DSM Dar es Salaam EDP Essential Drug Program ELISA Enzyme-linked Immunosorbent Assay EPI Expanded Program for Immunization FLE Family Life Education GDP Gross Domestic Product GPA WHO's Global Program on AIDS GTZ German Government's international aid agency HEP Health Education Project HEU Health Education Unit HIV Human Immunodeficiency Virus HLY's Healthy life years IEC Information, Education and Communication IMR Infant Mortality Rate jUWATA Trade Union of Tanzania KABP Knowledge, attitudes, beliefs and practices MCH/FP Maternal Child Health/Family Planning MOH Ministry of Health MTP Medium Term Plan NACP National AIDS Control Program NGO Non-Governmental Organization ODA Overseas Development Administration PHC Primary Health Care SIDA Swedish International Development Authority STI) Sexually Transmitted Disease TAC Technical Advisory Committee of the National AIDS Task Force TB Tuberculosis TBA Traditional Birth Attendant UMATI Family Planning Association of Tanzania UNICEF United Nations Children's Fund USAII) United States Agency for International Development UWT National Women's Organization of Tanzania WAMATA Tanzanian NGO to assist people with AIDS WAZAZI Tanzanian Parents' Association WHO World Health Organization FOR OMCIAL USE ONLY PREFACE The Tanzania AIDS Assessment and Planning Study was a joint undertaking of the Government of Tanzania, the World Health Organization (WHO) and the World Bank, growing out of discussions between Jonathan Mann (WHO) and Dean Jamison (Bank) in dhe late 1980s. The cost effectiveness framework was formulated by Mead Over, and the policy and implementation framework, by Jean-Louis Lamboray. The study comprised a series of desk and field studies (listed below) and a mission to Tanzania in April-May 1991 comprising Jealinette Murphy (senior operations officer and mission leader), Jean-Louis Lamboray (public health specialist), Richard Laing (public health consultant), Rami Chhabra (lEC consultant), Wendy Roseberry (pubiic health management consultant), Joy de Beyer (economist), and Larry Forgy (economist consultant). The report was drafted by Taryn Vian (consultant), drawing heavily on these background studies and on other references listed in the bibliography, and was finalized by Jeannette Murphy (task manager), following the main mission. Mead Over, Jean-Louis Lamboray and Martha Ainsworth contributed substantially to the study throughout its implementation. Professor Philip Hiza of the Ministry of Health coordinated Tanzania's participation; and invaluable assistance was prcvided by the National (Mainland) AIDS Control Program (NACP). Within WHO, Roy Widdus was responsible for overseeing the study, under the direction of Michael Merson. Within the World Bank, David de Ferranti and Stephen Denning were respectively the managing Division Chief and Department Director for this sector work. In its current form, the report reflects information available as of the end of December 1991. Background Studies 1. Chin, J. and Sonnenberg, F., "The Epidemiology and Projected Mortality of AIDS in the United Republic of Tanzania," WHO, Geneva, January 1991. 2. Bulatao, R., "The Demographic Impact of AIDS in Tanzania," draft mimeo, IBRD, Washington, D.C., June 13, 1990. 3. Cuddington, J., "Modelling the Macroeconomic Effects of AIDS with an Application to Tanzania," draft mimeo, Georgetown University, Washington, D.C., October 18, 1991. 4. Over, M. and Huq, M., "Economic Impact of AIDS on the Tanzania Economy," draft mimeo, IBRD, Washington, D.C., March 29, 1991. 5. Foote, D., "A Cost/Effectiveness Model for Comparing Interventions in Information, Education, and Communications about AIDS in Tanzania," Applied Communication Technology, Menlo Park CA, October 1990. This document has a restricted distribution and may be used by recipients only in the performance of their official duties. Its contents may not otherwise be disclesed without World Bank authorization. 6. Heald, G., "Estimating the Cost/Effectiveness of AIDS Information, Education and Communication Intorventions in Tanzania," on the basis of field work in November 1990, Communication Reeearch Center, Tallahassee FL, March 5, 1991. 7. Bjorkman, A., "Alternative Medical Interventions to Slow the Spread of HIV Infection," Karolinska Institute, Stockholm, January 1991. 8. Over, M., "Costs and Effects of STD Treatment as a Strategy for AIDS Prevention in Tanzania," on the basis of field work in November 1990, IBRD, Washington, D.C., February 12, 1991. 9. Over, M., "Costs and Effects of Blood Screening and Social Marketing of Condomis as Strategies for AIDS Prevention in Tanzania," on the basis of field work in November 1990, IBRD, Washington, D.C., April 10, 1991. 10. Laing, R. and Pallangyo, K., "Background Study on Alternative Approaches to Managing the Opportunistic Illnesses of HIV-Infected Persons: Costs and Burden on the Tanzanian Health Care System," on the basis of field work in August 1990, Management Sciences for Health, Boston, September 1990. 11. Bertozzi, S., Ankrah, M., Koda, G., Ngaiza, M., "Tanzania Assistance to Survivors of the AIDS Epidemic, a Review of Policy Options," on the basis of field work in March 1990, mimeo, IBRD, Washington, D.C., June 16, 1990. 12. Ainsworth, M. and Rwegarulira, A. "Coping with the AIDS Epidemic in Tanzania: Survivor Assistance," IBRD, Washington, D.C., December 1991. DEFINItIONS' AIDS Acquired Immune Deficiency Syndrome, a syndrome characterized by unusual opportunistic infections and rare malignancies in otherwise healthy individuals with no other reason for immune system compromise Case Fatality Rate Percentage of deaths due to a disease among the total number of people who have contracted the disease Child Mortality Rate Probability of dying between ages 1 and 5 Crude birth rate Number of births per 1,000 population in one year Crude death rate Number of deaths per 1,000 population in one year Dependency Ratio Population aged less than 15 and over 64 ("dependent population") divided by the population aged 15 to 64 ("productive population") HIV Human Immunodeficiency Virus, a retrovirus that causes the underlying damage to the human immune system which then permits opportunistic infections to cause virulent and fatal diseases in the HIV infected individual Incidence Number of new cases of a given disease which occur in a population during a specified period of time Incidence Rate Incidence expressed per specified unit of population (usually expressed as the number of cases per 1,000 persons) Infant Mortality Rate The number of deaths to infants (children under age one) in a given year per 1,000 live births in that same year Opportunistic Infections The many parasitic, bacterial, viral and fungal infections which are able to cause disease in an individual once the human immunodeficiency virus has damaged the immune system. These are the most common presenting clinical manifestations that establish a diagnosis of Sources: World Bank 1989, Volberding 1988, Center for International Health Information 1990, MacMahon 1970, UNICEF 1990. AIDS, and are characterized by an aggressive clinical course, resistance to therapy, and a high rate of relapse. Prevalence Frequency of a given disease at a designated point in time Prevalence Rate Proportion of a specified population which exhibits the disease at a specified time (often expressed per 1,000 persons or, in case of low figures, per 100,000 persons) Progression Rate Time from HIV infection to development of full-blown AIDS, sometimes expressed as the proportion of HIV infected individuals who will develop AIDS within a specified time period Seroincidence Number of cases whose blood shows evidence of antibodies to a given infectious agent during a specified period ot time Seroincidence Rate Seroincidence expressed per specified unit of population Seroprevalence Frequency of cases whose blood shows evidence of antibodies to a given infectious agent at a designated point in time Seroprevalence Rate Proportion of a specified population whose blood shows evidence of antibodies to a given infectious ageni ai a specified time Total fertility rate Average number of children that would be born alive to a woman during her lifetime given current age- specific fertility rates Under Five Mortality Rate The probability of dying between ages 0 and S TABLE OF CONTENTS EXECUTIVE SUMMARY i-xx INTRODUCTION I A. BACKGROUND I B. COUNTRY SETING 2 C. HEALTH STATUS AND HEALTH SYSTEM 2 - Morbidity 2 - Mortality 4 - Health System and Policies 6 nI. THE AIDS EPIDEMIC 9 A. CURRENT SITUATION 9 1. Data and Methods 9 2. Number of Cases, Infections and Deaths 10 - AIDS Cases 10 - Deaths from AIDS 12 - HIV Infection - 13 3. Distribution by Region, Gender, Age and Socio-Economic Status 16 - Geographic Distribution 16 - Gender and Age Distribr.tion i8 - Distribution by Socio-economic Status 19 - Prevalence Estimates by Special Risk Group 19 B. ROUTES OF TRANSMISSION AND ASSUMPTIONS REGARDING PROGRESSION 20 1. Major Transmission Routes 20 - Heterosexual 21 - Perinatal 21 - Blood Transfusion 22 - Contaminated Needles and Instruments 24 - Other Routes of Transmission 25 2. Assumptions Regarding Disease Progression 25 - Progression from HIV Infection to AIDS 25 - Progression from AIDS to Death 26 C. FUTURE DEVELOPMENT OF THE EPIDEMIC 27 1. Projections of Infections, Cases, Deaths and Distribution 27 - Infections and Cases 27 - CumulAtive Deaths 30 - Changes in Distribution 32 2. Impact on Mortality 33 - Crude Death Rate, IMR and Life Expectancy 33 - Empirical Data on HIV Prevalence and Mortality 37 3. Impact on Fertility 39 4. Migration 41 - Effect of Migration on AIDS 41 - Effect of AIDS on Migration 41 5. Population Size, Growth and Distribution 43 HI. CONSEQUENCES OF THE EPIDEMIC 48 A. MACRO-ECONOMIC IMPACT OF AIDS 48 - Introduction 48 - AIDS' Channels of Macroeconomic Influence 48 - A Simple Analytical Model 52 - The Simulation Results 53 B. IMPACT ON PUBLIC EXPENDITURES AND SERVICES 59 1. Health Services 59 - Impact on Health Care Delivery 59 - Cost of AIDS Care 64 2. Education Services 67 - Impact on Cohort Size 68 - Teacher Numbers, Turnover, and Training 69 - Impact on Quality and Efficiency of Education System 69 - Lost Investments in Education 69 - A Lower Demand for Education? 70 3. Impacts on Other Scawrs 70 C. IMPACT ON HOUSEHOLDS AND COMMUNITIES 71 IV. PREVENTING FUTURE HIV INFECTION 74 A. COSTS AND EFFECTS OF SELECTED INTERVENTIONS 74 1. STD Prevention and Control 74 - Effectiveness of STD Treatment 75 - Costs 82 - Cost-Effectiveness Estimates 84 - Implications for Policy Decisions 86 2. Condom Promotion 87 - Availability 88 - Use 89 - Costs 89 - Effectiveness of Condom Promotion and Distribution Programs 90 3. IEC and Counselling Activities 94 - Costs 95 - Effects 97 - Relative Costs, Effects and Cost-Effectiveness Ratios 97 - Implications for Policy Decisions 100 - Comment 102 4. Blood Screening 103 - Costs 103 - Effectiveness 104 5. Reduction in Blood Transfusions 106 - Use of Replacement Solutions 106 - Malaria Chemotherapy and Chemoprophylaxis 107 6. Possible Vaccine Development 109 V. COPING WITH THE AIDS EPIDEMIC 111 A. MEDICAL TREATMENT FOR PERSONS WITH AIDS 111 1. Treatment Options for Opportunistic illnesses 111 - Defining Treatment Objectives 112 - Defining Alternative Treatment Options 112 2. Episodes of Illness Per Case 113 3. Costs of Treatment 114 - Treatment Protocols and Cost Assumptions 114 - Cost Per Episode of Illness 115 - Cost Per AIDS Case 115 - Cost Implications of Alternative Treatment Scenarios 118 - Methodological Considerations 123 B. SURVIVORS 124 1. Growth in Numbers of Orphans and Other Dependent Survivors 124 - Numbers of Child Orphans 125 - Orphan Rates and Age of Child 125 - Orphan Rates and Adult Mortality 127 - Excess of Orphans Attributable to AIDS Epidemic 128 - Numbers of Widows, Widowers and "Elderly Orphans" 130 2. Survivor Assistance Programs 130 - Criteria for Selection of Alternatives 131 - Analysis of Options 131 - Implications for Policy Decisions 134 VI. THE NATIONAL AIDS CONTROL PROGRAM (NACP) 136 A. ESTABLISHMENT OF THE NACP 136 1. Background 136 2. NACP Organization 136 B. PROGRAM ACTIVITIES TO DATE 137 C. STD PREVENTION AND CONTROL 138 1. Efforts to Date 138 2. Planned Activities 139 3. Condom Promotion and Distribution 139 D. IEC INTERVENTIONS 1AO 1. Resources Available 140 2. Activities to Date 140 3. Effectiveness of IEC Efforts 141 4. Future Directions 141 E. ISSUES/RECOMMENDATIONS 142 1. Give the NACP the Status it Needs to be a Truly National Program 142 2. Concentrate NACP's Efforts on Recognized Priority Areas 142 3. Undertake an Aggressive National STD Prevention and Control Program 143 - Essential Elements 143 - Costs 145 4. Strengthen IEC for AIDS prevention 145 - A Communication Plan 145 - The IEC Unit within NACP 146 - AIDS Technical Committees 146 - Decentralization of IEC 147 - More Effective Electronic Software Development 147 - Development of a Multi-Media Campaign on AIDS 147 - Film Development 148 - More Emphasis on Youth 148 - Condom Promotion 149 S. Strengthen Intra-Sectoral Coordination within MOH 149 6. Strengthen Donor Coordination 151 7. Decentralize AIDS Planning 151 VII. RECOMMENDATIONS 153 BIBLIOGRAPHY 155 LIST OF TABLES 1.1 Disease Frequency in Sentinel Dispensaries, January 1984 - August 1987 3 1.2 Comparative Mortality Indicators, Tanzania and Neighboring Countries 5 1.3 Morbidity and Mortality Statistics, Diarrheal Diseases and Acute Respiratory Infections (ARI) 6 1.4 Total Health Facilities by Region and Management, Mainland Tanzania, 1988 7 1. 1 Seroprevalence in Blood Donors, by Region, 1987-1989 14 11.2 Cumulative Seroprevalence in Blood Donors by Region, 1990 15 II.3 Recent Estimates of HIV Seroprevalence Among Selected Risk Groups 20 1.4 Cumulative Percentage of HIV Infected Persons Who May Progress to AIDS Within a Given Period of Time From Infection, Adults and Infants 26 11.5 Cumulative Percentage of Persons with AIDS Who May Die from AIDS Within One and Two Years of Diagnosis, Adults and Children 26 11.6 Estimated Incidence and Prevalence of HIV in Mainland Tanzania 28 11.7 Projected Cumulative AIDS Cases and Deaths in Tanzania, All Ages 31 II.8 Changes in Mortality Indicators with AIDS by 2005-10, As Compared to Base Projections Without AIDS 35 11.9 Effect of AIDS on Fertility: Projected Change in Proportion of Women of Child-bearing Age by the Year 2020 40 HI. 1 Results of Simulation Modelling Macroeconomic Effects of AIDS in Tanzania 54 111.2 Output Projection: No-AIDS Scenario 55 111.3 Output Projection; With-AIDS Scenario (15% Monogamy) 56 M.4 Days of Care, By Episode and Treatment Setting, Adult AIDS Cases 61 111.5 Days of Care, By Episode and Treatment Setting, Pediatric AIDS 62 Cases 111.6 Estimated Costs of AIDS Treatment, 1991 65 III.7 Projected Expenditure for Treatment of AIDS Cases, No Resource Constraint 65 I.8 Projections of School-Age Cohorts 68 II1.9 Projected Teacher Needs for 34:1 Pupil:Teacher Ratio 69 m. 10 The Economic Impact of Adult Fatal Illness on the Household 72 IV. 1 Dynamic Effects on STD Transmission of Preventing 100 STD Cases in Core and Noncore Groups 78 IV.2 Discounted Healthy Life Years Saved Per Case Prevented or Cured When Epidemics are Independent: Core Vs. Non-Core 78 IV.3 Dynamic Effects on HIV Transmission of Preventing 100 STD Cases in Core vs. Noncore Groups. 80 IV.4 Discounted Healthy Life Years Saved Per Case Prevented or Cured When STDs Affect HIV Transmission: Core Vs. Non-Core 81 IV.5 Average Cost per Treatment of an STD Case 82 IV.6 Minimum Cost Per Effectively Treated Case of STDs: Sensitivity to Prevalence Rate and Diagnostic Procedure 83 IV.7 Cost Per Discounted Case of HIV Averted by STD Treatment: Sensitivity to Prevalence Rate and Core Vs. Non-Core Strategy 84 IV.8 Cost Per Discounted Healthy Life Year Saved by STD Treatment in the Presence of an HIV Epidemic: Sensitivity to Prevalence Rate and Core Vs. Non-Core Strategy 85 IV.9 Average Cost per Year of Protecton for a Condom Social Marketing Program in Tanzania 92 IV. 10 HIV Infections Averted Over Ten Years When 100 People Use Condoms for All Contacts for One Year 93 IV. 11 Cost-Effectiveness of Social Marketing Campaign to Induce 100 People to Use Condoms for All Contacts for One Year 94 IV.12 NACP IEC Costs (Total and Distributed Across Major IEC Activities/Media 96 IV. 13 Estimated Relative Cost/Effectiveness of Alternative IEC Interven.ions in Averting New HIV Infections (Intermediate Estimates) 98 IV. 14 Estimated Relative Cost/Effectiveness of Alternative IEC Interventions in Averting New HIV Infections (Optimistic Estimates) 98 IV. 15 Estimated Relative Cost/Effectiveness of Alternative IEC Interventions in Averting New HIV Infections (Pessimistic Estimates) 99 IV. 16 Summary of Relative Costs, Effects and C/E Ratios, IEC and Counselling 100 IV.17 Cost-Effectiveness of Anemia Prevention through Malaria Treatment and Prevention 109 V. 1 Average Number of Episodes of Opportunistic Illness per AIDS Case, Adults and Children 114 V.2 Costs per Episode of Opportunistic Illness Adult and Child AIDS Cases 116 V.3 Average Lifetime Cost per AIDS Case, Adults and Children 116 V.4 Days of Care, by Episode and Treatment Setting, Adult AIDS Cases., Decentralized Referral System 118 V.5 Days of Care, by Episode and Treatment Setting, Pediatric AIDS Cases, Decentralized Referral System 119 V.6 Implication of Alternating Treatment Scenarios on Cost per Case 120 V.7 Implications of Alternative Treatment Options on Total Direct Costs, 1990 122 V.8 Summary of Estimated National AIDS Treatment Costs in 1990 Under Alternative Treatment Scenarios 123 V.9 Estimated Proportion and Number of Orphans in Twenty Regions under Pour Scenarios, 1988 126 V.1 0 Analysis of Survivor Assistance Options 132 LISI OF FIGURES B. 1 Estimated Cumulative AIDS Cases, Adults and Children 12 11.2 Estimated HIV Prevalence Among Sexually Active Adults, Tanzania, 17 1989 11.3 Age Distribution of Reported AIDS Cases, Mbeya Region, Tanzania 18 A.4 Estimated Number of New HIV Infections Among Adults in Mainland Tanzania 29 11.5 Estimated Number of New HIV Infections Among Children in Mainland Tanzania 29 11.6 Estimated Cumulative AIDS Cases, Paediatric and Adult, in Mainland Tanzania 30 11.7 Cumulative AIDS Cases and Deaths 31 11.8 Annual New Adult HIV Infections, By Sex 32 11.9 Percent of Adult Males and Females Infected with HIV 32 11.10 Crude Death Rate (per thousand) 34 1.11 Infant Mortality Rate (per thousand) 34 11.12 Life Expectancy at Birth 34 11.13 Projected Change of Child Mortality due to HIV/AIDS in Tanzania 36 11.14 Projected Increase in Adult Mortality due to HIVIAIDS in Tanzania 36 11.15 Adult Mortality Rate, Twenty Regions of Tanzania, 1988 38 11.16 HIV Prevalence and Adult Mortality: Kagera Region, 1987-1988 39 11.17 Population Growth Rate Under Different Scenarios (percent) 43 lI.18 Total Population Under Different Scenarios 44 11.19 Dependency Ratio Under Different Scenarios 45 11.20 Youth Dependency Ratio, With and Without AIDS 46 11.21 Elderly Dependency Ratio, With and Without AIDS 46 11.22 Age Distribution in 2020, With and Without AIDS 47 iI. I Working-Aged Population, 1985 - 2010 50 111.2 Average Age of Labor Force, 1985-2020 51 I11.3 GDP Projection, with and without AIDS 57 11.4 Per Capita GDP Projection, with and without AIDS 57 m1.5 Capital to Labor Ratio Projection, with and without AIDS 58 mI.6 New Tuberculosis Cases in Tanzania, 1981-1989 63 111.7 Projected AIDS Treatment Costs Versus Available Public Funding (scenario 1) 66 111.8 Projected AIDS Treatment Costs Versus Available Public Funding (scenario 2) 66 111.9 Projected AIDS Treatment Costs Versus Available Public Funding (scenario 3) 67 IV. 1 Static Benefit of Preventing a Case of STD and Other Diseases in Sub- Saharan Africa 77 IV.2 Dynamic Benefit of Curing or Preventing 100 Cases of Gonorrhea in the Core vs. Non-Core Groups 79 IV.3 Total Health Benefit of Averting a Case of STD when STD's Exacerbate HIV Transmission: Core vs Noncore 82 IV.4 Cost per HIV Infection Averted by Blood Screening as a Function of Prevalence Rate and Test Cost 105 V.1 Contributions of Opportunistic Illnesses to AIDS Treatment Costs 117 V.2 Orphan Rates by Age Group, Six Regions of Tanzania, 1988 127 V.3 Adult Mortality and Orphan Rates, Six Regions of Tanzania, 1988 128 V.4 Orphan Rates and Adult Mortality, Kagera Region, 1987-88 129 EXECUTIVE SUMMARY OVERVIEW AND INTRODUCTION The rapid spread of AIDS will have far-reaching implications in Tanzania over the next several decades. This study assesses the current status, likely future development, and prospective demographic, economic, and other impacts of the AIDS epidemic, and examines the options available for doing something about it. The findings suggest that the epidemic will be severe and have major consequences. As will be described further below: - The number infected will reach 5.8 to 17.4 percent of the population by the year 2010, up from 1.4 to 5.3 percent currently. (These HIV-infected individuals will suffer debilitating illness, usually within one to ten years of infection, followed by certain death a year or two later, often in the prime of life or before.) In addition, a much larger number will be indirectly affected: as relatives, many of whom will incur significant costs on account of AIDS victims; as survivors, many of whom will be left in greater poverty; as earners, employers, or self-employers who will experience productivity losses; or as sufferers of other diseases in AIDS-induced resurgence (e.g., tuberculosis). - Demographic changes will alter the composition of the population and work force. Earners will have more dependents to provide for, as the ranks of working-age adults are thinned by rising mortality, while the young and the infirm become more numerous. The work force will become younger (average age 29 instead of 31 in 2010) and less experienced, and will have less education and training. Critical talents and rare skills - such as entrepreneurs, managers, and various professionals - will become even scarcer. - The economy will be adversely affected, mainly in the medium and long run. GDP will grow more slowly; 'ly 2010, it could be 14 to 24 percent lower than it would have been if there had been no AIDS epidemic. Per capita GDP also will be impacted, but more moderately, as the slower aggregate GDP growth is partially offset by slower population growth. - Some sectors, industries, regions, and subgroups will feel the effects much more than others. Kagera region, the worst hit area so far, will experience massive increases in mortality rates and unattached dependents (orphans, widows, etc.). It is already feeling the labor pinch, with crop production reportedly being adversely affected. Other regions may be similarly effected. Health services and budgets will come under increasingly intense pressure, as AIDS victims and other patients compete for the limited resources available. - Population growth, though slowed considerably by higher mortality, wil remain above 2.0 percent through 2010 or longer, and the dependency ratio will worsen slightly; population policies and programs, including family planning, will continue to be an urgent priority. - ii - The findings also indicate that effective action to slow the spread of the disease and improved coping measures X make a difference in mitigating the course and consequences of the epidemic. The most promising options are strengthening efforts to: - curb ith: sexually transmitted diseases (STDs) better, since their presence significantly increases the likelihood of AIDS transmission. Not only would a comprehensive STD/HIV prevention and control program decrease the risk of HIV infection for those with treatable STDs, it presents a unique opportunity for reaching those individuals at greatest risk of acquiring and spreading HIV with effective AIDS counselling; - increase condom usage, by actively promoting them for disease prevention as well as family planning purposes and by ensuring their ready availability; - design a variety of EEC messages specifically targetted to different audiences and utilizing a wide variety of media in order to create the synergistic effect necessary to change behavior; - reduce the need for blood transfusions (e.g., by improving prevention and treatment of other diseases - such as malaria - so that fewer patients reach the point of needing transfusions); - treat AIDS patients at primary facilities and at home rather than at hospitals, and thus realize significant cost savings. In addition, the adverse economic impacts can be moderated through actions to: - intensify public expenditure review and control, with a view to raising allocations to the health sector through reductions elsewhere, while also diminishing inefficiencies within the health system; - pursue vigorously the economic reform, so that the full growth potential of the economy can be realized; and - seek increased external funds, through donors and commercial investors, to compensate for the reduced domestic savings caused by AIDS. AIDS has come to Tanzania at a difficult time. Economic conditions and living standards (including health status) are still at very low levels, recovering from the setbacks of the early 1980s that reversed progress in the 1960s following independence. GNP per capita is about $100, and public spending on health per capita is under $4. Morbidity and mortality levels are unacceptably high. One out of every ten children born does not survive to its first birthday. More than three of every 1000 mothers-to-be die as a result of their pregnancy. The child mortality rate (prior to AIDS) was 147/1000 and the adult mortality rate, 7/1000, with an average life expectancy of about 48 years. - iii - EXTENT AND SEVERITY OF THE EPIDEMIC Cuffent Situation The National AIDS Control Program (NACP) estimates that about 800,000 people, or about 3.2 percent of the population, are currently infected with the disease (i.e., are HIV sero-positive). Of these, approximately 160,000 have already developed AIDS (i.e., have crossed the threshold from being infected to being ill). The remainder will develop AIDS sometime between less than one and up to twenty years from the date of infection (the median time for adults between infection and becoming ill appears to be eight to ten years in developed countries and may be less in developing countries). Death follows within a year or two of the onset of major symptoms, and often much sooner. Annual deaths from AIDS are presently estimated at between 20,000 and 30,000 which is 5-7 percent of total deaths. AIDS is believed to have recently surpassed malaria as the leading killer among diseases in adults, and is likely to do so for children in the very near future. Some parts of the country are much more affected than others. For example, in Kagera region, where the epidemic first manifested itself in Tanzania, reported AIDS deaths are about 16 percent of all deaths - and the actual figure may be higher due to underreporting. An estimated 17 percent of Kagera's urban population and 5 percent of its rural population are HIV infected; the proportions of productive adults infected is even higher. In contrast, in at least half of mainland Tanzania's 20 regions, only about 2.5 percent of the urban population and less than 1 percent of the rural population are thought to be infected. Whether these differences suggest different epidemics or different points in the same epidemic (i.e., whether all regions are likely to follow the Kagera pattern) is not yet known. This will depend on whether there are differences among regions with regard to the extent of sexual activity outside marriage and the prevalence of STDs in the populations, but information on these factors remains sketchy. There are also large disparities across different subgroups of the population. HIV infection levels are highest in people between the ages of 15 and 45, and in new-born infants. HIV infection rates are increasing especially rapidly among adolescents. Women and men are about equally infected. Women appear to become infected at a younger age than men on average. These facts are consistent with what is known now about how the disease is transmitted in Tanzania. By far the most important transmission route is heterosexual contact, which is estimated to account directly for close to 80 percent of all infections in the country. Blood transfusion and perinatal transmission have each been responsible for about 10 percent of infections; contaminated needles or other health care equipment is responsible for only 1-2 percent of infections; and transmission by other routes is negligible. This implies that those most at risk of acquiring AIDS are people with multiple sexual partners, their spouses and their unborn children. Future Prospects The likely future course of the epidemic will depend crucially on a number of vital and difficult-to-predict variables. By far the most important issue is what will happen to sexual behavioral patterns, and in particular, to the proportion of the adult population with multiple sexual partners. If, for example, both partners in 45 percent of married couples are "sexually monogamous", it has been estimated that, by the year 2000, about 1.2 million - iv - people will be carrying the virus and another 450,000 will have died of AIDS, with these numbers growing to about 2.3 million and 1.7 million respectively by the year 2010. If, on the other hand, only 15 percent of married couples are mutually monogamous, as many as 3.6 million could be infected by 2000 and AIDS deaths could be as high as 1.6 million, growing to 6.1 million infected and 5.6 million deaths by 2010. These figures imply that 3.9 to 12.4 percent of the population will be infected by 2000 and 5.8 to 17.4 percent by 2010. Even if massive efforts to reduce the spread of infection were successful in slowing transmission, the number of AIDS cases and deaths would still increase for the next several years as those already infected reach the stage of actually falling ill with the disease. In the extreme (and unlikely) case that transmission stopped completely as of today, the number of AIDS cases would still continue to grow to over 300,000 by 2000, and to more than 500,000 by 2010. Thus, no matter what happens, AIDS is going to have a major presence in Tanzania for many years to come, and the Government must plan now how to deal with that reality. IMPACTS OF THE EPIDEMIC The increases in both mortality (deaths caused by AIDS and by AIDS-induced growth in the prevalence of other diseases) and morbidity (illness - i.e., decline in healthiness - among those afflicted) will have a wide range of impacts, affecting in particular (1) demographic factors (population growth, age structure, dependency ratios, etc.), (2) the economy, (3) the health care system, and (4) other sectors. Impacts on Demographic Factors AIDS is expected to increase the adult mortality rate to about 25/1000 by 1994, from about 7/1000 in the late 1980s. The child mortality rate, which has been decreasing steadily for the past twenty years, is likely to begin increasing again within the next few years. Overall, the crude death rate could increase to as much as 24/1000 (from 20/1000 currently) and average life expectancy could drop to about 40 years (from 48). Population growth will remain positive, but the M of growth could decline from the current 2.8 - 3.0 percent annually to 2.0 - 2.6 percent by the year 2010. Thus, population growth pressures will still be a major problem, and concerted efforts to improve and expand ropulation policies and programs, especially family planning, will continue to be an urgent and high priority. Most of the epidemic's effect on population growth will be from increased mortlity; the impact on fertility is anticipated to be minimal. The age structure of the population will change in several ways. The 15-to-64 age group, i.e., working age adults, will become younger overall: their average age will decline to 29, instead of 31.5 in the absence of AIDS. The under-15 age group will be somewhat smaller than it would have been otherwise, due mainly to deaths of AIDS infants; but will have the same age composition. Because the losses in the under-IS group will be relatively fewer than the losses in the 15-64 group, the under-15 group will account for a larger proportion of the 0-64 population in the with-AIDS case than in the without-AIDS case. Thus the youth dependency rate will rise. The over-64 group will initially be unaffected; later, as the losses in the 15-64 group work their way up the age pyramid, the over-64's will be significantly reduced in number, compared to the no-AIDS case. These trends, together with the reduced size of the 15-64 group, will cause the elderly dependency ratio to be initially higher than it would have been, and then to be much lower. The overall dependency ratio, taking into account all these factors, will be slightly higher than in the absence of AIDS. In sum, the epidemic will lead to a younger (and therefore presumably less experienced) work force with a larger number of dependents to provide for. Significant increases will occur in the numbers of "unattached dependents' - orphans, widows, and families without able-bodied, working-age adults. In Kagera, there are a number of households consisting exclusively of dependents - young children and their grandparents. This development will impose extra burdens on others, including non-AIDS households, communities, and public authorities, as some victims move in with relatives and others become homeless scavengers. Impacts on the Economy The effects of the epidemic on mortality, morbidity, and demographic factors will affect the macroeconomy through multiple channels, starting with impacts on: - the labor force (the number of workers and potential workers, the age structure, the level of experience, the extent of training, etc.) and the dependent population (their number, needs, etc.) - the productivity of workers, as those who are ill, or caring for others who are ill, become less productive than if they were well - the human capital embodied in the work force, as well-trained, educated, experienced, and/or specially talented workers become less productive and die sooner than they would have otherwise. - expenditures occasioned by AIDS, including both public and private spending, at the individual household level and in aggregate.1' Flowing from these various effects will be others concerning the savings behavior of households and their non-AIDS-related spending. As victims, relatives, friends and others spend more on AIDS-related expenses, and as some of them also earn less because they work less, they will have fewer resources left for other purposes. This will undoubtedly result in a reduction in saving (or, equivalently, an increase in dissaving among those already spending more than their income). There may also be some reduction in expenditure for items unconnected with caring for victims, and possibly shifts in the composition of expenditure and saving. In addition, expenditure and saving behavior may be affected for other reasons too. For example, there may be an impact on the precautionary demand for savings by households who experience greater income variability in the presence of AIDS. 11 Expenditures on health services, though perhaps the main item here, will not be the only one. Households may incur many kinds of other "costs of caring" as well. In the end, too, there will be funeral expenses, which in Africa can often be extremely large relative to income. - vi - The effects on savings are especially important because economic growth prospects will be modified to the extent that savings - and investments -- are altered. In the likely event that the dominant effect on saving is the impact of increased spending to care for AIDS victims, the outcome will be a decline in saving. Other effects on saving might mitigate or exacerbate this decline, but the net result seems more probably to be a reduction. The effects on expenditure unrelated to AIDS may involve numerous adjustments across different markets. Some of these adjustments may be quite ordinary, in the sense of being little different, in their economic consequences, from the usual market shifts associated with changes in weather, tastes, external factors, etc. But a few may be of a more special nature. One might be a reduction in spending on education. If, as a consequence of spending more on health care and other things, people spend less on their children's schooling (and thus enrollments fall, dropouts rise, and learning achievement declines), the resulting setbacks in investment in human capital could constrain long term future growth. Taking all these effects together, and some further ramifications they may have, the study examined a number of basic questions about the possible economic implications of AIDS, including the following. Will the implications amount to much, or will they be minor? If more than minor, how will the time path of key macroeconomic variables (GDP, GDP per capita, saving, investment, real wages) in the AIDS-stricken economy differ over the next twenty or so years from what would have happened in the absence of the AIDS epidemic? To what extent can policy interventions alter the "with AIDS" outcomes? To analyze these issues, a simple conceptional framework was developed, and then extended to motivate a more involv;J simulation model that links conjectures about the demographic impacts of AIDS to the macroeconomy. Using two extensions of a well-known approach (the Solow growth model) and simulations that trace out the time paths of a "no AIDS" scenario and alternative "with AIDS" scenarios reflecting different policy options, the analysis considered how the various effects would interact with one another. Of particular interest were the conflicting influences of declining labor force growth, labor productivity, and national saving. Taken together, these and the other effects of AIDS might result in either increases or decreases in GDP per capita, the capital/labor ratio, wages, and other variables. The simulations suggest answers to these questions and indicate the likely magnitude and direction of key changes over time. The main results of this analysis suggest that if no new actions are taken (e.g., if no new Government measures or donor support are forthcoming), then: - GDP would grow more slowly than it would have in the absence of AIDS. The average real GDP growth rate through the year 2010 would be 2.9 - 3.4 percent, instead of 4.0 percent if there were no AIDS. By the end of that period, the level of real GDP would be 14 to 24 percent lower than it would have been otherwise. - Per capita GDP would also be affected but more moderately, as the impact of slower GDP growth is offset somewhat by slower population growth. The average real per capita GDP growth rate through 2010 would be 0.3 - 0.7 percent, instead of 0.7 percent in the absence of AIDS. Under some scenarios, it is possible that per capita GDP might remain unchanged, especially if the reduction in savings were small or nil. vii - - Capital/labor ratios would be affected slightly - possibly rising marginally more than it would have otherwise, as employers use non-labor inputs more so as to compensate for the losses In the numbers and productivity of workers. The time paths of these impacts are traced In the following figures for one of the simulation scenarios found to be among the most plausible, among the large number analyzed. Not surprisingly, the impacts start out as quite modest, but then build up over time as the prevalence, mortality, and morbidity from AIDS worsens. GDP Projection Ito, so - 70 s.o 40 0No &AM mb + Mo7d Uw Per Capita CDP Projection 2-2. 2.10 2 torn tom1 tifi torn t0 I rod un uIrd O Na Am * t~9 - viii - Capital to Labor Ratio Projection Tat lam 1 ....... .§ . . . . t o.,' p, 105 IO_ The effects within particular industries, sectors, regions, and subgroups of the population could of course be much more drastic. The loss of better trained, educated, or experienced personnel, or specially talented individuals could have major implications in some cases, especially where the activity is crucial to the overall development of the economy and/or there are pre-existing skill gaps and capacity scarcities. For example, losses of current or potential entrepreneurs, or among those farmers most predisposed to adopt new ideas, would be singularly damaging. Although nothing can be said with certainty about any of this, there is a substantial risk that as the epidemic spreads further, some vital elements of the economy could be severely affected, which would then make the aggregate impact even more pronounced. For individual households, too, the impact of illness and death of one or more family members is likely to be quite severe. In addition to the permanent loss of income or labor of the deceased - who may have been the primary source of support - household savings and/or equity may have been depleted caring for the patient. In a country as poor as Tanzania, it is likely that many families which had been managing, but , will be pushed into real poverty as a result of AIDS. In short, the ramifications of AIDS for the economy are likely to be by no means trivial. If no new actions are taken by Government and/or donors to counteract the consequences of the spread of the disease, GDP will grow much more slowly than previously projected, making the future development prospects of the country considerably more uncertain than otherwise. Per capita income levels will also rise more slowly, and although the per capita effects will be smaller than the aggregate effects, the reason for that outcome offers little solace: namely, AIDS deaths will reduce the number of people who otherwise would have been alive. Thus, the need to explore what can be done to contain the epidemic and to counterbalance its impacts is all the more urgent. Impacts on the Health Care Sys The consequences of the epidemic for health care provision and costs, both public and private, will be major, requiring difficult choices among competing demands for resources, - ix - not only between AIDS and other health issues (maternal and child health care, malaria, etc.) but also within AJDS care. Before dying, a typical adult AIDS patient wIll suffer two or more episodes of chronic diarrheal disease, ten episodes of oral thrush, and about three skin infections. In addition, approximately 50 percent of adult AIDS patients will fall Hl from pneumonia and/or septicaemia, and 15-25 percent will suffer from tuberculosis, severe headache, and/or neurological diseases, bringing the average number of episodes of illness per adult AIDS patient to 17 over the one to two years prior to death. All this sick time, if fully provided for, would require over 280 days of care, including home care. For infant AIDS patients, about six episodes of illness are expected, including a combination of chronic diarrhea, oral thrush, pneumonia and skin diseases, as well as recurrent fever and TB in some cases. The care burden for child cases would total about 200 days of care on average per patient over the year or less prior to death. It has been estimated that each adult AIDS case treated in the health care system absorbs about TSh 56,000 ($290) in nursing and drug costs, while the cost per pediatric case is about TSh 37,600 ($195). If all AIDS patients were being treated in health facilities, and sufficient drugs were available for this purpose, AIDS care would absorb about TSh 5.27 billion ($27.3 million) from both public and private giurces in FY 1991, which is equivalent to about half of the entire public health recurrent budget for that year. However, it is d:ficult to know how many AIDS cases are actually seeking care from the public or private health care system. On the one hand, only a fraction (10-15%) of the estimated number of AIDS cases have been recognized at health facilities as having AIDS, implying that many illness episodes are being treated by traditional healers or at home. On the other hand, many opportunistic diseases could be, and probably are, treated without being recognized as HIV-related. If only half of all AIDS cases were seeking treatment from public health facilities, AIDS expenditures would currently absorb about one-quarter of the Government's recurrent health budget. Given the projections of new AIDS cases in the future, the demnand for care is going to place extreme pressure on the health budget. It is unlikely that financing from the public sector for AIDS treatment will be sufficient, especially in view of the numerous other health needs of the population. Various projections of economic growth done as part of the recent economic report for Tanzania indicate that Government allocations for health will rise only marginally in the coming years unless the economic reform program is pursued more vigorously than at present. Only with a combination of vigorous economic reform and an effective AIDS control program to slow the epidemic is Tanzania likely to be in a position to provide reasonable care for all its AIDS victims while also meeting the health needs of the other 80-90+ percent of the population. Certainly, since the main threat of illness and death for the vast majority of Tanzanians will come from causes other than AIDS, most of them preventable, the treatment of AIDS cases must not be allowed to divert resources from primary health care programs, including those recommended in this report for preventing the spread of AIDS. Impacts on Other Sectors The impacts on other sectors besides health will be more complex. On the one hand, losses of personnel, especially trained, experienced workers and scarce talent, will create additional difficulties and costs. So will the extra expenditures that employers have to bear for sick workers. On the other hand, the slower growth rate of the population will mean that some services will not need to expand as rapidly as they would otherwise. Regarding the losses of personnel, much will depend on how easily those people can be replaced. For example, HIV infection is spreading very rapidly among truck drivers on the major transport routes, suggesting that this segment of the work force will lose a disproportionately large number of people. Replacements, in this case, should not prove too difficult. On the other hand, other sectors - such as energy and health, for example - already employ the relatively few highly-skilled and experienced people in the country with the specialized skills on which the sectors depend. The loss of only some of these people could affect the efficiency and productivity of these sectors; replacing them could take many years, suggesting that an even greater effort will be needed in building Tanzaria's human resource capacity. Regarding the impact of a slower population growth rate, many sectors are already falling so far short of meeting the needs of the existing population that a slightly slower growth rate will not have much practical effect. Needs for extension and improvel..ent of services will still be vast, stretching providers - and Government as a whole - to the limit of the resources available. Considering these and other possible sources of impacts, one of the most significantly affected sectors will be the education system. The slower-population-growth effect will be evident in the numbers of children to be educated. This will not mean, however, that expenditure on education can be reduced, since there is currently substantial under-funding relative to the need. Only 68 percent and 5 percent, of the relevant age groups attend primary and secondary school, respectively; books and other teaching materials are in very short supply; and teachers are underpaid. A sharp increase in per student expenditures on education will be required if quality is to be raised to the required level. There will also be a large losses of personnel as substantial numbers of teachers succumb to the epidemic. The loss of more experienced teachers is likey to result in diminished internal efficiency of the education system, raising the cost per graduate. In addition, the considerable investment already made in the education of those who die from AIDS will be lost, lowering the returns to education realized, on average, by graduates, their families, employers and communities. This could reduce the demand for education in the future, unless the loss of the highly educated drives up their wages enough to offset uncertainties about survival. Finally, the increased mortality rate among skilled people in all sectors implies that the rate of investment in human capital formation will have to be increased. RECOMMENDATIONS FOR ACTION AIDS will be a prominent and pervasive issue for a long time in Tanzania as elsewhere, but there a actions that Government, communities, individuals, and the international community can take to mitigate the severity and consequences of the epidemic. Much can be done through both prevention measures to slow the further spread of the disease as much as possible and coping measures to deal with the consequences that caniot be prevented. - xi - The prevention measurcs that are being or could be applied are diverse. An analysis of the various possibilities and their costs and effectiveness in the Tanzanian context was done for this study, the results of which are discussed in following sections under headings on: - STD prevention and control programs (i.e., strengthening of efforts to control the major, treatable sexually transmitted diseases, since having another STD vastly increases the chances of being infected by AIDS); - condom programs (enhancement of supply and distribution systems, and expansion of demand through active promotion); - IEC programs (improvement of information, Cducation, and gommunication activities directed at changing behaviors at high risk of facilitating AIDS transmission); - blood screening (to lessen the possibility of infection through donated blood); - reducing the need for blood transfusions (e.g., by improving prevention and treatment of other diseases - such as malaria - to lower the risk of becoming sick enough to require a transfusion). Immunization with an effective and affordable vaccine is, unfortunately, not an option now and cannot be expected to become one for a number of years, if at all. Recent information from the worldwide vaccine development effort is not encouraging: even when a vaccine is eventually discovered, fully tested, and evolved to the point of being ready for mass distribution (all of which could take a very long time), the cost may be prohibitively high and/or the difficulties of mounting a successful immunization campaign may be overwhelming (e.g., there could be special technical requirements in the handling of the vaccine, as m the case of the cold chain needed for the measles vaccine, and the logistics of delivering it could be as difficult as for the childhood vaccines, which still have not been successful in eradicating those diseases). Thus, strategies for responding to the AIDS problem, in Tanzania as elsewhere, should at this point poI be counting on new breakthroughs in vaccines, and instead should be moving ahead vigorou ly with appropriate other actions. To pin hopes on some coming vaccine "miracle" that never happens would only retard progress on other more promising options. Even with a vaccine, behavior change will be necessary to stem the epidemic. The coping measures also are diverse. The results of the analysis done for this study on the alternatives and their pros and cons are discussed belo-v in sections on: - Treatment protocols (i.e., revamping the pro( lures used in the handling of patients, depending on their problem, so as to Alocate resources more effectively; both for AIDS and non-AIDS cases) - Survivor assistance (to deal with the large increases expected in the numbers of orphans, widows, workerless households, etc., and with the adjustments that households and communities must make to higher mortality and morbidity rates) - Actions to address the economic impacts. In both the prevention and the coping measures, the findings of the analysis suggest that current efforts could be improved upon significantly and the limited resources available could be utilized much more efficiently. Also, the cost-effectiveness ef specific interventions varies substantially from one locality to another, depending upon a number of factors, the most important being the HIV prevalence in the population which is the target of the intervention and the number of trained personnel who can be spared for a specific activity, - xii - given other demands on their time. Thus, planning for prevention and coping measures needs to take place at the local level. In the selection of some measures, particularly for prevention, the fact that almost 90 percent of new infections occur as a result of heterosexual transmission (given that most women who transmit it perinatally were themselves infocted through sexual transmission) is a key factor in determining which options emerge as the most cost-effective. The best options, as described below, all relate to changing sexual behavior or reducing the risk of infection through sexual transmission. The success of efforts to slow the spread and impacts of AIDS will depend critically on how resistant the high risk groups are to altering practices and mores that have long been customary and widespread (e.g., as regards having more than one sexual partner and not using condoms). Prevention Measures STD Prevention and Control Programs. The presence of certain STDs other than HIV/AIDS in either partner has been estimated to increase the likelihood of transmitting the virus by 3 to 10 times, depending upon the STD in question. Since research has shown that the risk of heterosexual transmission is quite low in the developed world, the high prevalence of STDs may help to explain why the spread of AIDS heterosexually has been so rapid in Africa. The presence of certain STDs not only improves the biological environment for HIV transmission, it is also indicative of high-risk behavior. This means that an effective STD prevention and control program also presents an opportunity for one-on-one counselling in STD/HIV prevention to those most at risk of becoming infected, an extremely cost-effective approach to such counselling. The cost-effectiveness of treating different STDs as a means of averting HIV infection in Tanzania is presented below. As can be seen, with an HIV prevalence of 5 percent, the cost of treating various STDs (among people with multiple sexual partners) per case of HIV averted is only TSh 386 ($2.00) to TSh 2,900 ($15). This is a significant finding, especially when one realizes that the prevention and control of STDs has significant health and fertility benefits quite apart from HIV prevention. Table: STO Treatment Cost oer HIV Infection Averted (USS) HIV Sero-Prevalence Level in Population Considered 1X 5X 10X 25X A. Core arou8 Chancroid 39 a 4 2 Chlamydia (f) 74 15 8 3 (m) 24 5 2.80 1.40 Gonorrhea (f) 35 7 3.50 1.40 (m) 15 3.30 1.30 0.90 SyphlLis 11 2 1.20 0.40 B. General wooutation Chancroid 545 110 55 25 Chlamydfa (f) 887 177 90 37 (m) 293 60 33 17 Gonorrhea tf) 380 75 37 15 (m) 165 35 20 10 Syphilfs 123 25 13 5 Source: Over 1991. - xiii - Although, theoretically, STD prevention and control is included in the preventive health services offered by the Ministry of Health, and is specifically mentioned among the activities to be pursued by the NACP, the program has been largely dormant for many years. Since little reporting occurs and no formal STD program exists, one must assume that existing STD services are poor. The regular screening of pregnant women for syphilis was discontinued in 1985 and no education campaign to inform people about STD symptoms has been carried out in recent years. There are plans to establish five STD clinics, and to improve STD services in more than 100 health facilities, through a phased approach involving training and drug supply to be financed by EEC. This effort will need to be complemented by a comprehensive STD prevention and control program involving education and counselling, patient risk assessment, partner referral, monitoring and evaluation and operational research. It is recommended that NACP make implementation of a national STD/HIV prevention and control program the main thrust of its future activities. with first attention being given to those public and private facilities serving urban. young adult populations. Condoms Programs. Condoms, if of good quality and properly used, have been found to be very effective in preventing the transmission of HIV, and of other STDs. How cost-effective they are in slowing the AIDS epidemic depends upon the probability of one of the partners being HIV sero-positive, the quality of the condoms supplied, and whether they are used regularly and correctly. While condoms can be procured in bulk from manufacturers for as little as TSh 4-10 ($0.02-0.05), to this must be added the cost of distributing them. The cost per condom in a program in Tanzania has been estimated at TSh 73 ($0.38). At this cost, if one had sex once each with 100 different partners among whom the HIV sero-prevalence was 5 percent, the cost-effectiveness of using condoms would be TSh 1,467 ($7.60) per HIV infection averted. Because condom use is not very prevalent in Africa, cost-effectiveness analysis usually assumes that they will be made available through a social marketing program, the costs of which need to be added to the analysis. The costs of a social marketing program in Tanzania directed at the general public, if no charge is made for the condoms, is estimated to be quite high and is not by itself a cost-effective approach to averting HIV infection. However, condom promotion as part of an STD control program. and social marketing programs targetted to very high risk groups can be very cost-effective and should be pursued. For the general population. much more effective promotion of condoms for both family planning and disease prevention is required. and policies should be followed to bring the cost of condoms down to affordable levels for users by. for example. removing current taxes and imnort duties. and negotiating bulk purchases at low cost from suppliers. Information. Education. and Communication Programs. A wide range of options are possible here, involving media campaigns (radio, newspaper, etc.), distribution of materials (posters, calendars), changes in school curricula (in primary and post-primary education), community awareness efforts, family counselling, and more. In principle, IEC options ought to be highly cost-effective, given that (1) they focus on informing people and changing behavior which are crucial for slowing the spread of a disease like AIDS, where behavior is the key to transmission, and (2) they are low cost compared to most other options, especially those requiring labor and resource-intensive delivery of services. In practice, however, it is extremely difficult to design EEC messages which will be successful in actually changing - xiv - behavior. STD/HIV is particularly prone to this problem, since sexual practices are among the most resistant to change of all areas of behavior. Experience to date with AIDS IEC programs have consequently been disappointing, world-wide and in Tanzania. Nevertheless, IEC options should continue to be kept high on the list of priorities for prevention interventions and every effort should be made to acquire the special expertise required to design effective IEC programs. Considering how critical the behavior change element is to containing AIDS, even partial success in the IEC domain can have considerable importance, and a breakthrough to greater success would be enormously beneficial. Furthermore, where EEC has worked well, they have been capable of bringing about major improvements that would not have been possible otherwise. Continued searching for new and better ways to make IEC more effective - and overcome the resistance of old behaviors to modification - needs to be pursued vigorously. A knowledge, attitudes, beliefs and practices (KABP) survey carried out by researchers at the University of Dar-es-Salaam on a representative sample of Tanzanian households has shed light on the current situation in Tanzania. The survey found that 97 percent of respondents were aware of AIDS and over 80 percent knew the major routes of transmission. The vast majority of those who had heard of AIDS learned about it from the radio. Despite this high degree of awareness, high-risk sexual behavior remains common. A recent study on "partner relations" concluded that 20 percent of sexually active men and 15 percent of sexually active women had sexual partners other than their spouses, and an even greater proportion engaged in commercial sex, defined broadly as sex in exchange for money, gifts or other favors. Condom use was negligible both in and outside of marriage, and even by those who had STDs. In this study, 80 percent of people engaged in high-risk behavior did not perceive themselves to be at risk of acquiring AIDS. A smaller study of truck drivers, casual laborers and secondary school graduates in Dar es Salaam found consistent results. A high proportion of respondents engaged in extramarital sex (75%, 43% and 35%, respectively), and sex with barmaids or prostitutes was not uncommon (44%, 16% and 10%, respectively). Although knowledge of condoms was fairly high, they were rarely used. Tanzania has several [EC programs underway currently. Although data on them are limited, a program relying on radio spots was considered to be the most cost-effective in a recent review involving a committee of experts under the auspice of the NACP. The committee also concluded that posters/calendars, cartoons, newspaper "newspaper flashes", and family counseling were, respectively, 20, 18, 15 and 3 percent, as cost-effective as radio spots. This analysis used a mathematical model to assess the relativ costs and effectiveness of interventions. The absolttt cost-effectiveness of these programs is difficult to determine from the evidence available. Thus far, they have mostly been directed at the population at large, rather than at specific risk groups or target groups, and have been largely limited to informing people about the disease rather than changing behavior. IEC directed at chaneing behavior is known to be most effective when it is specifically designed for population sub-groups with different characteristics. A distinction should be made between IEC directed at encouraging more responsible sexual behavior (which can probably be achieved only over time), and that developed for saving lives immediately, with means found for delivering these messages to appropriate risk groups. An impoi =ant target group will be adolescents, among whom HIV - xv - prevalence is growing dramatically and who have been found to be less well informed than adults on matters concerning AIDS. Another crucial target group will be people with multiple sexual partners. Studies carried out for this report have shown that effective EEC campaigns directed at people with multiple sexual partners, who are therefore at extremely high risk of contracting and spreading AIDS, have a much bigger payoff (6-8 times) in slowing the spread of AIDS than do similar campaigns directed at the population at large. One such target group is prostitutes and their partners. A pilot project supported by AIDSTECH and AMREF to promote condom use by truck drivers and prostitutes along a major truck route in Tanzania is underway, but results are not yet available. A similar program in Kenya, however, was found to be very cost-effective, costing only TSh 1,200 ($6) per case of HIV infection averted. In addition to better targeting of IEC messages tailored for specific sub-mgoups. improvements are required in the content and delivery of EEC activities. Additional communication channels must be tapped, and more imaginative use made of different media. As a start, the Ministry of Health (MOH) should examnine each aspect of its operations for opportunities to deliver carefully constructed IEC messages on AIDS in the context of health care delivery; STD services are one obvious delivery point; fertility and family planning counselling for men and women is another. But a!! arms of the Government, the Party, NGOs and society in general should be utilized for reaching specific population groups. For example, AIDS education appropriate for different school-age children should be integrated into the family-life education curriculum now being developed and tested. The entertainment industry should be co-opted for reaching adolescents, both in and out of school, with messages that young people are likely to respond to. The cooperation and active participation of labor and industrial groups should be solicited to develop AIDS prevention programs for workers. The assistance of religious leaders should be sought, not just to exhort their followers to lead "moral" lives, but to develop a strategy for changing the sexual mores of society. The NACP's stated intention to decentralize IEC is sound, and this should be done in such a way that AIDS EEC can be fully integrated with other health education efforts once the planned reorganization of the MOH's health education unit has been realized. As a first step, it is recommended that the regional and district AIDS control coordinators be relieved of whatever clinical responsibility they now carry and be made primarily "communicators." with responsibility for working with other groups at the local level in developing EEC campaigns appropriate for that specific population. The NACP should probably limit its IEC activities to providing support to these local efforts. developing IEC messages and campaigns to be directed at specific high-risk groups. and mobilizing society at large through workshops with policy makers and opinion leaders. The support aspect of NACP's family counseling activities is probably more appropriately carried out by the Social Welfare Department in the Ministry of Labour and Youth Development, a shift which would free the few staff in NACP with communications expertise to concentrate on developing programs for high risk groups and advising other sectors/bodies in the design of IEC for their constituencies. Finally, the success of the AIDS IEC prevention campaign will depend critically on the visibility of the support it receives from the highest level of Government. Hopefully, policy makers are beginning to be aware of the devastation which the disease is reaping on - xvi - segments of the population, and the threat that its continued spread implies for the country as a whole. The epidemic needs to be treated as a crisis. The population must be made aware that it is not limited to a few unlucky people but that, the more the infection spreads, the higher the probability of any single person acquiring the disease, even those who are themselves sexually monogamous. Blood Screening. Although contaminated blood has been a relatively unimportant mode of HIV transmission in Tanzania, accounting for only 10-12 percent of infections to date, considerable resources and effort have been expended to reduce transmission via this route. One hundred sixty-eight blood transfusion centers have been equipped with simple rapid kits for testing HIV-1 antibody, and an estimated 80 percent of donated blood are now being screened prior to transfusion. The cost-effectiveness of blood screening for averting HIV-infection is very sensitive to the prevalence of HIV in the population. If the HIV sero-prevalence level is only 3 percent, as it is estimated to be in the Tanzanian population at large, screening blood prior to transfusion costs about TSh 96,500 ($500) for every case of HIV infection averted, because one is likely to test many blood samples before an infected sample is encountered. If HIV sero-prevalence is as high as 25 percent, however, the cost-effectiveness of blood screening drops to TSh 12,900 ($67). In contrast, decreasing the need for blood transfusions by using replacement solutions where this is possible, or by preventing anemia in the first place, can be very cost-effective. Reducing Blood Transfusioxls. It has been estimated that about 100,000 blood transfusions are given annually in Tanzania. Women, and children under five, are more likely to require blood transfusions than men, and therefore to acquire HIV via this route, because of the high prevalence of anemia among them, due to inadequate nutrition and to the severity of malaria attacks. Between 40-60 percent of transfusions are due to severe anemia in children, while 10-40 percent are due to severe anemia in pregnant women, before or after birth. Blood transfusion always carries risks, both because of the diseases which can be transmitted via this route (HIV, malaria, syphilis, hepatitis B & C, etc.) and because of the impaired healing and increased risk of secondary infections when done after surgery. For this reason, replacement solutions (plasma expanders) are often preferable to transfusion of blood for urzent volume replacement following acute bleeding. For anemia, however. prevention is preferable to treatment. The table below indicates the cost-effectiveness of these interventions for averting HIV infection. Preventing anemia in women and children, of course, has considerable benefit apart from slowing the AIDS epidemic. Table: Cost-Effectiveness of Selected Interventions for Averting HIV Infection* Us$ A. Blood screening for HIV 170.00 S. If blood is not screened - Use replacement solutions, when possible 20.00 - Treat malaria in children 4.50 - Give malaria prophylaxis to children 18.80 - Give malaria prophylaxis during pregnancy 2.20 Souce Biorkiman 1990. Ntent: * Assuming an HIV sero-prevalence of 10.2X in urban adults and of 4.7X in rural adults. The MOH supports an extensive maternal and child health program which is meant to identify women and children at risk. The high maternal and infant mortality figures for Tanzania indicate that it is only partially successftl, however. Recognition that effective - xvii - MCH services can help to stem the AIDS epidemic underscores the importance of these services for improving the health status of Tanzania's population. Coping Measures Care and Treatment. As stated above, it is estimated that the care and treatment of AIDS cases in Tanzania is currently costing about $290 per adult and about $195 per child. These costs are thought to be minimal in that they assume that the least expensive drug is used and only palliative treatment of opportunistic illnesses is offered, except in the case of communicable diseases such as tuberculosis and pneumonia, where the infective agent itself is treated. Notwithstanding this assessment, it is essential that ways be found to bring these costs down, on both equity and efficiency grounds. The average recurrent expenditure on health in Tanzania is only about $2 per capita, and money spent on treatment of the terminally ill is likely to be diverting resources from that which could save lives. Two cost-effective analyses of the current treatment regime have been carried out. The first compares it with a treatment alternative which concentrates the care of AIDS patients at lower level facilities, supplemented by home care, as compared to the current situation where suspected AIDS cases are sent to regional and national referral hospitals and cared for there until health authorities can do no more for them. Drugs were found to account for only 25 percent of the treatment costs; nursing and institutional care accounted for the other 75 percent. Because these latter costs are significantly greater at higher level facilities, considerable cost savings can be realized by caring for AIDS cases at health centers and dispensaries. It is estimated that the cost of treating adults could be reduced by 25 percent (from $290 to $218) while the cost of treating pediatric cases could be reduced by 21 percent (from $195 to $155). By the year 2000, such savings could total as much as TSh 9.5 billion ($49 million) annually, under the worst case scenario, and even more in the following years. If the economic reform program is vigorously pursued, there would be some hope that Tanzania could cope with the epidemic. AIDS treatment costs would total roughly TSh 9.5 billion ($49 million), as compared to a projected recurrent health budget of TSh 27.2 billion ($141 million). The second analysis compares the current treatment regime to one in which all of the WHO-recommended drugs are made available to a small proportion (1 %) of adult AIDS patients. This analysis was carried out, not because it appears to be a desirable option, but because pressure may be brought to bear on Tanzanian health authorities to use the more powerful drugs, given that they appear on the WHO-recommended list. Because the cost of these more powerful drugs is significantly greater than the alternatives being used in Tanzania, even this very limited use increases the average treatment cost for adults by 9 percent ($290 to $317). The expenditures associated with treatment of that one percent of the population, under this hypothetical situation, amount to TSh 538,500 ($2,790)/patient. By the year 2000, the impact upon the health budget would be to increase it by TSh 854.6 billion ($4,428 million), or 7 percent more than it would have been if the most expensive drugs are excluded entirely. These analyses indicate that the cost of caring for AIDS patients must be carefully monitored and controlled to ensure that they do not divert resources from preventiv measures. All possible measures should be aken to keep the cost of treatment low, by - xviii - limiting drug use. by strict adherence to treatment protocols, and by caring for patients at peripheral facilities and at home. Survivors. While there is a distinct possibility that one impact of AIDS will be to increase the extent of poverty in Tanzania, assistance programs targetted exclusively at AIDS survivors (orphans, widows, widowers, and surviving parents) will not necessarily reach those mo in need. For every AIDS orphan in Kagera in 1988, there were eight other children who had lost one or both parents due to other causes, and whose plight might be as or more desperate than those orphaned by AIDS. This report analyzes four options for rendering assistance to survivors of the AIDS epidemic, all of which are being undertaken on a limited scale in the country. These options are: (a) caring for orphans in orphanages; (b) direct transfers of cash or goods; (c) subsidies for school uniforms and/or iees; and (c) child care and feeding centers. Two issues will confront all programs targetted exclusively at AIDS survivors: cost and equity. Although the costs of the four options analyzed varied considerably, all would become prohibitive if such assistance were to be provided to all those adversely affected by the death of a loved one from AIDS. For example, the cost of caring for a child in an orphanage can run to TSh 125,000 ($650) to over TSh 173,000 ($930) a year. The costs of other programs are less (between TSh 4,000 and TSh 30,000 ($20-lS0) per child per year) but still unsustainable given the numbers of orphans likely to result from the AIDS epidemic. In Kagera region, alone, more than 5,600 children are thought to have been orphaned by the disease. Even a program costing only $20/child could not be afforded by Tanzania given the many other legitimate demands on its scarce resources. Because AIDS strikes rich and poor alike, AIDS victims and their families are not necessarily more deserving of assistance than their neighbors. Many of the victims come ftom better-off families; there may well be other families in the community untouched by AIDS in greater need of assistance. Assistance targetted to AIDS survivors - because of the large numbers involved - could well divert resources from programs designed to reach the poorest segments of society. Given the magnitude of the survivor problem in general and the limited resources for their support, it is imperative that programs to assist survivors: (1) establish criteria for identifying those most in need of assistance; and (2) adopt low-cost, sustainable interventions that will maxnimize the impact of program resources. The Progra=s most likely to assist thos AIDS survivors who are in greatest need. without diverting resources from others equally in need,are those tL tted at the poor. rather than those targeted to AIDS survivors. This suggests that resources should be directed to programs such as the "Iringa-style" child survival programs, targeted at communities rather than individuals, which depend upon community involvement and commitment and assist needy families generally. This underscores, once again, the fact that the decisions regarding allocation of scarce resources for coping with AIDS must be taken at the local level. Moderating the Adverse Economic Impacts Even with a concerted and successful effort to combat the spread of HIV infection as outlined in this report, AIDS and its effects will be felt on the Tanzanian economy for many years to come. What can the Government do to lessen these effects? In addition to the - xix - recommendations made in this report for slowing the spread of HIV, the report's analysis indicates that four actions are required: 1) the economic reform must be pursued vigorously and with determination; 2) mechanisms must be established to ensure that the care of AIDS patients does not divert resources from preventive measures to slow the spread of the disease and from other priority health interventions; 3) donor aid must be increased to offset the reduction in public and private savings and to respond to increased poverty; and 4) investment in human capital should remain a high priority. Chapter III describes the likely macro-economic impact of AIDS on the economy and explores the probable implications for public expenditures, comparing the increased demand for health services brought about by the disease with the expenditures likely to be available to the public health sector under three different reform scenarios. The simulations suggest that the faster the economy is able to adjust to market price signals, the smaller with be the negative consequences of the AIDS epidemic on economic growth. Further, the analysis indicates that, without rapid economic growth stimulated by the reform program, Tanzania cannot generate the public expenditures required to maintain its current public health effort, including more effective prevention and control of other STDs, and meet even a small proportion of the costs of caring for AIDS patients. The success of the fight against AIDS will therefore be closely linked with the determination of Tanzanian policymakers to pursue economic reform. including a maior restructuring of the budget. the key parastatals. and the onerative and financial sectors. Because caring for those who are ill will place inevitable pressure on health facilities, ways must be found to insulate the fund3 and gersonnel required for preventive health programs from the growing demands of patient care. This can be done, in part, by careful design and implementation of budgets, by improved planning, and by establishing and following treatment protocols which are resource saving. Of primary importance, however, will be acceptance by health planners and policy makers at every level that these hard choices are required for the ultimate good of Tanzanian society. Even if minimal public resources are used for the care of AIDS patients, the growing number of ill will inevitably reduce public and private saving that would otherwise be available for productive investment. Minimizing the adverse impact which this "dissaving" will have on Tanzania's future growth prospects will require increased financial assistance from multilateral and bilateral donors. The Bank should, in its economic work, attempt to estimate the magnitude of these increased resources and assist the Government in mobilizing them, to the extent that the country continues to pursue its economic reform program vigorously and imaginatively. To counter the slower growth of the economy due to AIDS, increased investment in both human and other capital will be required. While more study will be required to determine the appropriate balance between the two, it seems clear that increased attention to building the human resource base will be necessary, given the fact that certain skills are already in short supply and many experienced workers will die prematurely as a result of the - xx - epidemic. This argues for an even greater effort by the Government and donors to Improve and expand the education sector at all levels. Special attention will need to be given to girls' education, because of their particular vulnerability to increased demands at home and the positive links between the level of female education and other development goals. I. INTRODUCTION A. BACKGROUND 1.01 The National AIDS Control Program (NACP) estimates that about 800,000 Tanzanians are now infected with the human immunodeficiency virus (HIV), which causes AIDS.1' Experts believe that most, and probably all, people infected with HIV will eventually succumb to the acquired immunodeficiency syndrome after first suffering from sometimes long and usually very painful illnesses. AIDS has no cure. The morbidity and mortality caused by AIDS will have an enormous impact on Tanzania's economic and social development as well as on health services delivery and family welfare. 1.02 For the vast majority of Tanzanians who are not infected with the HIV virus, prevention activities are vital. Yet the epidemiological patterns of AIDS - how quickly and by what means the disease can spread, characteristics of those infected, regional variation in prevalence and risk behaviors - as well as countervailing demands of other serious health problems in Tanzania have made it difficult to choose the best strategies for intervention, let alone evaluate their success. 1.03 The purpose of the AIDS Assessment and Planning Study is two-fold: first, the study seeks to help Tanzanian decision makers develop clear intervention strategies to cope with and prevent further spread of the AIDS epidemic. Second, the study provides a framework for more effective decentralized health planning and resource allocation at the regional and district levels. 1.04 Specifically, the objectives of the study are to: 1. Assess the impact of the epidemic; that is, the health impacts, social service needs and economic burdens posed by HIV infection and AIDS; 2. Estimate the cost and probable effectiveness of interventions, both those to prevent further spread of the epidemic and case management and survivor assistance strategies to cope with the needs of existing and future AIDS patients and their families; 3. Provide instruments for local AIDS planning, by providing the tools to develop explicit guidelines to health sector decision makers, especially at the district level, to help them select and implement the best among the affordable interventions against HIV infection; and 4. Recommend institutional strenagening and other modifications at the national level to strengthen inter- and intra-sectoral coordination in the fight against AIDS and to support districts in the planning and implementation of local AIDS prevention and coping programs. 1.05 The remainder of this chapter gives a summary of the current health problems of the population in Tanzania and describes how the health sector is organized to deal with them. The second chapter presents what is known about the current state of the AIDS epidemic, including current level and distribution of HIV infection in Tanzania, routes of transmission, 1' NACP, AIDS Surveillance Report (No. 4) of March 1991. -2 - assumptions regarding progression of the epidemic, and estimates of future infection. Chapter three outlines the consequences of the AIDS epidemic, looking at the likely macro-economic impact of the disease over the next 20-25 years, as well as the impact on public sector expenditures. 1.06 Chapters four and five analyze prevention and coping strategies, respectively. Chapter four discusses options for prevention and suggests how to decide among different strategies, in part by weighing their relative costs and effects. Chapter five examines alternative scenarios for treating the opportunistic infections associated with AIDS, as well as strategies for helping survivors. 1.07 Chapter six describes the National AIDS Control Program, and makes a number of recommendations to enable it to put its efforts where they can do the most good, including organizational changes. It also outlines next steps for developing district planning guidelines for coping with AIDS and introduces the elements of a planning framework for decision- makers. Chapter seven summarizes the report's recommendations. B. COUNTRY SEFIlNG 1.08 Tanzania is a country of about 884,000 square kilometers in southern Africa. It's population of about 23.2 million (1988) is growing at 2.8-3.0 percent per annum. Following independence in the mid-1960's, the country embarked upon an ambitious program to provide basic social services to its largely rural population. By the early 1980's, access to health services had improved enormously, with more than 90 percent of the population within 10 kilometers of a health facility. But the economic situation of the country had, at the same time, suffered serious deterioration, so that the quality of those services began to deteriorate sharply for lack of basic inputs. Through structural adjustment measures and substantial donor assistance, the economy is now slowly recovering, but resources available for basic services continue to be constrained. In 1990, about 10 percent of Government's recurrent budget went to the health sector, but this translates into an expenditure of only about $2 per capita. GNP per capita in 1990 is estimated at $90-100, and is currently growing at about 1.5-2 percent per annum. C. HEALTH STATUS AND EALTH SYSTEM 1.09 While the focus of this study is the AIDS epidemic, AIDS is only one of the many health, population and nutrition problems faced by Tanzanians. As alarming as this current epidemic is, it must be viewed in the context of the many other diseases and social problems causing illness and death in the country. Morbidiq 1.10 Table I. 1 provides a summary of the principal causes of morbidity in Tanzania in the mid-1980's, as reported in a sentinel monitoring system established by the Essential Drugs Program.Y Although not recognized as such, it is possible that some of these illness episodes were manifestations of opportunistic illnesses in early AIDS victims. ' World Bank. Tanzania Population. Health and Nutrition Sector Review, Chapter m (Health). Washington, D.C.: World Bank. October 1989. -3 - TABLE 1.1: DISEASE FREQUENCY IN SENTINEL DISPENSARIES (JAN. 1984- AUG. 1997) Peroent of all Diagnoses Dloasae MaE Mm Median Malaria 30.9 21.4 26.1 Upper respiratory infootion 12.9 8.3 10.6 Diarrheal diseases 1 1.6 3.9 7.7 Eye diseases 9.4 3.3 6.3 Intestinal worms 7.8 1.6 4.7 Gonorrhea 7.2 1.7 4.4 Skin disease 6.7 3.5 5.1 Pneumonia 5.5 2.7 4.1 Acoidents 4.3 1.7 3.0 Anemias 3.3 1.5 2.4 Nutritional disorders 2.4 0.4 1.4 Minor oomplications of pregnancy 2.1 0.8 1.5 Ear diseases 2.1 1.0 1.5 Schistosomiasis 1.8 0.4 1.1 Mental disorders 1.6 0.0 0.8 Measles 1.3 0.2 0.7 Serious complications of pregnancy 1.0 0.3 0.6 Whooping cough 0.4 0.0 0.2 Adult tetanus 0.2 0.0 0.1 Acute poliomyelitis 0.2 0.0 0.1 Neonatal tetanus 0.0 0.0 0.0 All other diagnoses 18.2 10.2 14.2 Symptoms & ill-defined conditions 17.7 10.1 13.9 Source: World BanK 1989, p. 44 (citing Hedkvist, F.. EDP Monitoring Study. 1987, March 1988, GOT & UNICEF) 1.11 Malaria, respiratory infections, and diarrheal diseases account for over half of all diagnosed health problems, a pattern which is typical of tropical countries at Tanzania's stage of development due to widespread poverty and general lack of adequate water and sanitation services. The extensive network of health facilities has, however, had an impact, facilitating - for example - Tanzania's Expanded Programme of Immunization. Morbidity due to those infectious diseases for which vaccines have been developed has been cut considerably, though much remains to be done. 1.12 Other frequent health problems in Tanzania include skin diseases, accidents, undernutrition (malnutrition and nutrient-deficiency related disorders, such as anemia and Vitamin A deficiency) and pregnancy-related conditions. Malnutrition is a serious contributor of morbidity: a national, non-random nutritional survey conducted in 1986 and 1987 showed that almost half of children under five suffer from moderate to severe malnutrition, while 80 percent of pregnant women are anemic. Pregnant women also suffer from malnutrition (13%) and iodine deficiency (52%). 1.13 Sexually transmitted diseases (STDs) have been cited by the Ministry of Health (MOH) as being among the top ten causes of morbidity in Tanzania, and were ranked in the top three in some regions even before the AIDS epidemic. Gonorrhea is rated the sixth most frequently reported disease. Skin diseases, ranked seventh, is also a disease category 2Ministry of Health, Tanzania 1983 and Rukkwa Region 1986, as cited in World Bank 1989. -4- frequently used to report STDs, since many SIDs are characterized by rashes and skin ulcerations. About 12 percent of women attending prenatal clinics in Mbeya region in 1990 were found to be infected with syphilis.y In Mwanza region, population-based surveys in 1991 found evidence of recent syphilis infection In 7, 10 and 13 percent of those tested at rural, urban and roadside locations, respectively.1' 1.14 Unfortunately, health data from Tanzania do not provide population-based incidence and prevalence rates of most illnesses 0 However, the current disease burden is substantial. Malaria, alone, is estimated to account for about 1 million episodes of illness a year and measles, for about 250.000 episodes. These numbers far outweigh the estimated number of illness episodes caused by AIDS among adults and children annually, though this picture is likely to change in the years ahead. Mortality 1.15 The crude death rate (adults and children) in Tanzania is about 13 deaths per 1,000 population.2' Tanzania's crude death rate compares favorably with most neighboring countries, with the exception of Kenya, as seen in Table 1.2. The probability of a child dying between birth and his or her fifth birthday is between 13 and 15 percent. Infant mortality is about 104 deaths per 1,000 live births, higher than most neighbors except Malawi and Rwanda. 1.16 Throughout most of the 1980s, malari was thought to be the maor cause of death in Tanzania. In 1984, malaria was responsible for 14 percent of cild deaths and 13 percent of adult deaths. It is among the commonest causes of death in hospitals, accounting for 10 percent of hospital admissions and 5 percent of hospital deaths. Roughly one out of 100 malaria cases ends in death, with malaria accounming for approximately 10,000 deaths annually in the late 1980s. This picture is, however, changing with AIDS. In 1990, AIDS is thought to have overtaken malaria as the major killer of adults. By 1993. AIDS is expected to be the single greatest cause of death in both adults and children, and more than three times as many people may die of AIDS than of malaria. -NACP sentinel surveillance system for period 1/1/90-12/31/90, May 15, 1991. A'AMREF (Tanzania), Mwanza Regional Survey 1991, unpublished. -' Exceptions include anemia (prevalence ranges: 25-69% in surveys); diarrhea (incidence in children under five estimated at three to five episodes per year); and tuberculosis (annual risk of infection estimated at 1.2% in 1987, with incidence of smear-positive pulmonary TB at 60 per 100,000 population). 21 1988 estimate from the World DeveloDnment Report 1990 (13 per 1,000). UNICEF's State of the World's Children 1990 gives a figure of 14 per 1,000. -5 - TABLE 1.2 COMPARATIVE MORTALITY INDICATORS, TANZANIA AND NEIGHBORING COUNTRIES COUNTRY CRUDE DEATH UNDER FIVES INFANT MATERNAL RATE (1) MORTALITY MORTALITY MORTALITY RATE (2) RATE (3) RATE (4) Malawi 20 234-248 149 250 Rwanda 18 184-205 120 210 Uganda 17 147-167 101 300 Burundi 16 98-113 73 N/A Tanzania 13 135-153 104 370 (5) Zambia 13 108-124 78 110 Kenya 11 95-ll0 170 510 (5) SOURCE: World Development Report 1990, World Bank. Estimates are from 1988 in all cases except maternal mortality, where figures are from 1980. (1) Annual deaths per 1,000 population. (2) Annual deaths of children under 5, per 1,000 live births. Range shows males versus females (females first). (3) Annual deaths of infants under 1, per 1,000 live births. (4) Annual deaths of women from pregnancy related causes, per 100,000 live births. (5) Hospital based statistics, only. 1.17 As with morbidity, acute respiratory infections and diarrheal diseases are also high- ranking causes of death in Tanzania. This pattern is typical of other Sub-Saharan countries of similar income levels. Table 1.3 provides additional WHO data on morbidity and mortality from diarrheal diseases and acute respiratory infections (ARI) in Sub-Saharan Africa. 1.18 More effort is needed to analyze already collected data from Tanzania concerning death statistics. This will help policy makers draw conclusions about the relative importance of the AIDS epidemic within regions and localities, compared to other diseases. -6- TABLE 1.3 MORBIDfTY AND MORTALITY STATISTICS, DIARRHEAL DISEASES AND ARI DATA FROM AN ANALYSIS OF STUDIES IN SUB-SAHARAN AFRICA (Feachem and Jamison, eds. 1989) DISEASE ADULT/ INCIDENCE PROPOR- ANNUAL CASE PROPOR- CHILD (CASES PER TIONAL DEATHS FATALITY TIONAL PERSON MORBIDITY PER 1,000 RATE MORTALITY PER YEAR) RATE RATE DLAR- CHILDREN 4.9 - 5.1 25% 4.9 0.1 - 14.2 45% RHEAL DISEASES ADULTS 3-5 times 20% N/A N/A 3.8 - 21% lower than children l ACUTE CHILDREN 7.6 - 13.2 25.1% - 66% 6.2 17.9 7.6 - 22% RESPIRA- TORY INFECT- ADULTS N/A N/A N/A 2-4% 6-10% in all IONS age groups, adults and children NOTES: Incidence: Diarrhea, 4.9 = median of 73 estimates; 5.1 = Tanzania nationwide study May 1986 ARI, 7.6 = Ethiopia 1977; 13.2 = Burkina Faso 1986 (also 9.2 = Kenya 1984) Proportional Morbidity: Percentage of total cases of illness attributable to this disease. Diarrhea in children, median of 5 estimates; diarrhea in adults, est. from Imo, Nigeria 1987. ARI in children, 25.1 = Sierra Leone 1981; 66% = Burkina Faso 1986 (also 36.3% = Ethiopia 77) Annual Deaths: Expressed per 1,000 children or adults. Diarrhea, est. from Tanzania 1986, longitudinal study based on yearly recall. ARI, median of six studies. Case Patality: Diarrhea, 0.1 = Kenya 1984 community based study; 14.2 = median of 3 hospital studies. ARI in children, median of five studies; in adults, median of three studies. Proportional Mortality: Diarrhea in children, median of 40 estimates (WHO/CCD); in adults, 3.8% = Uganda 1986, 21% = Malumfashi, Nigeria, 1986. ARI in children, 7.6 = median of 7 estimates from longitudinal studies, 22% = median of five hospital studies. In adults, 6-10% is mode from nine estimates. Health System and Policies 1.19 Health policy in Tanzania emphasizes primary health care, especially targeted to rural areas and preventive services.Y In the 1970s and early 1980s, strategies for extending coverage focused on the expansion of rural health facilities. A great number of rural health centers and dispensaries were constructed during that period, and coverage improved substantially. In addition, many paraprofessional and auxiliary health staff were trained and placed in these peripheral facilities. A program promoting the use of village health workers F For a detailed review of the health sector in Tanzania, see World Bank 1989, op. ck., pages 48-65. in primary care was initiated. In more recent years, plans to extend coverage have had to be modified due to deteriorating economic conditions and resulting concerns about the financial sustainability of health services. 1.20 The health care referral system in Tanzania now includes over 3,300 facilities, from village health posts, dispensaries and rural health centers to district, regional, consultant and specialty hospitals. Table I.4 lists health facilities by region, distinguishing between those owned by the Government and those owned by private NGOs. TABLE 1.4: TOTAL HEALTH FACILITIES BY REGION AND AJAGENENT. AINLAND TANZAMIA 1988 Region Hospitals Hlth Ctrs Dispens. _osp Beds Clinic Beds Govt NGOs Govt mOos Govt MG0s Govt NGOs Govt |GOs Tanga 5 7 15 ... '36 67 945 626 364 - -- Coast 4 2 11 -... 90 44 394 196 180 --- LMorogoro _ 4 7 15 1 131 52 621 948 286 Lindi 4 3 12 ... 88 12 515 294 171 Iringa 5 8 16 . 91 50 629 975 217 Ruvuma 2 5 13 .. 93 36 347 933 355 --- Kiliman' 5 8 13 4 93 42 730 1170 393 ---__ Arusha 6 6 11 --- 115 65 683 622 230 . Dodoma 5 1 16 1 148 24 1461 250 328 Mara 3 4 11 --- 92 31 406 386 230 Rukwa 2 1 11 1 73 15 354 45 310 30 Singida 2 4 12 --- 75 46 344 578 177 Tabora 4 3 10 1 84 22 618 421 289 ... Kigom r 3 2 10 94 26 387 172 195 ... Mbeya 4 6 17 149 37 758 712 348 Mtwara 3 2 13 --- 99 9 743 500 214 . Mwanza 4 7 26 204 34 480 1724 667 --- Kagera 1 10 12 --- 132 15 250 1459 365 ... Shinyang 4 2 19 137 30 789 350 480 . D.S.M. 4 8 3 3 81 73 1982 300 50 52 [ Totals: 74 96 266 11 2205 730 13436 1 12661 5849 82 | SOURCE: World Bank 1989, p. 53, citing MON, February 1989. - 8 - 1.21 Although the private practice of medicine was officially banned in 1980, private, non- profit NGOs were excluded from this ban and many of their facilities are formally included in the health care system, with the Government providing some supervisory and substantial monetary support. The health care system nationally operates over 32,000 inpatient beds, and is staffed by more than 18,000 professional and other medical personnel. In addition, traditional healers and birth attendants are allowed to practice. The Ministry of Health has included traditional birth attendants in its training programs, including those organized by the NACP. 1.22 The Ministry of Health is responsible for :ormulating health policies and strategies, for health planning and for running centralized support services, including consultant and specialty hospitals. It currently comprises five departments responsible, respectively, for hospital services, preventive services, training, health planning, and administration. The NACP is one of 17 special programs established within the Department of Preventive Services to deal with specific health problems or challenges. Other important (and older) such special programs include the Maternal and Child Health Program, the Expanded Program of Immunization, the Tuberculosis and Leprosy Control Program, and the Diarrhea Control Program. Regional and district authorities are responsible for operating regional and district- level facilities, with all health facilities and staff at the local level - from the district hospital on down - coming under the jurisdiction of the district health team, headed by the District Medical Officer. 1.23 Theoretically, the District Medical Officer is responsible for establishing priorities among health interventions at the local level and for allocating staff and other resources in accordance with those priorities. Despite the Government's stated policy of decentralization, this has worked imperfectly due both to the separate funding and supervision of special programs operated from MOH headquarters, and to the control which the District Executive Director exerts over allocations for health which flow to the district through the budget of the Ministry of Local Government. 1.24 For public facilities at the district level and below, full decentralization of management responsibility is envisaged. In order to strengthen the districts' capacities to assume more such responsibility, Tanzania has launched an IDA-supported project, with urban and rural primary i.ealth care components administered by the Ministry of Local Government. Operating in ten pilot districts, the project provides assistance and funding which will allow development expenditures for health to be allocated in accordance with plans drawn up and implemented by the districts themselves. The process and the outcome will be carefully monitored and evaluated so that lessons learned might be applied more widely to strengthen district planning on a national basis. 1.25 Health expenditures are financed from budget allocations to the Ministry of Health and the Ministry of Local Government, from multilateral and bilateral donors, from local tax revenue, from NGO funds and from health care consumers. All health care is officially free at Government health facilities, although some small charges have been levied for higher quality rooms in hospitals. The Government is considering introducing some user fee charges to help defray rising costs. Most NGOs charge at least a nominal fee. Although data are not yet available from ongoing household expenditure surveys, private allocations for health care are thought to be significant. - 9 - II. THE AIDS EPIDEMIC 2.01 This chapter discusses what is currently known about AIDS in Tanzania: how many people are infected, which population groups and regions are affected most, how the HIV virus is transmitted, how quickly people who are infected will develop AIDS, and how long people survive with AIDS. 2.02 In addition to reviewing the current state of the epidemic, the chapter looks forward to estimate how the epidemic will change over time. While much less is known about the future of the epidemic, it is here that policies must be focused in order to reduce the major economic and social impacts of the disease. A. CURRENT SITUATION 1. Data and Methods 2.03 The report cites several sources in estimating the level and distribution of HIV infection and AIDS cases at present and in the future. NACP estimates are given, and these are supplemented by different models adapted by Bulatao (1990) and Chin and Sonnenberg (1991). Some aspects of Bulatao's modeling are described briefly here to provide a better understanding of how the estimates were derived. Differences between this model and the others are also highlighted. 2.04 Bulatao's model was used to simulate two alternative scenarios. The model divides the population into risk groups and specifies sexual and other risk behaviors (e.g. rate of sexual partner change, frequency of blood transfusions) for each risk group. Wherever possible, the model uses Tanzania-specific assumptions, although some "Pan-African" assumptions are used where Tanzanian data were not available. For example, transfusions per person per year is estimated to be 0.01, using data from a study by Ryder, et. al. (1989) in the Mbeya Region. Condom use is estimated at 1 percent, based on local data. 2.05 The two Bulatao model scenarios vary only in one key assumption: the percentage of the population who have only one sexual partner (i.e., "monogamous" for short). This assumption was varied for two reasons: first, the model is very sensitive to this assumption, and a change in the value of the monogamy variable has a great impact on the number of cases of HIV infection predicted; and second, this variable is one of the least well known parameters used in the model. The uncertainty about the true value of the monogamy parameter may be lessened as data from survey research on knowledge, attitudes, beliefs and practices (KABP) becomes available, but for now it is prudent to model more than one assumption. Thus, the two model scenarios from Bulatao are named "15% Monogamy" and "45% Monogamy," indicating the low and high estimates of the percentage of married couples in which both partners are monogamous. "Pan-African" assumptions were used in estimating the percentage of the female, single population who are prostitutes (4.3%) and the annual number of new partners per year for female prostitutes (143) used in the 15% monogamy scenario. l/ ltIn his paper, Bulatao refers to the "15% Monogamy" scenario as the "Modified Standard Scenario," for reasons explained in his text. In that scenario, 26% of all married men and 50% of all married women are monogamous, with both partners in a union being faithful in - 10- 2.06 The Chin and Sonnenberg projections are based on a WHO model which uses point- prevalence estimates of HIV infection and progression rates from infection to AIDS to calculate the number of past AIDS cases and make short-term estimates of future cases. The WHO model does not incorporate biologic and behavioral variables describing HIV transmission and natural history, as does the Bulatao model. Further information about the WHO model may be found in Annex 1 of the Chin and Sonnenberg paper. 2.07 The NACP estimates are based on seroprevalence data from blood screening programs (i.e., the proportion of screened blood which contains HIV antibodies, indicating that the donor has been infected). Data from each of the regions has been adjusted for age, sex and degree of reporting to give overall estimates of disease prevalence in the population. These have then been used to project future AIDS cases on the basis of estimated progressions rates. 2.08 While some estimation of the magnitude and spread of the AIDS epidemic must be made to assist policy makers and planners, projections need to be interpreted cautiously with the understanding that many uncertainties still exist, and that gaps in our knowledge of the natural history of the epidemic and of HIV prevalence in Tanzania are wide.V 2. Number of Cases, Infections and Deaths AIDS Case 2.09 The first suspected case of AIDS was diagnosed in Tanzania's Kagera Region in 1983. However, extensive spread of the human immunodeficiency virus (HIV) probably began in the late 1970s and early 1980s. Experts now understand that a person can be infected with HIV for up to 20 years before he or she develops AIDS, although the median time of infection before AIDS manifests itself appears to be 8-10 years in an adult. During this time, the HIV infected person is without symptoms, and appears to be in good health, although his or her immune system is becoming progressively weaker. Gradually, the person becomes more susceptible to all kinds of illnesses; i.e., 'opportunistic' infections (viral, bacterial or fungal) which may be quite common but would normally be successfully destroyed by the immune system. The presence of AIDS may not be suspected until these illnesses become especially frequent or fail to respond to treatment. To help health workers recognize AIDS, only 15% of couples, while these percentages increase to 39 and 58, respectively in the "45% Monogamy' scenario. Details can be found in Bulatao's paper. 2' F. Sonnenburg, personal correspondence to J. Murphy, World Bank, January 23, 1991. See also Anderson (1988) for an enumeration of many of the uncertainties which complicate scientific assessment of the impact of AIDS in LDCs. Y Chin and Sonnenberg 1991, citing Mhalu 1987. - 1t - criteria have now been established which define the onset of AIDS, especially in those circumstances when confirmatory blood tests cannot be carried out.*' 2.10 Following recognition of the first AIDS case in 1983, reported cases escalated rapidly. As of end 1990, a cumulative total of 21,175 AIDS cases had been reported throughout the country.-' The number of officially reported cases is thought to understate considerably the true number, given the rapid increase in reported cases as knowledge of the disease has improved; the current HIV sero-prevalence and incidence rates which suggest an older epidemic than reported AIDS cases would indicate; and current knowledge of the disease's transmission and progression rates. Such underreporting would, in fact, be expected, due to the inadequate access to health facilities by the population; the lack of resources for management of the case reporting system; and, despite NACP's best efforts, the continued insufficiencies in laboratory facilities, supplies, and training of clinical staff. The NACP estimates that the true number of AIDS cases from the start of the epidemic through 1990 is more than 100,000 (or about 5 times the number of reported cases)AY Chin and Sonnenberg's model, based on data through 1989, projects a total of about 64,700 cases through 1990,!' while Bulatao's projections through 1990 range between 46,300 and 224,200 cases, depending upon the scenario selectedY 2.11 AIDS is a disease that affects mainly the sexually active members of the population, and infants. About 94 percent of all reported AIDS cases have been adults between the ages of 15 and 55 years, and 4 percent have been children under five.Y Because of the difficulty in recognizing the disease in infants, however, pediatric AIDS is thought to be even more seriously underreported than adult AIDS. Children under 5 probably represent about one- third of AIDS cases to date. This is considerably higher than one might assume given the much lower transmission rates in children (see Section B, below), but is likely because the disease progresses more quicldy in infants than it does in adults. Figure 11.1 is a graphic representation of estimated cumulative cases - paediatric and adult - through 1990. - In Tanzania, an AIDS case is diagnosed using "modified Bangui criteria." The Bangui criteria are Africa-specific clinical AIDS indicators (usually the presence of infections such as oral thrush, pneumonia, and other diseases which frequently strike AIDS patients) developed during a workshop held by WHO in Bangui, Central African Republic, in 1985. The Bangui criteria require at least two major and one minor symptom, in order to allow an AIDS diagnosis. In Tanzania, these criteria have been tightened so that two major and two minor symptoms are required. In reality, only 35% of newly reported cases in the first semester of 1990 fulfilled the specified Tanzanian criteria, presumably due to inaccuracies in record keeping. (NACP 1990) E NACP (1991), gO. i. Y Ibd. Z1 Tables 2 and 3, Chin and Sonnenberg, 1991. IV Bulatao, 1990, Appendix, p. 15 and 24 (15% and 45% Monogamy scenarios). 2' NACP (1991). . 12 - FIGURE lA.l - ptimated Cumulative AIDS Cases. Adults and Children Number (Thousands) 140 - 120 - 100 80 60 40- 0 -i~ 78 79 80 81 82 83 84 85 86 87 88 89 90 Year El] Paediatric AIDS ELI Adult AIDS TOTAL Sre: Chin and Sonnenberg, 1991, Figure 4. Deaths from AIDS 2.12 Reported deaths from AIDS are not recorded in NACP reports, but given an average life expectancy of one year after diagnosis,E NACP estimates that the disease had been responsible for about 60,000 deaths by the end of 1990, about 26,000 of them in that year alone.W Models by Bulatao estimate roughly 33,000 to 165,000 cumulative AIDS deaths by 1990. 2.13 AIDS may very well have become the number one cause of deaths among adults in Tanzania by 1990.)1 In the absence of AIDS, approximately 91,000 adult deaths would have been expected in Tanzania in 1990 (given an adult mortality rate of about 7/1000 in the early 1980s). However, adlt deaths due to AIDS probably ranged between 8,000 and 35,000 Z Chin and Sonnenberg cite evidence from Dar es Salaam that the average life expectancy after diagnosis might be as short as three months (Pallangyo, et. al. 1989). 11NACP, 1991. W Chin and Sonnenberg 1991, p. 6. - 13 - in 1990 (if Bulatao's two scenarios represent the extremes), increasing the adult mortality by between 9 and 38 percent.23' HIV Infetion 2.14 HIV-1 infection has now been detected in all regions of the country, although prevalence varies enormously. The presence of HIV-2 (a less pathogenic West African variant of HIV) is thought to be very small.Y 2.15 On the basis of information available to it, NACP estimates that, in addition to those who currently have AIDS, about 800,000 people are infected with HIV (or about 3 percent of the total population). The estimates from the Chin-Sonnenberg model predict that about 500,000 people are infected, while Bulatao's estimates range from about 400,000 to almost 1.4 million currently infected, but not ill.11' Of these, between two and three percent are estimated to be children under age 14. 2.16 Using WHO data for Sub-Saharan African countries, Tanzania is ranked sixth in terms of total estimated number of HIV infected persons, after Uganda, Zaire, Zambia, Cote d'Ivoire and Malawi. Prevalence rates in Tanzania fall roughly in the range of those in Zambia, Zimbabwe, Malawi, Uganda and Rwanda. These rates are much higher than those found in North Africa and most Sahelian countries. Thus far, Central and East Africa seem to be hardest hit by the AIDS epidemic. 2.17 Estimates of the prevalence of HIV infection in the population are based on tests which have been carried out on blood drawn from a number of different groups. Blood is screened regularly for HIV antibodies at 168 blood transfusion centers around the country, while blood screening for surveillance purposes is done at about a dozen different prenatal clinics. In addition, tests have been carried out on a number of specific groups thought to be at high risk for acquiring the disease. Seroprevalence among blood donors, by region, are shown in Tables I.1 and 11.2, below. 13/ The true baseline adult mortality rate in Tanzania is unknown, but it is expected to be somewhat higher than the estimated adult mortality for Sub-Saharan Africa as a whole (between 5.6 and 6.5 deaths per 1,000 adults over age 15, Bulatao and Stephens 1989). Expected deaths in the absence of AIDS are calculated by applying the estimated baseline adult death rate in Tanzania (7 deaths per 1,000 adults) to the projected adult population in 1990 (about 13 million). Chin and Sonnenberg 1991, p. 7. AIDS deaths in 1990 are obtained from Bulatao 1990, Appendix, top right tables on pages 15 and 24. LI Chin and Sonnenberg 1991, p. 1. M' It must be stressed that these are estimates based on projections from epidemiologic- demographic models, each constructed in slightly different ways. The models have been calibrated with data on HIV prevalence from studies conducted in the Kagera, Mbeya, Mwanza, Kilimanjaro and Dar es Salaam Regions, as well as observed annual progression rates from HIV infection to the development of AIDS. Where prevalence data from other regions were available, these were incorporated as well. Models must be continually recalibrated based on updated information from surveillance systems and survey research. - 14 - TABLE I. 1 - Seroreyalene in Blood Donors. by Region. 1987 - 1989* (in percent) REGION J 1987_ 1988 ) 1989 Arusha -- 0 Coast 2.22 5.43 3.81 DSM 4.65 7.11 7.46 Dodoma -. = 1.56 Iringa - 20 Kagera _ - 13.14 Kigoma - 0.84 Kili'jaro _ - 3.18 Lindi _ 2.84 Mam - 6.01 Mbeya 4.36 4.68 4.68 Morogoro - 10.37 6.66 Mtwara - 5.04 Mwanza 5.06 7.55 Rukwa - 14.29 Ruvuma - 13.40 Singida - 4.35 Shinyanga 15 Tabora - 2.46 Tanga - 6.67 TANZANIA 4.55 7.23 6.77 Adjusted prevalence**: | 5.22 SOURCE: NACP 1990, Table Sd * While the NACP states that aU regions and most hospitals report seroprevalence of blood donors (NACP 1990), this does not mean that all blood donations are being screened. In some regions, the number of screened donations over five years does not exceed 100 (see Table 11.2). ** The source document does not state explicitly how prevalence has been adjusted, but adjustment is probably for sample size, as is the case in Table 11.2. Other tables in the source document (NACP 1990) make adjustments in prevalence for sex and age. Less than 6% of aU donors are female, and blood donors are younger than the general population. The net effect of these adjustments is to reduce prevalence to 5.22% in 1989. - 15 - TABLB 11.2 - Cumulative Sero ivalence in Blood Donors by Region. 1990 Region Adult Total Positive Prev. % Rank Expected HIV+ Population Tested _ __ Arusha 716,388 177 0 0.00% 20 0 Coast 338,148 1,063 44 4.14% 14 13,997 DSM 656,044 28,533 2,031 7.12% 8 46,698 Dodoma 721,251 128 2 1.56% 18 11,270 lringa 640,724 90 18 20.00% 1 128,145 Kagem 702,877 137 18 13.14% 5 92,349 Kigoma 453,053 356 3 0.84% 19 3,818 Kili'jaro 587,610 273 7 2.56% 16 15,067 Lindi 342,672 176 5 2.84% 15 9,735 Mara 514,599 2,238 111 4.96% 12 25,523 Mbeya 782,385 3,593 221 6.15% 10 48,123 Morogoro 648,051 3,578 297 8.30% 6 53,793 Mtwara 471,432 119 6 5.04% 11 23,770 Mwanza 995,484 1,620 114 7.04% 9 70,053 Rukwa 368,336 98 14 14.29% 3 52,619 Ruvuma 415,163 194 26 13.40% 4 55,640 Shinyanga 939,451 60 9 15.00% 2 140,918 Singida 419,661 115 5 4.35% 13 18,246 Tabora 549,235 928 22 2.37% 17 13,021 Tanga 680,327 4,668 341 7.31% 7 49,698 TANZANIA 11,942,891 | 48,144 3,294 6.84% _ 872,481 i~ - . .. - . , . Prevalence adjusted for sample size: 7.31% SOURCE: NACP 1990, Table Sc * Based on 1988 population - 16 - 2.18 One distinctive feature of the AIDS epidemic is that HIV infection is not randomly distributed in any human population.W The next section will discuss evidence of uneven distribution patterns in Tanzania, according to region, gender, age, socio-economic status, and special risk groups. 3. Distribution by Region. Gender. Ae and Socio-economic Status Geographic Distribution 2.19 Figure II.2 presents WHO's estimate of HIV infection among adults by region in 1989.12 Compared to the national prevalence estimates of 1.4-5.3 percent (adults and children), urban HIV prevalence rates of 7-10 percent and higher are found in several areas. The region estimated to have the highest prevalence rate is the Kagera Region. Located in the northwest corner of Tanzania, Kagera borders on regions of Uganda (the Rakai District), Rwanda, and Burundi, which together comprise the area of the world with the largest concentration of HIV infected individuals. In Bukoba, the capital city of the Kagera Region, the prevalence rate among adults 25 to 34 years old in 1987 was approximately 31 percent."' Compared to the estimated national adult prevalence of about 7 percent,1' this is extremely high. 2.20 The next most affected regions are Mwanza, adjacent to Kagera by the lake, Mbeya, on the main transportation route from Dar es Salaam to Zambia, and the city of Dar es Salaam. Other regions are currently estimated to have smaller, but still significant levels of infections. 2.21 In Tanzania, as elsewhere in Africa, studies have consistently found markedly higher HIV prevalence among urban populations, as compared to rural areas.2' Research conducted in Africa has found generally that rural seroprevalence is about one tenth of urban seroprevalence, excepting rural areas with heavily traveled routes passing through.2' This provides evidence in support of targeting AIDS interventions to urban areas, at least in the near future. If, as more recent studies suggest, the spread of HIV infection in rural areas is time-delayed, we can expect to see rural seroprevalence rates climb over time, which will create a greater need fbr AIDS interventions in rural areas.V Chin, J. "The Epidemiology and Projected Mortality of AIDS," Chapter 13.1 in Feachem and Jamison (eds.), Disease and Mortality in Sub-Saharan Africa. Vol. I. Washington, D.C.: World Bank. Nov. 1989. Ul' Zanzibar has been excluded, due to lack of data. IF Killewo, et. al. (1990) e NACP 1991. V Chin and Sonnenberg 1991, p. 4; Killewo 1990. 1' Chin 1989, op. cite., p. 256. W Chin and Sonnenberg 1991, p. 4. - 17 - FIGURE I1.2 - Esimated HIV Prevalece Among Sexuall Active Tanzania. 18 . . 9 a a 0 f 1V:" ":0 " wid 1.3% I*lala~~~~~~~~~~W Mozabtu n#rd '1 % nia1 c nF fUid < ts1 - Sore Chin and Sonnenberg, 1991, Figure 1. - 18 - Gender and Age Distribution 2.22 As in other African countries the numbers of male and female infected individuals are approximately equal. Age distribution by gender differs markedly, however: while both groups are stricken during their prime productive years, the male cases are generally older than the female cases. This is illustrated in Figure 11.3, which shows a pyramid diagram of reported AIDS cases in one region of Tanzania by age and sex. Again, the same difference in prevalence rates by age for males versus females has been documented in many other African countries. FIGURE 11.3 - e Distributign of Regqrte AIDS Cases. Mbeya Region ' 6 6-9 1a-t294 with known age lot total 1,661 10-14 Cumulative tOputs up to Sep. 16. 1990 16-19 ' 20-24 26-29 30-34 36-39 40-44 46-49 60-64 66-69 '60 250 200 160 100 60 0 60 100 150 200 t female inmale Source: Chin and Sonnenberg, 1991, Annex 4, Figure 8. 2.23 The age group most affected is sexually active adults in the 25-34 age group; however, an alarming increase in seroprevalence has been noted among blood donors in the 15-19 and 20-24 age groups since 1987.A Overall, adults in the 15-44 age group account for about 87 percent of reported AIDS cases in Tanzania to date; children 0-4 years old ' Chin (1989) provides an interesting discussion of why this pattern differs from that observed in Western Industrialized countries, where males are a much higher proportion of AIDS cases. His answer highlights the role of co-factors including high prevalence of infectious and parasitic diseases, and STDs. ' Chin 1989, p. 256. 9 NACP 1990, p. 9. 19 - account for about 4 percent of cases. Very few infections have been found in 5-14 year olds, in Tanzania or elsewhere in Africa.2y Distribution by Sociop-lonomic Status 2.24 The distribution of HIV infection by socioeconomic status in part reflects differences in geographic distribution and distribution by age. That is, adults who live in cities and along major transportation routes have on average higher incomes than adults living in more isolated, rural areas. Furthermore, middle-aged men tend to have higher incomes than do younger men. These facts combined with the fact that a high socio-economic status cannot protect against HIV infection (as it can against many other infectious diseases whose spread is facilitated by poor hygiene and sanitation, lack of potable water, etc.) already imply that the distribution of AIDS infection will not resemble that of other diseases, which show a heavier impact on lower income brackets. 2.25 If the prevalence of HIV infection were just as high among better educated, higher status Tanzanians as it was among lower status ones, this would already distinguish AIDS from most other infectious diseases. In addition, though, limited evidence from neighboring countries suggests that higher socio-economic groups have even higher prevalence rates than do lower ones.2l' For example in Rwanda urban adults with more than primary education had a 50 percent higher HIV prevalence rate than those with primary or less. In Zaire, the executives of an urban textile factory were twice as likely to be infected with HIV than factory workers, with foremen scoring halfway in between. Finally, among Zambian hospital patients, personnel and blood donors, those with five to nine years of education were twice as likely to be infected as those with four or fewer years, and those with ten or more years of education had a seroprevalence rate three times higher. 2.26 Although Tanzanian seroprevalence data have not been analyzed for correlation with socioeconomic status, researchers expect a similar pattern to hold true. The explanation of this correlation is unclear: some researchers suggest that the pattern is explained by higher income populations having greater access to foreign travel or higher rates of sexual partner change. Prevalence Estimates by Special Risk Groups 2.27 Since the HIV virus is transmitted sexually and through blood and blood products, certain segments of the population are at higher risk for HIV infection than others. These groups include sexually active adults - especially those who change partners frequently or who have other STD infectons - and their spouses/partners and unborn children; recipients of blood transfusions and health care workers also face some risk, the latter because they could become infected through the handling of blood products or contaminated needles. For these 2' Chin 1990, Chin 1989. ;' Over and Piot, 1990, Table 1O.X, p. lOa. Hunt (1989) cites other studies (Georges et. al, 1989; Harden, 1987 and Van de Perre, 1984) which have reported that AIDS patients in eastern Africa are more well-to-do than the average population, although this might reflect ease of access to the medical system rather than true differences in prevalence rates among the more affluent vs. poorer classes (p. 361). -20 - reasons, seroprevalence studies to track the spread of HIV infection often target prostitutes, STD patients, pregnant women, blood donors and health care workers. Results of prevalence studies among these special risk groups in Tanzania are presented in Table 11.3. TABLE 1.3 - Resent Estimates of HV SeroDrevalence Among Selected Risk Groups I Risk Group Prevalence Place Date/Source of Study Estimate STD Patients 13X Mwanza Dolmans 1989 Pregnant Women 10X Mbeya-urban 1989, NACP ' 16X Mbeya-urban 1990. NACP 3X Mbeya-rural 1989. NACP 7X Nbeya-rural 1990I NACP __ ___ 11X Morogoro City Christiansen 1989 BTlood Donors (see aso 5X Mwanza Dolumans 1989 Table 11.2) _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ Bar Maids 29X Dar es Salaam 1986, cited in Smith 1988 Ibid. OX Arushe Ibid. 137X Bukoba _ Health Clinic Staff 18X Morogoro City Christiansen 1989 B. ROUES OF TRANSMISSON AND ASSU1MPONS REGARDING PROGRESSION 1. Major Transmission Routes 2.28 In Tanzania, as in most of Africa, there are four principal routes of transmission of the HIV virus: heterosexual, perinatal (also called mother to child), blood transfusions, and contaminated health care equipment. Of these, by far the most important is heterosexual intercourse, which is directly responsible for between 75-80 percent of all AIDS infections. It is probably indirectly responsible for most perinatal transmission, accounting for another 10 percent of infections, since most mothers were probably infected through sexual contact. Blood transfusions are also estimated to be responsible for about 10 percent of infections to ' See Appendix B for a complete listing of prevalence studies conducted to date, along with results. Dates of studies may refer to date of publication of study results, and not the period for which data were collected. Studies in table were originally cited in Bjorkman 1991, unless otherwise stated. a' Mbeya survey data is cited in NACP 1990. Text and tables vary for the 1989 estimate in Mbeya-urban. While the text states prevalence rate was 10% in 1989 and 16% in 1990 (p. 9), Figure 2a on page 20 shows rates of about 9.5% and slightly under 15%, respectively. -21 - date. Contaminated health care equipment is thought to account for a relatively small number of infections.-' Hetesexual 2.29 The probability of heterosexual transmission is not known precisely, but on the basis of research to date is estimated to be .03 (3:100) during any one contact between an infected male and an uninfected female, and .01 (1:100) during any one contact between an infected female and an uninfected male.A' 2.30 In addition, several studies have indicated that an existing ulcerating classical sexually transmitted disease (e.g. syphilis or chancroid) in either partner can multiply the probability of transmitting the HIV virus by three to ten times.;2 2.31 Obviously, the more frequent the number of sexual contacts between a non-infected person and an HIV-infected partner, the more likely it is that the non-infected person will seroconvert. Due to the many differences in sexual behaviors and prevalence rates among different population groups in Tanzania, models which allow variation in behavioral parameters can potentially provide more accurate estimates than models which project future prevalence rates based on historical rates and progression patterns alone. A great deal more behavioral research is needed, however, to determine the values for assumptions such as percentage of adults who are monogamous, frequency of partner change among non- monogamous males and females, and use of condoms. Perinatal 2.32 HIV infection is present in large numbers of sexually active, otherwise healthy women of child-bearing age. When these women give birth, their children may contract HIV infection in the womb, during or immediately after birth. An estimated 10 percent of HIV infections in Tanzania may be the result of perinatal transmission. This accounts for the second most probable route of transmission, followed closely by blood transfusions. LI Bjorkman 1991, p. 5, citing Ngaly. AL' Over and Piot 1990, Table 10.11, p. 17a. Elsewhere Over and Piot state that "the basic risks of acquiring HIV infection through vaginal intercourse is 0.1 % to 0.5%, and through receptive anal intercourse is probably 10 times higher' (p. 7, citing Piot, et. al. 1987 and Johnson and Laga 1988). These probabilities assume one partner is infected and the other is not. Chin and Sonnenberg estimate that the probability may be between 1:10 to 1:100 per single sexual encounter between an HIV-infected and non-infected person (p. 2). V Bjorkman 1991; Over and Piot 1990, citing Piot et al 1988 and Pepin et al 1989; Chin 1989, citing Simonsen et al 1988 and Hemed et al 1989. Over and Piot note that presence of other STDs increases both susceptibility to HIV infection (of the non-infected partner) and infectiousness of the HIV-infected partner. (p. 7) V Ngaly 1990 and Chin and Sonnenberg 1991, cited in Bjorkman 1991. - 22 - 2.33 The probability of perinatal transmission is difficult to estimate because all infants born of HIV-infected mothers, even uninfected infants, will carry their mother's antibodies against the HIV virus for six to twelve months after birth. Since the only simple tests for HIV infection are designed to detect antibodies, performing tests on babies of infected mothers will yield uniformly positive results for infected and uninfected babies alike. As an uninfected baby ages he or she will lose the antibodies, and will no longer test positive for HIV infection. But many children die of conmmon childhood diseases such as malaria and measles before one can be sure whether they are truly infected with HIV or not. The infant mortality rate (IMR) in Tanzania is currently estimated to be between 105 and 118 deaths per 1,000 live births, and in 1983 (prior to the epidemic) the rate was at a similar or perhaps even higher level.-! 2.34 Taking into account these difficulties, probabilities of perinatal transmission range from 12 to 50 percent, with the higher transmission rates being observed in women who are at more advanced stages of HIV infection.; The mean probability may be 30-40 percent.;' Blood Transfusion 2.35 Blood transfusions are estimated to account for only about 10 percent of HIV infections in Tanzania,3" even though transfusion of HIV-infected blood to an uninfected person is known to be a highly efficient method of transmitting the infection, with a transmission probability of almost 100 percent for a single transfusion.3' 2.36 Unlike child bearing and sexual intercourse, though, blood transfusion is a relatively rare event for most Tanzanians. A survey in the Mbeya region showed that about 12,000 blood transfusions per year were performed in a population of 1.5 million, or 8 transfusions per 1,000 population Oess than 1 %).-9' The NACP estimates that the likelihood of blood 3' High end estimate is used by Bulatao, 1990 (Appendix, p. 19). UNICEF estimated a lower IMR of 105 for 1988 (State of the World's Children, 1990), compared to 115 in 1983 (State of the World's Children, 1986). V' Chin and Sonnenberg estimate that the probability of perinatal transmission ranges from 12- 39%, while Over and Piot note a range of 25-50%. Both these sources cite Ryder 1989, while Over and Piot also cite Lallemont et al, 1989. ;V Chin and Sonnenberg use a mean probability of 30% in their modeling, while Bulatao uses 40%. fl' Chin and Sonnenberg 1991, p. 2. It is not clear from the authors' paper if blQod transfusions account for 10% of new infections each year, or whether this figure is cumulative to date. Presumably blood transfusions will account for a declining proportion of total infections as coverage of blood screening programs increases. ' Bulatao 1990. t' Chin and Sonnenberg 1991, citing Ryder, et. al. 1989. - 23 - transfusion is 6 per 1,000 population per year nationwide.-' Once hospitalized, the likelihood of transfusion may be high. In Zaire, blood transfusion was an extremely frequent occurrence in 1986: a cross-sectional survey of 167 hospitalized children (under age 12) showed that 47 percent had received blood transfusions during the current hospitalizationA. A survey of hospital workers in Kigali, Rwanda in 1985 revealed that 16 of 150 (11%) had a history of blood transfusion in the previous five years.AZ 2.37 The prevalence of HIV infection in blood donors in Tanzania was low before 1987 (< 1-5%), and since 1987 blood screening has further reduced the likelihood of transfusing infected blood (the NACP estimates that about 80 percent of the nation's blood supply is currently being effectively screened prior to transfusion). It is estimated that blood screening is 90 percent effective in detecting HIV-infected blood.43' Tanzania's screening program will become more important for averting HIV transmission as the prevalence of HIV infection in blood donors increases. 2.38 The likelihood of HIV transmission through blood transfusion is higher among children under age five and women in child-bearing years, compared to other age groups. The Mbeya data showed that 40 percent of all transfusions were performed on children, most of which were to treat anemia caused by malaria. Similar statistics were found in Zaire, where 44 percent of transfusions at one major teaching hospital were performed on children under age five, three-quarters of which were for malaria treatment. Data from other countries show that pregnant or parturient women are also likely to receive blood transfusions (10-40 percent of all transfusions, depending on the area studied) due to acute blood loss during delivery or abortion, or high incidence of anemia associated with malaria or obstetrical problems. 40 NACP 1991, p. 9. du Greenberg, et. al. 1988. V Lepage and Van de Perre 1988. ' Chin and Sonnenberg 1991, citing Naegele, et. al. 1989. Jager, et. al. 1990. 4 Over and Piot 1990, citing Greenberg et al, 1988 and Ryder and Mhaly 1988. The records review by Jager et. al. (1990) in Zaire noted that pregnancy-related complications were the most important cause for transfusions in adults aged 15-40, accounting for 43% of the transfusions in this age group, or 15% of transfusions in all age groups. - 24 - Contaminated Needles and Instruments 2.39 Unfortunately, the general lack of resources in the health care sector of Tanzania increases the likelihood that needles and other instruments used in the provision of health care will not be sterile. Disposable instruments may have to be reused, or fuel may not be available to boil water. Protective clothing is often not available. In addition, certain traditional ceremonial and medical practices use invasive instruments which are rarely sterile. Such conditions suggest a high risk that instruments contaminated with HIV-infected blood could result in HIV transmission to uninfected individuals. 2.40 On the other hand, skin piercing does not appear to be a particularly efficient method of transmitting HIV. Studies of needle-stick incidents in health care workers in developed countries suggest that the risk after an "average" exposure or puncture wound from a contaminated needle is between 0.3 and 0.5 percent, or about one in 200-300.)' Needle puncture wounds are not a frequent occurrence among health workers: three different studies between 1981 and 1984 showed rates of 8, 10, and 16 per person per year.4' Studies in Sub-Saharan Africa indicate that use of unsterile needles, syringes and other skin piercing and invasive equipment, both in modern medicine and traditional practices, account for a very small portion (1-1.5%) of HIV infections in this part of the worldA There are no indications that Tanzania differs from this pattern. 2.41 While the risks of HIV transmission through needle wounds and traditional practices in Tanzania therefore appears to be quite low, the importance of these routes of transmission should be monitored since HIV infection levels and numbers of AIDS cases are rising nationwide, meaning that the likelihood of any used needle or surgical instrument being contaminated with HIV-infected blood is increasing as well. 2.42 Sterilization, especially using a heat method, is very effective at inactivating HIV. Tlbrefore, promotion of proper sterilization of medical equipment is an important control measure for decreasing this mode of transmission. Use of disposable needles and development of guidelines to prevent needle stick injuries are other measures which have been promoted to decrease the risk of HIV transmission. Finally, traditional practices using invasive instruments for purposes which have no positive health value (for example, the practice of female circumcision, which in fact poses serious health dangers) should be strongly discouraged.4' 4' Chin and Sonnenberg 1991 (one in 200) and Bjorkman 1991, citing Goldsmith 1990 (one in 300). 47' Bjorkman 1991. Bjorkman also cites a 1988 study which showed that devices requiring disassembly had rates of needle stick injuries five times the rate for disposable syringes (Jagger et al, 1988). This is part of the justification for using disposable needles to reduce risk of HIV transmission. !' Chin 1989; Bjorkman 1991, citing Ngaly 1990 and Chin and Sonnenberg 1991. 42' See also Tanzania: Women and Development, World Bank 1991, p. 45. - 25 - Other Routes of Transmission 2.43 The following routes of transmission are notable only because they are considered to be insignificant in Tanzania, either because of their inefficiency or because the numbers of individuals affected are very small. Evidence to date suggests that contaminated breastmilk does not contribute significantly to transmission of HIV infection from mother to child.-Q In addition, homosexual transmission, needle sharing among intravenous drug abusers, and treatment of hemophiliacs are all modes of transmission which need not be of concern to Tanzanian policy makers because the numbers of susceptible individuals in these risk groups are very small.51' 2. Assumtlions Regarding Disease Progression 2.44 Two types of assumptions are important for predicting the impact of HIV infection on AIDS morbidity and mortality in Tanzania. These are the rate of progression from HIV infection to AIDS, and the progression rate from AIDS to death. Since country-specific data are not yet available to allow calculation of these rates in Tanzania, international estimates (derived from cohort studies of homosexual men and hemophiliacs in developed countries) have been used in the epidemiologic-demographic models referenced in this report. Past research in Africa has suggested that African estimates may not vary greatly from these international estimates, although recent Tanzania-specific data indicate a somewhat faster progression.-W Pgression from HIV Infection to AIDS 2.45 Table II.4 shows the cumulative percentage of HIV infected persons who would be expected to progress to AIDS within a given time period from infection (adults and infants). The table illustrates the long latency period of the viral infection. In adults, only 15-20 percent of HIV infected persons are expected to progress to AIDS in the first five years after infection; only 50 percent may develop AIDS within 10 years, and an estimated 5-10 percent may still be AIDS-free 20 years after infection. In children progression to disease is faster: 25 percent of perinatally infected infants are expected to develop AIDS before their first birthday, while about 80 percent may have AIDS by the time they are five years old. It is assumed that these infant progression rates can also be applied to children who may become infected through blood transfusion. 2.46 The above research suggests that the rapidity of the disease progression from infection to onset of AIDS can vary from several months to 20 years or more. -' Over and Piot 1990, citing Ryder 1990. AV Chin and Sonnenberg 1991; Chin 1989; Over and Piot 1990; Bjorkman 1991. 2' Chin 1989, citing Piot et al, 1988, notes similar estimates among international and African research. Killewo, et. al. 1990, cited in Ainsworth and Rwegarulira 1991, provides evidence that in Tanzania, the progression rate may be faster. The epidemiologic-demographic models summarized in this report assume that the progression rates in Tanzania are equal to international estimates, since Killewo's data were not available at the time these models were run. - 26 - TABLE L.4 CUMULATIVE PERCENTAGE OF HIV INFECTED PERSONS WHO MAY PROGRESS TO AIDS WITIN A GIVEN PERIOD OF TIME FROM INFTION, ADULTS AND INFANTS Years from HIV Adults Infants Infection 1 0.2% 25% 2 0.5% 45% 3 10% 60% 4 15% 70% 5 20% 80% 10 50% 100% 15 75% - 20 90-95% __ Sources: Chin and Sonnenberg 1991; Chin 1989; and Over, personal communication, 1/15/91 Prorssion from AIDS to Death 2.47 Once a patient has AIDS the time to death is not long, as shown in Table H.5. An estimated 80 percent of adult AIDS cases in Tanzania are likely to die within 12 months, and 90 percent may die within 24 months. Again, children are expected to die faster: approximately 90 percent of children with AIDS will probably die within 12 months; it is presumed that almost all will die within 24 months. TABLE 11.5 CUMULATIVE PERCENTAGE OF PERSONS W1TH AIDS WHO MAY DIE FROM AIDS WTHIN ONE AND TWO YEARS OF DIAGNOSIS, ADULTS AND CHILDREN Years from Adults Children AIDS diagnosis l 1 80% 90% 2 90% 100% Source: Chin and Sonnenberg 1991 - 27 - 2.48 Before dying, a typical adult Tanzanian AIDS patient will suffer two or more episodes of chronic diarrheal disease, ten episodes of oral thrush, and about three skin infections. In addition, approximately 50 percent of adult AIDS patients will fall ill from pneumonia and/or septicaemia, and 15-25 percent will suffer from tuberculosis, severe headache, and/or neurological diseases, bringing the average number of epigodes of illness per adult AIDS patient to 17. These episodes of opportunistic infections will require over 280 days of care, including home care. 2.49 Infants will also suffer greatly before dying. About six episodes of Ulness are expected per child AIDS case, including a combination of chronic diarrhea, oral thrush, pneumonia and skin diseases, as well as recurrent fever and TB in some cases. As with adults, the care burden will be high, with 200 days of care required on average per patient. C. FUTURE DEVELOPMENT OF THE EPIDEMIC 1. Projections of Infections. Cases, Deaths and Distribution Infections and Cases 2.50 Using Bulatao's model with the 15% monogamy scenario, by the year 2010, the number of new infections annually in Tanzania will be about one million, implying HIV prevalence among adults of about 30 Dercent in 2010. Cumulative AIDS cases will have reached 6.4 million, with about 5.6 million cumulative deaths. 2.51 The 45% monogamy scenario predicts less infection overall, although a still serious epidemic: 400.000 annual new infections by the year 2010. with 10 prcent of adults in the count infected with HV. Cumulative AIDS cases would total about 2 million, with 1.7 million deaths. 2.52 The Chin-Sonnenberg model, meant for much shorter term projections (3-5 years), projects a cumulative total of 125.000 AIDS cases by the year 1992 (80,000 in adults and 45,000 in children), with about 103,000 annual new infections and HIV prevalence rates of 11 percent among urban adults and about 6 percent among rural adults (see Table II.6 and Figures I.4-6).P N Estimates are from Pallangyo and Laing (1990) and are based on a survey of a small number of Tanzanian clinicians. The authors note that some patients may suffer from multiple illnesses concurrently, which would reduce the number of individual episodes somewhat. See Chapter III for fiuther discussion. & Days of care required assumes that the current referral system is in place. Pallangyo and Laing also estimate the days of care required if treatment protocols and referral criteria are changed to encourage more decentralized treatment in peripheral facilities. See Chapter Im for details. a' Chin and Sonnenberg 1991, p. 6. - 28 - TA;BLE li.6 - Faiae IndneadPeaec fHVI nnadTnai :~~ ~~ jJ ! I ; I1,,L ++iII i ;lI!III!!II I! !~ 3 6 t1 ~-E i _ i lj - 29 - FIGURE 1.4 - Estimated Number of New HIV Infections Among Adults In Minland Tanzania HIV Inftlone/year 120,000 100,000 d_e tood bwbn tula; d_ to heteornx Ianhi^slon 60,000 40,000 20,000 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 Year Note: Blood screening began In 1987 As_wd covee: 197 10%, 1968 25%, 1969 50%, 1990 75%. 199 + 90% Sourc : Chin and Sonnenberg 1991, Figure 2. FIGURE ILS - Estimated Number of New NIV Infections Among Children in Mainland Tanzania HIV Infections/year 40o000 Pa mOlb HIV hdoo due to blood Ir_'bn Padilo HV I_ onlf -due lo 30,000 _ 20,000 0 _ 78 79 80 81 82 83 84 85 86 87 88 69 90 91 92 Year Iebt Wood g beWgn I 17 Am_d coveage: 19710%, 92%, 19 I %, 19907%, 1+ Sour: Chin and Sonnenberg 1991, Figure 3. - 30 - FIGURE 11.6 - FAdmated Cumulative A_S Cases PAdlatric and Adult - *Mnland Tanani Number (Thousands) 140 120 100 80 7 60- 40 0 1- , - --I---i- -r-r-|~- Il-W2r -- -f~ 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 Year [ii] Paediatric AIDS LE] Adult AIDS _ TOTAL Sour : Chin and Sonnenberg, 1991, Figure 4. 2.53 The NACP has made two projections of the epidemic on the basis of current estimated HIV prevalence levels. The first concerns projected AIDS cases if no further transmission were to take place. Even in that instance, cumulative cases would rise from about 80,000 at present to some 450,000 by the year 2000, with about 35,000 new cases per year. The second projection assumes that new infections continue to occur at a rate of one percent per year up to 1995, in which case cumulative AIDS cases would number about 750,000 by the year 2000. Cumulative Deaths 2.54 Table H.7 and Figure I.7 illustrate the projected cumulative burden of AIDS deaths according to Bulatao's model, ranging from 1.7 to 5.6 million, through the next 20 years. According to the 15% monogamy scenario, cumulative AIDS deaths may surpass half a million by the year 1994. The alternative scenario (45% monogamy) shows a slower growth in the burden of deaths, but both projections predict that AIDS deaths will exceed one million by the year 2006-perhaps even as early as 1998. - 31 - TABLE 11.7 - PrWlet Cumulative AS Case, and Deaths in Taana. AII (in thousands) Prolectian Scenario 15% monooaK vecmytV C.uulativi CiMiletv e CuUetetuIv CwletIve Yeor Ca Oeaths _ Cases DeathL 1988 109.15 76.07 19.76 13.75 1989 160.78 115.50 30.88 21.77 1990 224.22 165.41 46.31 33.21 1991 301.32 226.93 66.55 48.63 1992 394.82 301.91 91.93 68.45 1993 507.44 392.&Z 122.45 93.03 1994 641.75 501.31 159.80 122.82 1995 800.28 630.77 203.70 158.54 2000 1,987.70 1,630.21 548.77 446.63 2005 3,866.97 3,267.54 1,131.80 951.13 2010 6,424.61 5,554.05 1,966.52 1,707.22 Sou ~rce uatao 1990. FIGURE 11.7 - Cumulati_e AIDS Case and Daths Und Diffent SCenari Sourc Bulata 1990, Figure 6. 2.55 Indirect effects on mortality may be anticipate as well; for exaimple, uninfected 8mall chfldren whose mothers die of AIDS may be more likely to die from other causes than thiey would have been had their mother lived.- Orphaned children of all ages may have difficulty finding resources to pay for food and shelter, which may lead to higher mortality. Ibese indirect effects have not been quantified in Tanzana or elsewhere in Africa, but should be considered nonetheless in policy decisions. -W Ainsworth and Rwegaruira 1991. See also Carballo and Carael 1988. - 32 - Changes in Distribution 2.56 Evidence from seroprevalence surveys suggests that the discrepancy in age of HIV infected males versus females (with females being ,nfected at younger ages than m; !s, as shown in Figure 11.3) may be widening over time, as is also predicted by epidemiologic- demographic modeling in Figures I.S and II.9. This may be because (already infected) men are seeking ever younger partners in an attempt to avoid infection, as has been reported in a number of countries.2' FIGURE L.8 - Annual New Adult MV Infections, by Sex 4. Source Bulatao 1990, Figure 4. FlIGURE II. - Penm of Adult Males ,and Females Infected wilb SIV a.~~~~~~~~~~~~~~~~~ Soure Bulatao 1990, Figure S. Btlamptey and Potts, 'Targeting o4f Prevention Programs in Afica,' (p. 146), from Handbook .4r Al)S Preveigon in Af4rica, P. Lamptey and P. Piot, editors. - 33 - 2.57 Regarding rural versus urban prevalence rates, studies in Tanzania and other parts of East-Africa indicate that HIV infection in rural areas is invreasing, while seroprevalence levels in urban areas such as Dar es Salaam and Mwanza may be reaching a plateau.8' 2.58 The level at which HIV prevalence will plateau is difficult to predict, for it depends on behavioral as well as epidemiological characteristics. For example, in Kinshasa, Zaire, general seroprevalence seems to be leveling at 3-6 percent, while in Kigali, Rwanda, seroprevalence is leveling off at 33 percent.. One factor which accounts for different prevalence equilibrium rates is the size and degree of interaction between various "core' and "non-core" population groups. Core groups, characterized by higher rates of partner change, are at higher risk for HIV transmission and have higher prevalence rates than non-core groups. Since core group members interact with non-ore group members, they increase the prevalence of HIV in the non-core group as well, although the effect of core/non-core interaction cannot yet be predicted with accuracy in Tanzania due to lack of data. 2. ImDact on Mortalitv Crude Death Rate. IMR. and Life Eectancy 2.59 Figures I. 10-12 show three indicators of mortality in the absence of AIDS and with AIDS (according to the two Bulatao projection models). These indicators include crude death rate (Fig. I. 10), infant mortality rate, or IMR (Fig. 11. 11), and life expectancy at birth (Fig. 1. 12). In the absence of AIDS, the indicators show a projected steady decline in mortality. -ihe presence of AIDS slows this decline, and in fact reverses it in some cases. The changes in mortality indicators by 2005-10 are summarized in Table 11.8, which shows that the 15% monogamy scenario (which assumes that only 15 percent of married couples are mutually monogamous) predicts the greatest effects on mortality. ' Chin and Sonnenberg 1991. World Bank, "AIDS and Population and Family Planning Policies and Programs in Africa." It is unknown whether the -ttmates are for all adults, sexually active" adults only (i.e. ages 1549), or for the total p-b .wulation, adults and children. - 34 - FIGURE 11.10 - Crude Death Rate (per IhonuaLd) B , a . 400 Sorc: Bulatao 1990, Figure S. F IGUR 1111.2- kS^pt EotaUt Rate Brthosad 4.. 40 aogrc: Bulatao 1990, Figure 9. FIGURE 11.11- I LiSn EMxprtaflt Rate Bi~rthouad 433__ 40. em o _a.a . __ 0 *_0 * 11.3 : Bulao 1990, Figure 8. - 35 - TABLE I.S - Change. in Mortality Indicators with AIDS by 2005-10. As Compared to Base Projecns Without AIIDS Projection Scenario Change in Crude Change in Infant Change in Life Death Rate Mortality Rate Expectancy at Binh 15% Monogamy +100% +30% -35% 45% Monogamy +35% +10% -% Source: BuWatao 1990, from text on p. 4. 2.60 Using a different projection model, Chin-Sonnenberg predict that AlIDS deaths will increase child mortality by 10 to 15 percent during the early to mid 1990s (see Figure 11.13).& Adult mortality (age 15 and above) is expected to almost double by 1992 (from 7 deaths per 1,000 adults to 13.7 deaths per 1,000 adults), and to triple by the year 1994 (reaching 24.8 deaths per 1,000 adults, as shown in Figure 11.14).W In both children and adults, AIDS is expected to surpass measles and malaria as the major cause of death early in the 1990s. 9 Chin and Sonnenberg 1991, p. 6. This is an increase over the current child mortality rate, assumed by the authors to be 200 child deaths per 1,000 live births. The rate of 200/1,000 is higher than the estimates of under fives mortality (age 0-5) provided in the Word Development Report 1990 and in UNICEF's State of the World's Children 1990 (135-153 and 176, respectively). The authors do not define whether childm refers to all children under age five, or only to children age 1-4. In addition, the figures do not compare mortality with AIDS to projIected mortality without AIDS, a baseline which would presumably be lower than current mortality rates, as predicted in Bulatao's scenarios. IL Compared to the Bulatao projecton scOarios, the WIHO model predicts less infection overall, but higher mortality, indicating a problem with the construction of the demographic portion of the WHO/GPA model. -36 - FIGURE li.13 - proJeted Change of Child Mortality DMe to EIV/AJDS in Tanzania CM/1000 live birth F .JCM* without HIV/AIDS FP CM with HIV/AIDS 170 156 1 | -- - - ~~.14715 136 1980 1984 1988 1992 1995 * Projections of the UN ourc: Figure 6, Chin and Sonnenberg 1991. FIGURE II.14 - Projected Increase In Adult MortaUtv Due to HIV/AIDS In Tanzania Adull Mortality /1000 Adults 35 30 24.8 25 20 13.7 .. 15 8.8 7 10 5 0 1982 1988 199~~~~I2 1994 Year _______ F]AM without HIV fl AM with HIV Sou: Figure 5, Chin and Sonnenbeg 1991. - 37 - Empirical Data on HIV Prevalence and Mortality 2.61 Tbe relationship between HIV prevalence and mortality has been estimated empirically, both in developed countries and in Tanzania. Extrapolations from the progression rates from HIV to AIDS (Table II.4) based on U.S. data suggest that with every 10 percent increase in the HIV prevalence rate among sexually active adults, the adult mortality rate (ages 1549) will increase by about 5 per thousand.- If one assumes a baseline of 5 deaths per 1,000 adults age 15-49 without AIDS (an average for Sub-Saharan Africa), a 10 percent HIV prevalence rate implies a doubling of the mortality rate among adults in this age group, while a 20 percent HIV prevalence implies triple the deaths normally expected in this population age group.1y 2.62 The impact of AIDS on mortality in Tanzania can also be assessed directly through analysis of locally collected information. Data from the 1988 Census of Population has allowed the calculation of mortality rates for adults aged 15-49 for all twenty regions of mainland Tanzania, shown in Figure II. 15. The overall mortality rate for adults 15-49 in Tanzania was 5.8 per thousand in 1988. Tbe mortality rates range from a low of 3.4 deaths per 1,000 adults in Singida, to a high of 9.4 deaths per 1,000 in the Kagera Region. While ten years before, Singida and Kagera had similar life expectancies, by 1988 Kagera, hard-bit by the AIDS epidemic, had an adult mortality rate almost three times higher than Singida's.&2 2.63 Within the Kagera Region, further empirical analysis has established the relationship between HIV prevalence rates and mortality rates by district. Figure H. 16 illustrates that the two districts with the lowest adult mortality rates (4.6-4.7 deaths per 1,000 adults ages 15- 49), Biharamulo and Ngara, also have the lowest HIV prevalence rates (ess than 1%). On the other hand, high mortality districts such as Muleba and Bukoba (13.2-14.8 deaths per 1,000) have much higher HIV prevalence (9.7 - 13.1%, respectively). The analysis found that a one percentage point increase in HIV prevalence is associated with an increase of 0.91 deaths per thousand adults aged 15-49. This relationship tested statistically significant at p < .02. An increase in the seroprevalence rate of 10 percentage points would thus be associated with an increase of 9.1 deaths per 1,000 adults, compared to the 5 additional deaths S This relationship is based on empirical data from cohort studies in the United States and is explained in further detail in Ainsworth and Rwegarulira 1991, p. 2-3. e The baseline of 5 deaths per 1,000 is used by Ainsworth and Rweganlira to make the Dlustration clearer. Chin and Sonnenberg use a baseline of 7 deaths per 1,000 for Tanzania, which may be closer to the true value. Bulatao does not explicitly input a baseline adult mortality rate in his models; rather, adult deaths are calculated as a result of other assumptions in the demographic and epidemiologic modeling (infant and child mortality rates, crude death rates, etc.). & This discussion is summarized from Ainsworth and Rwegarulira 1991. Tbey note that while the 1988 census data "is clearly an underestimawc of true mortality levels,' the relative differences among regions may be indicative of the impact of AIDS. w It is not possible to calculate adult mortality rates for the earlier time period, since only life expectancy data were given in the source article (Sembajwe 1983, cited in Ainsworth and Rwegarulira). e Ainsworth and Rwegarulira 1991, p. S. -38 - per 1,000 adults extrapolated from the United States progression data described in paragraph 2.60. The relationship between HIV prevalence and mortality therefore seems to be much stronger in Africa than in the United States. FlGURE I.15- Adult Mortality Rate. Mainland Tanzania. 1988 Adult mortality rate, 15-49, per 1000 Twenty regions, Mainland Tanzania, 1988 1 94 §8 7.9 7A a 7- 6.6 ~~~~~~ ~ ~ ~~~~6.3 __ R 5- 4-~~~~~~~~~~~~~~~~~~~~~ ,2- 3.5 . 3.4 0 3- E Anu Coa oar D iri Ke a uUn mar Mb. Mor'MwwRkRYS i TabTa Region §~Adukdeath rate Ace: Winsworth and Rwegarulira 1991, Figure 1, based on raw Census data. - 39 - FIGURE H.16 - Prevalence and Adult Mortality. Kagera Regiln. 1987.88 Seroprevalence and adult mortality egera Region, by District, 1987-88 6e- _,.. i 6 , 14- _14 0 i 2 0.4 g . , | E~~~~1.3 0 1122 12 12. 12 | Adulf sero rate - Adult dea tot 6-~~ 3. 4mac 4on 4etiit 4.6 4- E m *~~~~K# km U/R Adult sero rate MAdult death rate Sou~ Ainsworth and Rwegarulira 1991, Figure 2, based on Killewo, et al (1990) and the 1988 Tanzania census. 3. Imnac on Fertility F2 2.64 'The impact of AIDS on fertility is uncertain, since it depends not only on the proportion of the population who will be alive to bear children, but also on how couples may modify their preferences and change their reproductive decisions in light of the AIDS epidemic. It is easier to predict the former than the latter. 2.65 The proportion of women in the child-bearing age group may not change very much, given the counterbalancing mortality among children infected with HIV. Bulatao's models predict that by the year 2020, the proportion of all women who are women of child-bearing age may decrease by 1-1.5 percent. As a percentage of the total population, the proportion of women of child-bearing age could decrease by about 0.5 percent. These figures are shown in Table 11.9. 0' This discussion draws heavily on World Bank 1990, "AIDS and Population and Family Planning Policies and Programs in Africa." - 40 - TABLE 1.9 - Effect of AIDS on Fertilit: Projected Chanee in Proportion of Women of Child-beardng Age by the Year 2020 Women Ages Tote' Total Pop., Women 15-49 Women 15-49 15-49 Female Pop. Male and as a X of ss a X of Projectfon Scenario (000) (000) Female Total Fem. Tot. Pop. l___________ ____________ (000) Poputation talelFemale Without AIDS 16.890 34,101 67.784 49.53X 24.92X AIDS. 15X Monogamy 11,768 24,169 U.067 48.69X 24.48 AIDS, 45X Monogamy 14,256 29,736 58,498 47.94X 24.37X Projection scenarios with AIDs are from Bulatao 1990, and have been described elsewhere. Projection scenarios without AIDS are from Bulatao, personal communication (1/2/91). 2.66 The effect of AIDS on reproductive decisions is less clear. On the one hand, couples who, due to their own experience or that of their neighbors, perceive an increase in child and adult mortality, may wish to have more children to offset this possible loss. Even women who know they are infected and may pass the infection to their offspring may wish to continue having children, since there is a 60-70 percent chance their child will not be infected with HIV. This may be especially true for childless women. Hoping to ensure that their wives are uninfected with HIV, men may marry younger women who will consequently have a longer number of years in which to bear children. All these factors may help to increase the total fertility rate. 2.67 On the other hand, couples who take in children orphaaed by AIDS may be reluctant to increase their family size any further and so may refrain from having more children themselves. Even families who have not yet fostered children may begin to anticipate this situation as they see neighboring families take in children who have lost a parent or parents from AIDS. Women who know they are infected with HIV may refrain from having children so as not to take the 30-40 percent risk of infecting their baby. These factors may help decrease the total fertility rate. The combined effect of both positive and negative influences on fertility cannot be measured. Further research on this topic may provide additional insight to reproductive decision making among infected and non-infected couples. - 41 - 4. Mitgrion L' Effect of Migration on AIDS 2.68 Many studies have produced evidence suggesting a direct link between migration and increased seroprevalence, although this hypothesis is not espoused without controversy.69' The role of migration in increasing the spread of AIDS is suggested by cohort studies which ask about travel history, as well as by the existence of urban-rural differentials in prevalence, high seroprevalence rates found in areas nearby major roads and commercial transport routes, and increased prevalence of HIV-2 infection in areas of West Africa which seemed to be previously uninfected with this form of the immunodeficiency virus. 2.69 One survey in Tanzania found that travelling outside the region but within the country was associated with an increased risk of HIV infection in rural areas; however, travelling was associated with lower risk of infection in urban areas.m 2.70 Yet, studies in South Africa and Cape Verde Islands have shown that migration does not necessarily increase seroprevalence.2L' The existence of conflicting evidence and theory suggests that further research is needed to fully understand the role of migration in spreading AIDS. Unfortunately, despite WHO sponsored collaboration to campaign against obstacles to international travel, many countries seem to be adopting "better safe than sorry" policies for AIDS prevention which include segregation of persons with AIDS, deportation of infected migrant workers, and myriad obstacles to travel.-' Effect of AIDS on Migration 2.71 Most literature about AIDS and migration describes the effect of migration and international population movements on HIV prevalence rates and AIDS, rather than the reverse relationship, i.e. the effect of the AIDS epidemic on migratory patterns. Two authors who do touch on this topic, albeit briefly, are Carballo and Carael (1988), whose arguments are summarized below. a This discussion draws heavily from chapter VII, "Migration and Health,' of Sharon Russell, et. al.'s World Bank Discussion Paper, "International Migration and Development in Sub-Saharan Africa (Vol. I, Overview), in addition to the other references cited. 69' Hunt 1989; Micklin and Sly 1988; Chin 1988; Nzilambi, De Cock et. al. 1988, De Cock 1988, Simonsen et. al. 1988, Gachihi and Mueke 1988 and others cited in Russell, et. al. 1990. Russell discusses the controversy (documented in an exchange of letters in the &M England Journal of Medicine! on page 78, noting especially the opinion that urbanization and social disruption in Africa existed long before the AIDS epidemic. ' Killewo et. al. 1988, cited in Russell et. al. 1990. ' Brink et. al. 1987, Lyons et. al. 1987, Brun-Vezinet et. al. 1987, cited in Russell et. al. 1990. N Parks 1987, WHO n.d., cited in Russell et. al. 1990. -42 - 2.72 Carballo and Carael note that in recent years mortality differentials between urban and rural areas have emerged (10-20 percent lower in urban than rural areas) in many developing countries due to higher education levels, environmental sanitation, and access to medical care in urban areas. Such differentials may be part of the motivation behind rural-urban migratory patterns. The authors suggest that with the AIDS epidemic, urban and rural mortality rates may begin to converge, which could in turn affect how rural people perceive urban centers and might decrease the rate of rural-urban migration (p. 84). 2.73 On the other hand, it is conceivable that deaths in urban areas due to AIDS will drive up urban wages, as experienced workers become scarce. A widening wage differential batween rural and urban areas would tend to increase rural-urban migration. 2.74 The AIDS epidernic might also increase the rate of urban-rural migration in the following way. Migrants to urban centers are typically characterized as young and single, often "marginally employed and employable." Lacking community affiliation, these individuals have few social support resources to draw on for health problems, should they become infected with HIV. According to Carballo and Carael, "anecdotal evidence from some countries suggests that resultant 'out' migrations of people seeking care and support from families still living in rural areas can be anticipated," but the magnitude of such an effect is difficult to estimate (p. 84-85). This phenomenon is also noted by Hunt (1989), who suggests that HIV infected, sick migrant laborers and prostitutes returning to their rural homes is probably a major cause for the spread of the AIDS epidemic in rural areas.' Researchers in Ghana have noted increases in HIV prevalence in areas of the country where many Ghanaian prostitutes who worked in Cote d'Ivoire had returned home, often in order to seek care for AIDS-related opportunistic illnesses. 2.75 In general, Carballo and Carael believe that the AIDS epidemic is unlikely to affect the overall urbanization patterns in Central and Eastern Africa, where urban growth rates now reach 6-8 percent per year. It is not so far-fetched, however, to believe that cities could become identified with AIDS infection in the eyes of certain more mobile groups, such as traders, seasonal migrants, and young people seeking higher education, and that these population groups might become more hesitant to move to urban areas (p. 85). On the other hand, AIDS patients and families of persons with AIDS may be encouraged to move from rural to urban areas, seeking specialized health care services available only in tertiary hospitals. 2.76 In their article on the impact of AIDS on food production in Eastern and Central Africa, Abel, et. al. suggest that the level of intensity of contact between rural and urban areas can influence the progression of the AIDS epidemic. Urban-rural contact may vary among and within countries. For example, Urban-based civil servants in Kenya with shambas in Central Province may ensure the diffusion of AIDS to rural areas faster than, for example, in Western Province of Zambia where such urban-rural links do not exist, Z' The author cites evidence from Doyal 1981 as well, on movements of sick people from urban to rural areas. W Neequaye, et. al. 1988. - 43 - despite the possibly higher ambient incidence of AIDS in Zambia than Kenya. Urban-rural trade and labour movement paths may also determine the spatial pattern of diffusion. An additional set of risk factors might be proximity to rural dispensaries or ease of transport to larger medical facilities where infection can occur through blood transfusion or injection. (p. 148) 2.77 The authors believe that in discussing the implications of AIDS on the national economy, the effect of AIDS on migration patterns is one of the greatest areas of uncertaint. Labor shortages and breakdowns in family coping and care systems will not occur evenly across the country, and will probably influence rural-urban, urban-rural and rural-rural migration. Further research is needed to estimate the possible effects of AIDS on migration. S. Population Size. Growth and Distrdbution 2.78 Since the effects of AIDS on fertility are very uncertain, it is difficult to project changes to population size and growth with accuracy. Nonetheless, models by WHO and the World Bank anticipate that population growth will slow because of AIDS, although the growth rate will still be positive. Bulatao estimates that the population growth rate will fall by between 0.3 and 1.1 percent, from 3.1 percent in his base projection for the years 2005-2010 to between 2.0 and 2.8 percent (see Figure il.17). This represents a 10-35% drop in the population growth rate.;F FIGURE II.17 - Popuaton Growth Rate Under Different Scenarios (prnt) 3.. Soure Bulatao 1990, Figure lO. 2.79 Chin and Sonnenberg estimate that by 1992, deaths due tD AIDS may decrease the expected incrementa growth in the urban adult population by 20 percent, while decreasing the expected increase in rural adults by 8 percent (p. 7). 2.80 Total population size will be less affected than the population growth rate. III Bulatao's most severe scenario, total population will be IS percent less in 2010 than one would expect given base projections without AIDS. Nonetheless, population would grow 2E Bulatao 1990, P. 4. -44 - from 2541 million over the 20 year period. His second scenario projects a decrease in total population of only 4 percent, as shown in Figure H. 18. FIGURE II.18 - Ttal Population Under Different Senarios do Source Bulatao 1990, Figure 11. 2.81 It is unclear ex ante just how irnportant shift in the population structure due to AIDS will be. Standard demographic analyses suggest that dianges in birth rates have a much more pronounced imnpact on the age pyramid than do shifts in mortality rates, at least in cases where the latter have roughly similar irnpacts on all age cohorts. To the extent that AIDS deaths involve infants, the effect on population growth is similar to a reduction in birth rates; that is, large changes in the age pyramid might be expected. To the extent that AIIDS increases mortality rates over a broad span of the adult population, on the other hand, we would expect the impact on the age structure to be less pronounced. 2.82 If AIIDS is most prevalent among the sexually active age cohorts, as appears to be the case, a significant proportion of them will not reach old age. The net result should be a reduction in the proportions of the population in the older and older age cohorts. The result would be a progression towards a younger population due to AIIDS. .-V 2.83 An analysis of the age structtre of the population in the no-AIDS and various with- AIDS scenarios for Tanzania offers some interesting conclusions. According to Bulatao's sirnulations, the epidemic is predicted to have only a modest effect on the dependency ratio (i.e., the population under age 15 and over age 64, as a proportion of the population age 15- 64) since deaths will occur among both adults and children. 'Me dependency ratio depends much more importantly on fertility, which in Bulatao's model is assumed to begin decreasing Zv For recent analyses of the macroeconomic effects of the aging of the population in major inutilcountries (s the post-war *baby boom* generation matures), see Masson and Tryon (1990) and the references listed there. To the extent that AIDS causes the age distribution to shift away from the elderly, one might expect macroeconomic consequences that are exactly the opposite of what Masn and Tryon predict in the case of an aging popuation. - 45 - in the mid-1990's due to family planning efforts. As illustrated in Figure I. 19, the dependency ratio rises gradually from 98.6 percent in 1985 until the turn of the century, when it begins to decline. By the year 2020, it falls to 85.3 percent under the no-AIDS scenario verses 86.6 percent in the with-AIDS scenario. FIGURE 11.19 - DeRendency Ratio under DiMerent Scenarios I, 4as 4g5 91~ 3 we.a.o . _e, . *' , 4SU@ Source: Bulatao 1990, Figure 12. 2.84 ThIs time profile for the overall dependency ratio, however, hides larger offsetting changes in the youth and elderly dependency ratios. As Fig. 11.20 shows, the yDUIh dependency ratio rises by a couple of percentage points relative to the no-AIDS scenario between 1985 and the year 2000. The elderly dependency ratio (Figure Il.21) also rises slightly initially (presumably because the current elderly were spared from infection because the prevalence of HIV was extremely small when they were sexually active). Between 2000 and 2020, however, the elderly dependency ratio is about 2 percent less in the with-AIDS cases as compared to the no-AIDS case. The fact that the two ratios shift in opposite directions makes the MndlM ratio appear relatively unaffected. 2.85 The change in the age structure of the population is much more noticeable when 5- year age cohorts are examined. Figure 11.22 shows the percentage share of the population in each cohort in the year 2020, with and without AIDS. The effect of the AIDS epidemic will be to reduce the overall size of the population and to significantly shift the composidon of that population towards the young. The demographic shifts can be expected to have important macroeconomic consequences, which are discussed in Chapter m. -46 - FIGURE 1120 - Youth Dependency Ratio. Wlth and Without AIDS I 01 099 0.98 0 97 096 0 95 0 94 0 93 0 92 0 91 0 9 089 088s 087 086 085 084 083 082 081 1985 1988 1991 1994 1997 2000 2003 2006 2009 2012 2015 2018 0 No AIDS scenario + Madified Standard SaurEc: Cuddington 1991. FIGURE H.21 - Ejderly DPednc Ratio. Wth and VIhU _ AI 0 064 - 0062 006 0058 t 0056 - 0054 - 0052- 0.05 0.048 - 0046 - 0044 - 0042 - 004 - 0 038 1- . . . . . . . . . . . 1985 1988 1991 1994 1997 2000 2003 200S 2009 2012 2015 20183 O No AIDS scenario + Modifled Standard Aoure: Cuddington 1991. -47 - FIGURE 11.22 - AaL istribution In 2020. With and &Wjho t AMS 017 0 o 0155t9X3 4-9606 5 014 0.13 012 Oil 01I 009 008 007 006 005 0.04 003 002 001 0-4 15-19 30-34 45-49 60-64 75+ 0 No AIDS scenario + Modified Standard Sourc: Cuddington 1991. - 48 - II. CONSEQUENCES OF THE EPIDEMIC 3.01 In addition to increasing the morbidity and mortality burden of disease in Tanzania, the AIDS epidemic is likely to have significant influence on economic activity and development as well as on the delivery of services in the public sector. A. MACRO-ECONOMIC IMPACT OF AIDS Introduction 3.02 Given the alarming incidence of AIDS within various regions in Tanzania, and the projected growth of the epidemic over the next twenty-plus years, a crucial question is: will AIDS have important macroeconomic effects on the economy as the epidemic runs its course? To what extent can various policy interventions alter these macroeconomics consequences? 3.03 In an attempt to answer these important questions, a macro-economic model has been constructed which attempts to compare a "no AIDS" scenario to a "with AIDS" scenario.Y/ The following sorts of questions suggest what is meant by the 'macrjeconomic" effects of AIDS: a. How will the time path of potential output in the AIDS-stricken economy differ from what would have happened in the absence of the AIDS epidemic? b. Will er caita income be lower or higher in the "with AIDS" steady state and during the transition to this steady state? C. How will AIDS affect domestic (public and private) saving behavior? d. How will capital investment and the economy's capital/worker ratio be affected over time? Will the composition of national output shift in favor or against capital intensive sectors of the economy? e. Should we expec' higher or lower real wages? 3.04 In this report, the macroeconomic effects considered are limited to these highly aggregated variables, abstracting from sectoral/allocational considerations. AIDS' Channels of Macroeconomic Influence 3.05 The rising prevalence of AIDS in the population can be expected to affect the macroeconomy through a number of separate channels. At the most basic level, the effects of AIDS may be grouped into two categories, those associated with: (1) declining "healthiness" 1' Bulatao's (1990) demographic simulations were used as input to this model, dtcheloped in Cuddington (1991). The "with AIDS" scenario used is his most pessimistic variant, the "modified standard," also referred to in this report as the "15% monogamy" scenario. - 49 - of the population, and (2) rising mortality rates for particular age cohorts, especially sexually active adults and infants (infected at birth). 3.06 The decline in healthiness, in turn, has four major effects: it will (1) reduce labor productivity, (2) increase health care expenditure, (3) reduce saving, and (4) reduce human capital investment. Labor productivity falls because sick or worried workers are less productive than happy, healthy ones. Even the productivity of those who do not have AIDS may be affected as infection rates and illness among friends, families and coworkers rise. Higher health care expenditures are incurred by households and the (public and/or private) health care systems to assist AIDS patients and their families in coping with their worsening health condition. 3.07 The rise in health care expenditure is likely to have important consequences for other expenditures, current and future. One direct consequence will undoubtedly be reduced saying as society meets a large part of the increase in health spending by consuming a greater proportion of current income.2' Whether private or public savings is more severely reduced depends to a large extent on the nature (f the health care delivery system. The public sector has financed a large proportion of healt} care expenditures to date but, due to budgetary constraints, there is increasing pressure so recover additional amounts from users. Some of the increase in health spending may also be accomplished through reductions in other current expenditures. Anecdotal evidence suggests that reduced spending on educational investment for children may be especially important. Needless to say, the potential for slower economic growth due to AIDS is heightened to the extent that saving and investment in human capital are reduced. 3.08 Two additional effects of AIDS are caused by the increase in mortality rates, especially for sexually active adults and children (who contact AIDS at birth): (1) a reduced population growth rate and (2) a change in the age structure. The population growth rate declines as survival probabilities for various age groups decline. As discussed in Chapter 11, the population growth rate in Tanzania is likely to fall by between 0.3 and 1.01 percentage points as a result of the AIDS epidemic. There will also be important shifts in the age structure of the population. In Chapter 11, the youth dependency ratio was shown to increase by about two percent, and the elderly dependency ratio to decrease by about 0.5 percent over the next 20 years. Perhaps more important are two additional consequences: the size of the labor force (the working age population) grows more slowly than it would without AIDS, and the composition of the population shifts toward the young, with the labor force being composed of younger, less experienced workers. Figure 11. I shows that by the year 2010, the size of the work force will be only 80 percent of what it would be without AIDS, while Figure 111.2 indicates that the mean age of the working-age population (15-64) will fall, in the with-AIDS scenario, from 32 years to 29 years by 2010, and to 28 years by 2020, verses about 31 years without AIDS. 2 Care most be taken to consider the difference in health expenditures in the no-AIDS and with AIDS scenarios, because non-AIDS-related health care costs may be reduced if individuals die earlier and more quickly due to AIDS. 2' The direct dissaving effect is by no means the only impact of AIDS on saving. For example, higher AIDS prevalence may raise the precautionary demand for savings by households who experience greater income variability in the presence of AIDS. so - FIGURE m. - Woking Aged Population. 1935-20t1 24- 23 . 21 In 17- la -)l O N 9b + 14 13 12. It Ism Ism list IOU4 Ii? 2OWO 2005 2C 0 No AM Sweum + Mcdfiad Smviw Sourcs: Cuddington 1991. 3.09 Shifts in the age structure towards the young have at least four important consequences. They affect the size of the working age population, the labor force participation rate,.Y the composidon of public sector expenditures, and saving rates. A decrease in the elderly dependency ratio, for example, would have a dampening effect on per capita expenditures on health,F while an increase in the youth dependency ratio, particularly in developing countries, tends to increase per capita expenditures on health. As discussed in Section B. I of this Chapter, any savings in health care costs associated with the changing age structure in Tanzania will be more than offset by the higher health care costs due to AIDS. The smaller total size of the population (relative to a 'no-AIDS' scenario) suggests that total expenditures on education will be smaller tha. would otherwise be the case, but an increase in the youth dependency ratio implies an increase in education expenditures in per capita terms (discussed in Section B.2). Al Besides the direct effect of shifts in the fraction of the population that is working age, children of AIDS victims may be forced to enter the labor force earlier and therefore be unable to continue schooling. -' Per capita expenditures on pensions and social security would also decrease with a reduction in the elderly dependency ratio, but these are not significant in Tanzania. - 51 - FIGURE M.2 - Avage Age of Labor Fore. I20 m_ 215 255 278S I9B 19M 1991 1994 1997 200 2O. 2DO 200 212 DI5 2Dw o No AM nwwIb + Mofbi 9Iwtd.d Source: Cuddington 1991. 3.10 According to economic theory, the shift in Tanzania toward a younger population is likely to have a negative influence on savings rates. Standard life cycle savings channels predict that both an increase in the youth dependency ratio and a decrease in the elderly dependency ratio, as are likely to happen in Tanzania, will reduce household saving rates. The more children in a household, the greater the proportion of household income which must be used for consumption (of food, clothing, health and education). Household saving may also be reduced because of the relatively smaller number of elderly that will need to bi cared for in the future. Another factor likely to have a dampening effect on savings rates is the severing of intertemporal linkages among dynastic' households (as parents die, children are left either as orphans or in the care of grandparents of other rdatives); savings are not as great when the extended family shrinks. The increase in orphans also imply higher government costs (if not total societal costs) of raising children. Some of these negative effects on savings may, however, be moderated with time. As the epidemic progresses and people increasingly anticipate the need for greater health care expenditures in the future, savings rates may go up. 3.11 The third impact of rising mortality rates due to ADS is loss in uman ai. This refers to the premature loss of well-trained, educated and experienced workers. Tbis effect is distinct from the AlDS-induced shifts in the perceived costs and benefits from undetalking new human capital investments, discussed in Section B.2, below. The former is a 'stock' effect, which results from the unexpected early death of educated persons after the investments are made. This effect disappears over time as the reality of AIDS is increasingly -52 - taken into account and the economy approaches a new steady state. The latter effect is an ongoing "flow" effect; it remains important even once the with-AIDS steady state is reached. The human capital loss effect is exemplified by the drop in the average age and experience of the labor force, mentioned above, and revealed dramatically in Figure 11.22 in Chapter II. Labor force productivity is likely to fall, compounding the reduced productivity caused by deteriorating health conditions as AIDS spreads (i.e. the negative productivity effect discussed above). 3.12 The loss in human capital may be especially pronounced in the case of certain people with highly specialized skills. The illness and death of experienced managers and entrepreneurs, for example, may lead to productivity declines among others far in excess of their individual productivity. Replacing highly specialized people, such as doctors and nurses, will be especially difficult given the years of training and experience involved. A Simple Analytical Model 3.13 A simple analytical model was constructed to simulate the above effects of AIDS on the Tanzanian economy. The basic demographic input for the macro simulation model includes Bulatao's (1990) projections for the size of the total and working-age populations in each year from 1985 through 2014 in the no-AIDS scenario and his "modified standard" AIDS scenario, referred to in this report as the 15% monogamy scenario.Y Labor force is taken to be synonymous with the working-age population. His model provides information on the total number of AIDS cases (i.e., those who are ill with AIDS, not merely infected with HIV) in children and adult populations, with the latter being used in the model to describe the proportion of the working population with AIDS at any point in time. Both child and adult cases of AIDS are considered in the calculation of the annual medical costs of treating AIDS patients, which are derived from Pallangyo and Laing (assuming full drug availability), with costs assumed to be spread over 1.5 years to reflect the average length of time an AIDS patient lives. 3.14 Historical data on GDP, gross fixed investment, foreign capital inflows, gross domestic saving were used to get rough orders of magnitude for key parameters and future time series for certain macro variables in the model. Historically, domestic saving has been roughly 5 percent of GDP for the period 1968-1988. In the same period foreign capital inflows represented roughly 17 percent of GDP. 3.15 The simulations assume that foreign capital inflows would continue at their historical rate of 17 percent of GDP, and domestic saving would remain at its historical norm of 5 percent in the no-AIDS scenario. Various reduced saving rates are assumed in the presence of AIDS, as discussed in the previous section. Implicitly, it has been assumed that the economic reform program will continue. Total domestic plus foreign saving determines domestic capital formation. 3.16 Estimates of the capital stock are typically unavailable in developing countries. Therefore, an estimate was made of the capital stock in 1985 using Cobb-Douglas technology, 0' This may be considered as a "worst-case" scenario, as it produces the most pessimistic results of any of the AIDS scenarios modeled. - 53 - from which an estimate of the capital stock in each future year was obtained. A depreciation of 5 percent was assumed. 3.17 Given the initial capital stock, a constant was chosen to represent the rate of exogenous technological change in the production function which yielded a value of GDP equal to the actual value, and a rate of growth in per capita output under the no-AIDS scenario equal to roughly one half of one percent between 1990 and 2000.1' This was done in order to generate a no-AIDS scenario that seemed plausible. The analysis then compared this no-AIDS case to the situation where AIDS prevalence gradually worsens in order to determ.ine the net effect of AIDS on the economy. the Simulation Results 3.18 Simulations of the "with-AIDS" scenario were run using various values for the most important variables in the model; namely, the productivity parameter (z), "annual labor productivity loss per AIDS patient," and the saving parameter (x), 'the proportion of annual AIDS-related medical expenditures met through reduced saving." A value of 1.0 for z, for example, would assume that each AIDS case resulted in an annual loss of worker productivity, either because the ill person was unable to work for a year or because someone caring for him or her was unable to work, or both were unable to work for one-half year. A value of 2.0 would indicate that two full years of worker productivity were assumed lost for each AIDS case per year. A value of 0.5 for x would assume that half of the cost of caring for each AIDS case was drawn from savings (with the other half coming from reduced consumption), while a value of 1.0 would assume that all of the costs were drawn from savings. Values greater than I for the saving parameter were tried, in order to see what the effect might be if saving was reduced by more than the cumulative cost of treatment, due for example to the changing age structure of the population. Different combinations of these values yield different results for GDP, average growth rate of GDP, per capita GDP, average growth rate of per capita GDP, and the saving ratio in the year 2010. 3.19 The simulation results are shown in Table E1.1. Various values for the productivity and saving parameters form the axes of the matrix displayed in the table, and the resulting GDP, per capita GDP, and growth rates, etc. for the year 2010 are shown in the boxes of the matrix. These may be compared with the "no-AIDS" case, shown at the top of the table. The shaded box of the matrix, where z = 0.5 (implying that each AIDS patient results in an annual loss of one-half year of worker productivity), and x = I (implying that all AIDS- related health care costs are financed by reduced saving) is considered to be the most plausible. How representable it is of possible results can be seen by comparing it to other cases displayed in the matrix. The full results of that simulation are given in Table III.2 and m.3, for the "no AIDS" and "with AIDS" scenarios, respectively. Figures m.3-s compare the results with the "no-AIDS" case in graphic form. Z'This is the rate forecasted by the World Development Report. 1990 for Sub-Saharan Africa. - 54 - TABLE M.l - Results of Simulation Modelling Macroeconomic Effects of AIDS In Tanina A. No-AIDS Scenario 106,077- GDP In 2010 (in millions of constant 1980 TSh) 4.0 - average yearly GDP growth rate 1985-2010 (in percent) 2.23 - per capita GDP fn 2010 cin thousands of constant 1980 TSh) 0.7 - average yearly per capita GDP growth 1985-2010 (in percent) 0.05 - saving rate in 2010 B. With AIDS Scenario* LABOR PRODUCTIVITY LOSS PER AMDS A CASE I 0 0.5 1 1.5 2 D- S 91,206 90,036 88,863 87,688 86,510 3.4 3.3 3.3 3.2 3.2 C 0 2.23 2.20 2.17 2.14 2.12 0 0.7 0.7 0.6 0.6 0.5 S 0.05 0.03 0.05 0.05 0.05 T S 89,909 88,649 87,485 86,320 85,151 3.3 3.3 3.2 3.2 3.1 M 0.5 2.20 2.17 2.14 2.11 2.08 E 0.7 0.6 0.6 0.5 0.5 T 0.03 0.03 0.03 0.03 0.03 F 88,355 . 86,050 84,893 83,734 Dt 3.3 3.2 3.1 3.0 o 1 2.16 2.10 2.08 2.05 M 0.6 0.5 0.5 0.4 0.02 0.02 0.02 0.02 R D 86,836 85,694 84,549 83,402 82,253 D 3.2 3.1 3.1 3.0 3.0 u 1.5 2.12 2.10 2.07 2.04 2.01 c 0.6 0.5 0.4 0.4 0.3 E ~~~0.00 0.00 0.00 I 0.00 0.00 D ~~~85,246 84,112 82,977 81,839 80,700 3.1 3.1 3.0 3.0 2.9 s 2 2.08 2.06 2.03 2.00 1.97 A 0.5 0.4 0.4 0.3 0.3 v -0.02 -0.02 -0.02 -0.02 -0.02 1 N G S NaZi: * Bulatoo s modified-standard scenario, assuming 15X monogamy In Populoation. Eal IBIS to output ratio fn 1985 (K/tY) 0 Lrsshar ofotubets) o.r ProductIvity growth rate (pma-1) 0.008 Depreciation rate (theta) 0.05 Initial savIng rate (s) 0.05 Rate of cap tal lnflow (t*) 0.17 Medical cost of treating adult AIDS patient per year (ma) 3.23 Medical cost of treating child AIDS patient per year (mc) 2.47 Age efficiency intercept 0.8 Age efficiency linear gafn 0.02 Age efficlency quadratic gain -0.0002 Output level In 1985 39,131 Profil, of exoerfence pain In efficlency: 15-19 20.24 25-29 30-34 35-39 40-U 45-49 50-54 55-59 60-6 0.82 0.90 0.98 1.06 1.12 1.18 1.22 1.26 1.28 1.30 - 55 - TABLE M.2 - Output PrQiection: No-AIDS Scenario Key values assumed in simulatlon: Saving loss parameter (x) na Productivity loss parameter (z) na Capital to output ratio in 1985 (K/Y) 3 Output scale factor (alpha) 1.756829 Labor's share of output (beta) 0.7 Productivity growth factor (gamia-1) 0.008 Depreciation rate (theta) 0.05 Initial saving rate (s) 0.05 Capital inflow to output ratio (s*) 0.17 Medical cost of treating adult AIDS case per year (ma) na Medical cost of treating child AIDS case per year (mc) na Age efficiency intercept (rhol) 0.8 Age efficiency linear gain (rho2) 0.02 Age efficiency quadratic gain (rho3) -0.0002 Gross Labor GDP Domes. Adult Effi- GDP per Capital Fixed AIDS clency per Capita GDP Year Inflow Inves. Prey. GDP Cacital Labor Units EL Capita Growth Growth 1985 na na 0 39131 117393 10655 10925 11.02 1.85 na na 1986 6652 8609 0 40443 120132 1097 11211 10.95 1.85 0.07 3.35 1987 6875 8897 0 41853 123023 11300 11522 10.89 1.85 0.20 3.49 1988 7115 9208 0 43371 126080 11637 11861 10.83 1.86 0.34 3.63 1989 7373 9542 0 45005 129317 11984 12230 10.79 1.87 0.47 3.77 1990 7651 9901 0 46764 132752 12341 12630 10.76 1.88 0.61 3.91 1991 7950 10288 0 48437 136403 12719 12978 10.72 1.88 0.19 3.58 1992 8234 10656 0 50225 140239 13110 13353 10.70 1.89 0.31 3.69 1993 8538 11050 0 52139 144276 13512 13758 10.68 1.90 0.42 3.81 1994 8864 11471 0 54189 148533 13926 14195 10.67 1.91 0.54 3.93 1995 9212 11922 0 56386 153028 14353 14665 10.66 1.92 0.65 4.05 1996 9586 12405 0 58642 157782 14851 15135 10.62 1.93 0.53 4.00 1997 9969 12901 0 61046 162794 15367 15637 10.59 1.94 0.63 4.10 1998 10378 13430 0 63611 168084 15901 16173 10.57 1.96 0.72 4.20 1999 10814 13994 0 66347 173674 16453 16745 10.56 1.97 0.82 4.30 2000 11279 14596 0 69269 179587 17024 17356 10.55 1.99 0.92 4.40 2001 11776 15239 0 72085 185847 17599 17900 10.56 2.01 0.77 4.06 2002 12254 15859 0 75068 192413 18193 18476 10.58 2.02 0.84 4.14 2003 12762 16515 0 78230 199307 18808 19086 10.60 2.04 0.92 4.21 2004 13299 17211 0 81584 206553 19443 19732 10.62 2.06 0.99 4.29 2005 13869 17948 0 85143 214174 20100 20416 10.66 2.08 1.06 4.36 2006 14474 18731 0 88863 222196 20828 21121 10.67 2.11 1.21 4.37 2007 15107 19550 0 92802 230636 21583 21865 10.69 2.14 1.27 4.43 2008 15776 20417 0 96974 239521 22365 22649 10.71 2.16 1.34 4.49 2009 16486 21334 0 101393 248879 23175 23475 10.74 2.19 1.40 4.56 2010 17237 22306 0 106077 258742 24015 24348 10.77 2.23 1.46 4.62 Source: Cudington, 1991. - 56 - TABLE m3 - Output Projection: With-AIDS Scenario (15% Monogamy) Key values assumed in simulation: Saving toss parameter (x) I Productivity loss parameter (z) 0.5 Capital to output ratio in 1985 (K/Y) 3 Output scale factor (atpha) 1.757385 Labor's share of output (beta) 0.7 Productivity growth factor (gamma-1) 0.008 Depreciation rate (theta) 0.05 Initial saving rate (s) 0.05 Capital inflow to output ratio (s*) 0.17 Medical cost of treating adult AIDS case per year (me) 3.230 Medical cost of treating child AIDS case per year (mc) 2.467 Age efficiency intercept (rhol) 0.8 Age efficiency linear gain (rho2) 0.02 Age efficiency quadratic gain (rho3) -0.0002 Gross Labor GOP Domes. Adult EffI- GDP per Capftal Fixed AIDS ciency per Capita GOP year Inflow Inves. Prev. GOP CaDitl Labor MLtL Capita Growth Growt 1985 na na 0.09 39131 117393 10655 10920 11.02 1.85 no na 1986 6652 8568 0.16 40385 120091 10958 11185 10.96 1.85 0.48 3.20 1987 6865 8811 0.25 41727 122898 11269 11473 10.91 1.85 0.15 3.32 1988 7094 9061 0.35 43163 125814 11590 11786 10.86 1.86 0.27 3.44 1989 7338 9322 0.46 44705 128846 11919 12127 10.81 1.86 0.39 3.57 1990 7600 9604 0.56 46359 132008 12258 12498 10.77 1.87 0.52 3.70 1991 7881 9908 0.68 47812 135351 12585 12777 10.75 1.88 0.09 3.13 1992 8128 10160 0.80 49358 138710 12921 13080 10.74 1.88 0.19 3.23 1993 8391 10422 0.94 51002 142196 13266 13409 10.72 1.88 0.28 3.33 1994 8670 10693 1.09 52752 145779 13620 13764 10.70 1.89 0.38 3.43 995 8968 10974 1.25 54616 147464 13983 14148 10.69 1.90 0.48 3.53 1996 9285 11267 1.41 56355 153585 14371 14472 10.66 1.98 0.36 3.18 1997 9580 11524 1.57 58197 157118 14769 14822 10.64 1.92 0.44 3.27 1998 9893 11797 1.72 60149 161059 15179 15199 10.61 1.93 0.52 3.35 1999 10225 12091 1.86 62220 165097 15599 15605 10.58 1.94 0.61 3.44 2000 10577 12411 2.00 64418 169254 16032 16041 10.56 1.95 0.70 3.53 2001 10951 12755 2.14 66290 173546 16418 16346 10.57 1.96 0.54 2.91 2002 11269 13023 2.28 68264 177892 16814 16675 10.58 1.97 0.61 2.98 2003 11605 13314 2.41 70346 182312 17219 17029 10.59 1.99 0.68 3.05 2004 11959 13628 2.53 72542 186824 17634 17409 10.59 2.00 0.76 3.12 2005 12332 13968 2.66 74859 191459 18059 17815 10.60 2.02 0.83 3.19 2006 12726 14330 2.77 7705 196209 18500 18171 10.61 2.04 0.96 2.96 2007 13103 14670 2.88 79410 201069 18951 18552 10.61 2.06 0.03 3.03 2008 13500 15041 2.98 81872 206056 19414 18960 10.61 2.08 1.10 3.10 2009 13918 15444 3.06 84468 211198 19888 19394 10.62 2.11 1.17 3.17 2010 14360 15883 3.15 87203 21652G 20374 19855 10.63 2.13 1.23 3.24 Source: Cuddington 1991. . 57 - FIGURE m3 - GDP Projection, with and without AIDS 10~ 100 sgo J70 40 ~.a tiM tim i991 1994 1997 2000 201 00 2C61 a No b " =W= + M Sourc: Cuddington 1991. FIGURE mA - Per Cagita GDP Proiection, with and without AIDS 2M~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ - 22- 2.15 2 I..6 IM IBM low 199 1994 19O97 2000 201C; 2005 20' O Meb AM keflw + Maifhd S2dm SjgM: CTddtington 1991. - 58 - FIGURE m.5 - Capital to labor Ratio ProJect wlth and without AIDS 11. . ois 10.U 10.75- loam im nit 194 1997 w0 00 am 2 r3 Nob m erm ld 9 zlz.cn: Cuddington 1991. 3.20 The presence of AIDS reduces the average real GDP growth rate in the 1985-2010 period by between 15 and 28 percent, from 4.0 percent per annum to 2.9 - 3.4 percent per annum (depending on the productivity and saving parameters chosen). Over a 25-year period this decreases potential output by between (1980) TSh 15 billion and TSh 25 billion. The impact on potential per capita GDP is more moderate, decreasing it by only 12 percent in the worst case scenario. Under the simulation, per capita GDP is forecast to grow at an average annual rate of 0.7 percent in the hypothetical situation without AIDS, while with AIDS growth rates range between 0.3 and 0.8 percent per annum. The more plausible outcomes near the upper left of the matrix exhibit the lowest declines in per capita GDP. For the- extreme cases where the negative saving effect due to AIDS is nil and morbidity results in no productivity decline (x=0, z=O), per capita GDP achieves the same level in 2010 of TSh 2,230 (in constant 1980 terms) with or without AIDS. 3.21 In summary, the simulation suggests that the economy wIll grow more slowly with AIDS, with average annual GDP growth over the period from 1985 to 2010 being as much as one percentage point lower than would be the case without AIDS. The effect on per capita GDP will not be as great, as the impact of slower GDP growth is offset somewhat by slower population growth. - 59 - B. IMPACT ON PUBLIC EXPENDITURE AND SERVICES 3.22 The presence of AIDS in any society is likely to impact on public expenditures, but the extent of that impact will depend very much on how those expenditures are distributed across sectors and among different population groups. In a country like Tanzania, which has a history of allocating significant public resources to the social sectors, particularly health and education, the impact is likely to be significant.. In the following sections an attempt is made to outline just what that impact will be. 1. Health Services Impact on Healh Care Delivery 3.23 As mentioned in Chapter II, Tanzanian clinicians interviewed by Pallangyo and Laing (1990) estimate that on average, each adult infected with HIV will experience almost 17 episodes of illness related to this infection before dying from AIDS; each pediatric case will suffer about 6.5 episodes of illness related to HIV infection. This means that people thought to be currently experiencing HIV-related illnesses will have suffered from an estimated 1.4 million episodes of illness before dying. Many of these illnesses will be painful and moderately to severely debilitating. Some will result in permanent impairment of mobility and reduction of productivity even after treatment. 3.24 To put the morbidity burden of HIV infection in perspective, it is helpful to compare it to the burden of illness from other diseases, for example, acute diarrheal disease.9' In the absence of HIV, Tanzanian children have an average of 5.1 episodes of diarrheal disease per year.1 Studies elsewhere in Africa have estimated that diarrheal disease may account for 25 percent of all childhood illness, which would mean that non-HIV infected children can expect to suffer about 20 episodes of illness per year from all other causes besides HIV/AIDS.1-' HIV-related opportunistic infections may increase the burden of morbidity in those children with AIDS by about 32 percent, therefore. 3.25 Population-based data on adult morbidity are not available for Tanzania, but studies in other countries predict that adult diarrhea morbidity rates may be three to five times lower than child rates. This means that in Tanzania, adults may suffer one to two episodes of diarrheal disease per year. One study in Nigeria indicated that diarrhea accounted for 20 percent of all adult illnesses. Applying this proportional morbidity rate to Tanzania suggests IV The authors note that some patients may suffer from multiple illnesses concurrently, which would reduce the number of individual episodes somewhat, and might also reduce costs accordingly. For simplicity, in this report, it is assumed that there are no multiple illnesses suffered concurrently. LI Comparative morbidity statistics cited in this and subsequent paragraphs are from Table I.3, first presented in Chapter I. -'° Tanzanian nationwide study, May 1986, cited in Kirkwood (1989). wl From Table 1.3, the proportional morbidity rate for diarrheal disease (25%) is the median of five estimates from different countries. Total expected episodes of illness per child is calculated as episodes of diarrhea divided by the proportional morbidity rate, or 5.1/0.25 = 20.4 total episodes. A similar total episodes figure is reached using 1986 acute respiratory illness (ARI) data from Burkina Faso. Total episodes = 13.2 episodes ARI divided by 66% proportional morbidity, or 20 episodes. - 60 - that adults may suffer between five and ten episodes of illness per year from all causes, in the absence of HIV infection. Assuming that all 17 episodes of HIV-related illness occur in the last year of life of the AIDS patient, HIV-associated morbidity may account for a 170 to 340 percent increase over baseline morbidity rates in adults with AIDS.w 3.26 Days of Care Needed for AIDS Patients. Table 11.4 shows the total estimated days of care which may be needed for each HIV infected adult to treat the opportunistic illnesses associated with AIDS.2 Pallangyo and Laing have again relied on expert opinion from Tanzanian clinicians to develop these care-seeking patterns, although in the absence of population-based data on household health care demand behaviors it is possible that utilization of modem health care facilities has been overestimated somewhat.14 The table assumes that the existing health care referral system is in place, i.e. that AIDS patients, once recognized, tend to be referred frequently to secondary and tertiary facilities for treatment. Under these assumptions, an estimated 286 days of care may be needed per case. 20% of which could be hospital days (both inpatient and outpatient care). Over 60% of care will probably be provided at home, however, by family members who must adjust their own schedules accordingly. '2Of course, not all HIV-related illnesses will occur in the last year of life of the AIDS patient, but since the distribution of episodes of illness over the lifetime of an HIV infected person is as yet unknown, and empirical evidence shows most AIDS patients are very ill in the last year of life, this assumption may not be too far from the truth. Because of the fast progression from HIV infection to AIDS in children, the assumption is most plausible when applied to this age group. 3! Some assumptions regarding the objectives of case management for terminal illnesses are implicit in these estimates of days of care required for treatment of HIV infected patients. For a fuller discussion of this issue, see Chapter V. ly For example, data from a household health care demand study in Zaire indicated that only 57% of respondents who were ill went to a health facility to seek care. About two-thirds of these people chose public providers, while the rest went to private dispensaries (not officially- recognized providers). Reported traditional care was negligible, although this might have been a problem with survey implementation rather than a reflection of actual use of traditional providers. (Bitran 1989) - 61 - , TABLE m.4 - Days of Care, by Episode and Treatment Settilng Adult A S Case. -No. Total Totat Days Ipf. Hoop Nowe Care _o__ _t Prfm Home Total per Care Care per HospWt X Home oIa Care Care Esse Dav -R 6aV8 ao a Chronic 7 0 10 17.0 2.40 16.8 24.0 40.8 41X 59X iarrhoea Orat Thrush & 1 3 7 11.0 10.00 10.0 70.0 110.0 9X 64X Cand. - Pneumonia 9 2 5 16.0 0.50 4.5 2.5 8.0 56% 31X Tuberculosfs 45 30 ISO 2S5.0 0.25 11.3 45.0 63.8 18X _71 Severe Headache is_ 3 30 48.0 015 2.3 4.5 7.2 31X 63X Yfrat sygn Inf. 1 4 30 _5. 0.10 0.1 3.0 3.5 3% 86% BeFt. Skfn Inf. 1 4 5 1C.0 1.05 1.1 5.3 10.5 10X 50X Eczema 3 6 _ 10 19.0 0.45 1.4 4.5 8.6 16X S3X Pruriao _ 0 10 15.0 1.05 0.0 10.5 15.8 0X 67X KaaOSf Sarcaoa 6 O 10 16.0 0.02 0.1 .2j 0.3 38X 63X Neurot. Disease is Q. is .j,0 0.20. 3.0.9.L0L12.0j. 252 .....fL.. Seoticaemfa a 0 0 8.0 0.70 5.6 0.0 5.6 1002 02 Total 16.87-I 56.0 178.5 286.0 202 62X Sourc: Pallangyo and Laing, 1990, Appendix IA. NIote: Hospital refers to patients treated in reference, regional or district hospitals. Primary care refers to rural health center or dispensary treatment. Home care is care given at home by household members. No distinction is made between patients treated as outpatients and those treated as inpatients. 3.27 Pallangyo and Laing estimate that approximately 200 days of care may be required for pediatric AIDS cases, as shown in Table m.s. Again, the majority of care will be delivered at home. -62 - TA1ELE m.5 - PAy of Car.e. by andrmt Pediatric AIS Cas Illness Days of Care Per Episode** Total l . _ ^ No. TotaL Total Days H08fit Prim Home Totat ~~~~~~~~~~~~~~~~~~~~~~~~~~Epi. Hasp Homie Care i Hospit Prim Home Total Epr Care Care per Hos t Home a CaFe Care Days Davs Case a Chronic S 2 5 12.0 1.00 5.0 5.0 12.0 42X 42Z Diarrhoea - - - - - Oral Thrush & 3 4 5 12.0 2.00 6.0 10.0 24.0 25X 42X Cand. Pneumonia 7 4 - ° 1 ! 1.00 70.0 11. 0 6 Skin Diseases 3 3 . 1.40 4.2 .4.2 12.6 33X Recurrent Fever 7 2 5 _14.0 0.75 5.3 3.8 10.5 50X 36X Tuberculosis 30 11 330 371.0 0.35 10.5 -115.5 129.9 8X -aQx-2 - Total 6.5 38.0 138.5 200.0 19% 69X Source: Pallangyo and Laing 1990, Appendix IA. NMt: Hospital refers to patients treated in reference, regional or district hospitals. Primaiy care refers to rural health center (RHC) or dispensary treatment. Home care is care given at home by household members. No distinction is made between patients treated as outpatients and those treated as inpatients. 3.28 It is important to remember that most AIDS cases will not be diagnosed as such, either because patients will seek care infrequently or from many providers, or because of technical or training problems in applying diagnostic criteria adequately.j' Nevertheless, many AIDS cases will present at health facilities throughout the country for treatmnent of opportunistic infections associated with AIDS, especially chronic diarrheal disease, thrush, pneumonia, tuberculosis, skin conditions and generalized pain. Consuming pharmaceuticals and nursing care, these undiagnosed and diagnosed cases will create an important burden on the health care system. Even if care is sought for only 20% of episodes of illness per year this results in an incremental burden to Tnzania's health care system of an estimated 72S500 to 325,000 episodes of illness, many of which will require expensive inpatient care. Ways to minimize the cost of this care are analyzed further in Chapter V. 3.29 Tuberculosis 1-t. Another important impact of AIDS on the health care system is an expected increase in tuberculosis cases due to the interaction between AIDS and tuberculosis. Three factors are important in defining this effect: 1) increased risk of TB among persons infected with HIV; 2) increased rate of progression to overt TB in persons infected with HIV; and 3) increased risk of TB infection to the population at large. In addition to increasing the number of cases of TB, AIDS will affect tuberculosis control programs by requiring additional health worker safety precautions and modified treatment protocols for treating TB in HIV infected people. 1' Pallangyo and Laing 1990 w' This discussion is drawn from the paper by Slutkin, Gary, J. Leowski and J. Mann. 'The Effects of the AIDS Epidemic on the Tuberculosis Problem and Tuberculosis Programmes," pp. 21-25 in Fleming, et. al. (eds.) The Global Impact of AIDS. New York: Alan R. Liss, Inc. 1988. -63 - 3.30 Data from several countries show that ljt.t of HIV Inftion In la natimt are 3 to 20 times the rates of HIV in the general npulation. For example, in Zaire, TB patients were seven times more likely to be HIV infected than a control grwup matched for sex, age and residence. Evidence from a cohort study in New York City cited by Slutkin et. al. (¶988) further documents this increased risk. 3.31 Researchers have known for some time that immunosuppression worsens the course of tuberculosis Infection. It is therefore not surprising to find evidence of an increased rate of progression from asymptomatic to overt tuberculosis for persons co-infected wCtA HIV. This wil probably result in an increasing number of cases of TB in coming years, especially in regions with high HIV prevalence rates. 3.32 Tanzania is one of two countries cited by Slutkin, et. al. as having a national tuberculosis reporting system. By 1982, tree years after beginning the national TB reporting system, Tanzania had complete reporting. At first, with complete reporting the system showed evidence of a "flattening of the curve," indicating a relatively constant rate of new TB cases in the country. Since 1984, however, the number of new TB cases being reported annually has been increasing steadily (See F ,ure m.6). Initially, there was some thought that the availability of a newly introduced short course chemotherapy might havf influenced more individuals with TB to seek treatment, but the continued rise in dhe rate at which TB is spreading suggests a causal link with the HIV epidemic. FIGURE m.6c- New Tuberculosis C_ es In TanzanIa. l11l NEW CASES (Thousands) ACTUAL CAM 10 _. . .. _ . _ -.0- - .... __§ . CA WNO VW 5~~~~.* -- ... 1o8s 1968 1963 1984 10" l" 1907 l98 1989 Sur=: Laing and Pallangyo, 1990. 3.33 In Tanzania, the annual risk of infection with the organism which causes TB ("M. tuberculosis") was 1.2 percent in 1987; for Sub-Saharan Africa as a whole, the risk is 1-3 percent. TB case rates are normally much lower than infection rates, given the body's normal ability to combat this infection. Although in some countries TB case rates may be as high as 100 to 300 per 100,000 population, in Tanzania the rat of smear-positive pulmonary TaB was estimated at 6Q per 100.000 population in 1987. ly Slutkin, et. al. note that HIV co-infected Individuals often have more non-pulmonary presentations, and more serious presentations, than TB patients who are not co-infected with HIV. This suggests that both the risk of TB infection and the risk of developing serious presentations of the disease are likely to rise in Tanzania as a result of AIDS. ly World Bank 1989. -64 - 3.34 Tuberculosis Program Effects. Health workers in tuberculosis programs will have more HIV infected patients than other health workers. As needles are used in testing for TB and in treatment, special precautions will be needed to ensure occupational safety. Also of concern are clinical protocols for diagnosing and treating TB in HIV co-irfected individuals. Slutkin et. al. note that individuals who are HIV positive are often allergic to TB skin testing. Clinical and radiographic pattems of TB are different in HIV co-infected individuals, as well, which may pose problems for diagnosis. A third concern is the frequency with which HIV co-infected individuals experience relapses when provided with conventional anti-TB therapy. Modifications in dosages or provision of different drugs altogether may be needed to adequately treat TB in AIDS patients. Finally, pediatric AIDS patients often have a reaction to the BCG exam, so it is not recommended for HIV co-infected persons. 3.35 Slutkin and his co-authors call for better surveillance and more research to define the magnitude of the TB problem associated with AIDS, and to forecast the fiuture impact of the expected increased TB caseload. Cost of AIDS Care 3.36 On the basis of current treatment protocols and referral patterns, Pallangyo and Laing (1990) estimate that the total. p~er case. lifetime treatment cost for AIDS patients in Tanzania is TSh 55.917 ($290) for adults and TSh 37.651 ($195) for children. If the health care system is experiencing a shortage of drugs, as frequently happens, the average cost per case is somewhat reduced. If only 60% of the recommended drugs are available for treatment, the total, per case, cost is TSh 50,158 ($260) for adults, and TSh 34,412 ($178) for children. 3.37 These average costs should not obscure the fact that great variation may exist. Over, et. al. note that direct cost per patient may vary: "depending not only on the particular clinical symptoms with which HIV infection manifests itself, but also on socioeconomic characteristics of the health care options available to that patient." (Over, et. al. 1988, p. 124) In other words, poor people living in rural areas with limited access to modern health care may incur lower health care costs (and perhaps die sooner) than higher-income populations in urban areas %!ith access to tertiary level facilities. It should also be noted that these estimates of direct costs of AIDS do not include the cost to patients for transport to seek treatment or the cost of care (often drug therapy) foregone by patients with other illnesses who may have been crowded out of the system by AIDS patients.L' This is an issue for further investigation. 3.38 Although precise estimates are not available, AIDS treatment costs in Tanzania probably do not exceed per patient treatment costs of other illnesses. Data from Kenya show that the average cost of hospitalization for a pneumonia patient at a major teaching hospital in 1986 was about 2,700 KSh ($166, at 16.23 KSh to one USD). This cost included nursing (average length of stay of 5.25 days), medicines, laboratory and radiology costs, but not other medical procedures.2 IF Over, et. al. 1988, p. 125. 192 At the teaching hospital in question, pneumonia was the third most frequent cause of hospitalization. Other frequent causes of hospitalization, along with their cost and average length of stay (ALOS) were malaria (fourth most frequent cause, 1,580 KSh ($97); 3.87 ALOS), normal delivery (first ranking cause, 1,323 KSh ($82); 2.87 ALOS), and abdominal surgery (second most common surgical procedure, 5,393 KSh ($332); 8.70 ALOS). - 65 - 3.39 The total expenditure required to traat AIDS patients obviousiy depends upon the rapidity with which the virus spreads, and the number of HIV-seropositive people at any given time whose immune systems have deteriorated to such a state that they can no longer fight off opportunistic illnesses. As indicated in Chapter 11, the actual number of living AIDS cases at any time cannot be known with certainty, either because they have not presented themselves to a health facility or because they are not recognized as having AIDS when they do so. All of the opportunistic illnesses are quite common in Tanzania, and might not be recognized as AIDS-related unless they are particularly virulent, or recurrent, or non- responsive to treatment. 3.40 Table II.6 presents the potential public expenditure on AIDS cases in 1991, taking the NACP's estimate of persons currently living with AIDS; namely, about 69,800 adults and about 36,000 children. Two cost scenarios are presented, the first assuming 100 percent drug availability and the second, only 60 percent drug availability. TABLE 111.6 - Estimated Costs of AIDS Treatment. 1991 (US$) 100% Drua Availabilit 0% Drua Avaitabiit Per Case TotatI O'U§Pr Case ToMa C'0) AduLts 290 20,242 260 18,148 Children 195 7 C-0 178 6 408 'total 2Z626 2 3.41 In FY90/91, the Government allocated 10.9 percent of its recurrent budget to health, or a total of about TSh 11.25 billion ($58.29 million). If the above estimates of patient numbers and treatment costs are accurate, AIDS treatment could be absorbing as much as 40- 50 percent of the Government's recurrent budget for health. While many of the AIDS patients will seek treatment from private practitioners, some of whom do not receive subventions from Government, it is clear that AIDS cases are already placing an enormous burden on public finances. If only half of all AIDS cases are seeking treatment from publicly-financed health facilities, AIDS expenditures currently absorb about one-quarter of the Government's recurrent health budget. 3.42 One can get a rough approximation of the likely costs of treatment for future AIDS cases by multiplying the projected cases under Bulatao's scenarios by the treatment costs estimated by Laing and Pallangyo, assuming full drug availability. These are summarized in Table 111.7 below, assuming no resource constraint. TABLE Im.7 - Projected Expenditure for Treatment of AIDS Cases No Resource Constraint Year 45X Monogamy Scenario 15% Monogamy Scnar5io Cases u Cost cuss m nM cases VuYuu ost c mtnx 1995 26/10 9,490 90/40 33,900 2000 52/21 19,175 164/85 64,135 2005 83/37 31,285 247/134 97.760 2010 119/55 45,235 332/176 130,600 2015 160/75 61,025 400/200 155,000 Source: Cases, adults/children, taken from Bulatao Costs derived from Pallangyo and Laing. - 66- 3.43 The projected costs associated with treatment of AIDS cases, assuming current referral patterns, are likely to quickly outdistance financing expected to be available from the public sector. Figures M.7, m.8 and m.9 below illustrate the divergence between expenditure "needs" for AIDS treatment as currently defined, and projected recurrent expenditures for health from Government sources under various reform scenarios described in the recent economic report for Tanzania.~& As is evident, the resources available to finance recurrent health expenditures differ significantly from one scenario to another, and only under Scenario 3 is there any hope of sufficient resources being available to meet a substantial portion of these needs, without diverting funds from other (arguably more important) health activities. Under that optimistic "reform" scenario, GDP grows at 5 percent annually, and 14.6 percent of the Government's recurrent budget is allocated to health. FIGURE I11.7: Prolected AIDS Treatment Costs Versus Available Public Fundina (Scenario 1) TSh billion 40- AIDS treatment costs (15X monogay) 35- 30- 25- 20- T otal recurrent health budget 15- _ fAlpS treatment costs '0 oX~~~~~~~~~~~(5 monogamy) 0- 1990 1995 2000 2005 2010 2015 2020 Y E A R FIGURE 1II.8 - Proected AIIDS Treatment Costs Versus vailable Public Fndin (Scenario 2) TSh btllton 50- Total recurrent health budget ' - 6 325° AIDS treatment costs (15% monogamy) 30- 25- 20- 5- ~~~~~~~~~~AIDS treatment costs (45K monogamyf) 1990 1995 2000 2005 2010 2015 2020 ; Tanzania Economic Reror: Towards Sustainable Develonment in the 1990s, World Bank, June 11, 1991. - 67 - FIGURE m.9 - Projected AIDS Treatment Costs Versus Available Public Funding (Scenario 3) TSh billion 80- 75- Total recurrent health budget 70- 65- // 60- / 55- / 50-/ 45- AIDS treatment costs 40- (15X monogamy) 20- / ~ AIDS treatment costs 1-5o / / < ~~~~~~~~~~(45X monogamy) 0- 1990 1995 2000 2005 2010 2015 2020 Y E A R 3.44 A rough analysis of the Government's recurrent health budget suggests that about half of such resources are currently being channeled to the nation's hospitals (district, regional and referral hospitals, including private hospitals subsidized by Government). Since the main threat of illness and death for the vast majority of the Tanzanian population will come from causes other than AIDS, most of them preventable, the treatment of AIDS cases must not be allowed to divert resources from Tanzania's "preventive" programs, such as maternal and child health care, the expanded program of immunization, diarrheal disease control, family planning, malaria control, etc., which are - for the most part - handled by primary facilities. For this reason, the Government may wish to ensure that no more than 50 percent of its recurrent health budget is available for hospital services in the future, despite the increased pressure on these facilities due to AIDS. The proportion of funds allocated to hospitals which is absorbed by AIDS patients and the proportion which remains available to treat illness and injuries from causes other than AIDS (Table 1. 1) will depend on treatment protocols adopted and how patients are managed by the health care system. Chapter IV explores options for decreasing the cost of AIDS treatment. 2. Education Services 3.45 AIDS can affect education in Tanzania in a number of ways: through an impact on cohort sizes and enrollment numbers; teacher needs, turnover and training; parent's willingness and ability to pay for schooling; the efficiency of the education system; and the economic returns to education. This section will also discuss the implications for education budgetary requirements. - 68 - Impact on Cohort Size 3.46 Table I11.8 below shows the projected numbers for the primary and secondary school-age cohorts, without AIDS, and for the two projection scenarios used elsewhere in this study (45% and 15% monogamy). There is very little demographic impact by 2000, but by 2020, the presence of AIDS in the population is likely to reduce significantly the number of school-age children as compared to the "no AIDS" scenario; in the worst-case scenario, at the primary level there would be 22 percent fewer children to be educated, and at the secondary level, the relevant age groups would be reduced by about 14 percent. TABLE HI.8 - Prolections of School-Age Cohorts ('000) No AIDS 45% Monoaamy 15% Monoasm Primary School Age 7-14 1990 5,298 5,296 5,288 2000 7,677 7,614 7,410 2010 10,685 10,351 9,399 2020 13,323 12,436 10,420 Secondary School Age 15-19 1990 2,513 2,513 2,513 2000 3,679 3,677 3,667 2010 5.289 5,228 5,033 2020 7,199 6,925 6,161 3.47 At present (1990/91 budget), recurrent expenditure from Treasury (through the Ministry of Regional Administration and Local Government) is approximately TSh 2,35iO (roughly $11.75) per primary school pupil. There are 3.6 million primary school pupils, implying an enrollment rate of about 68 percent of the primary age cohort.21' In nominal terms, in order to maintain the present level of provision of primary education (with places for 68 percent of the cohort), the budget for the primary level would just about need to double by 2020 under the 15% scenario instead of increasing to 2.5 times the 1990/91 amount in the hypothetical case without AIDS - a difference of about 4,650 million shillings. This should not be seen as a source of fiscal relief, however, for two reasons. First, education is seriously underfunded at present, especially at the primary level, even taking into account the additional revenues contributed by parents. Slower enrollment growth without any reduction in budget grodih would allow much needed improvements in educational quality, e.g. through better provision of textbooks and other teaching materials, in-service training and teacher support, and would make it easier to reach the Government's goal of universal primary enrollment. Second, the fact that-the populationas a whole is growing more slowly and getting younger (the age structure effect discussed in Section A) decreases the number of productive members of society who must support education of the young. It may therefore prove difficult for the economy to generate sufficient funds to maintain even the current level of investment in education. In fact, while the absolut expenditure on primary education will be reduced under the "with-AIDS" scenarios compared to that without AIDS, per caita expenditures will remain largely unchanged. 3.48 As regards the secondary level, even with the recent upsurge of community schools, at present there are places in Form I for only 15 percent of pupils who complete primary school, and less than 5 percent of the relevant age group. Even a large reduction in the cohort size would do little to ease the enormous pressure for secondary school places. Li f Tanzania, Teachers and the Financing of Education, World Bank, May 10, 1991. - 69 - Teacher Numbers. Turnover, and Traning 3.49 At present there are about 99,000 primary school teachers country-wide. Lower numbers of school-aged children compared to the 'no AIDS" projections imply that fewer teachers will be needed to maintain the country average pupil:teacher ratio of 34:1. Table 11I.9 below sets out estimated teacher numbers needed in 2000, 2010 and 2020. Under the 15% monogamy scenario, compared to the hypothetical situation without AIDS, 25,700 fewer primary school teachers would be needed in 2010 and 58,100 fewer in 2020. However, this will be partially offset by increased attrition from the teaching force as HIV-positive teachers fall ill and die. If it is assumed that the rate of HIV infection among teachers is the same as among the general adult population, and that teachers will constitute 0.39 percent of all those who die each year of AIDS, mirroring the percentage of the total population who are teachers, then by 2010, 14,460 teachers will have died from AIDS, with the number mounting to 27,000 by 2020. The average cost per year of teacher training is TSh 140,000.2' To train 27,000 additional teachers for two years each to replace those who die would cost TSh 7,560 million in recurrent costs alone. 3.50 Another possible effect of the AIDS epidemic on education could be the inability to attract teachers to those areas most severely affected. This has not proven to be the case, to date. For economic and cultural reasons not related to AIDS, the Kagera and Dar es Salaam Regions, which have experienced the most AIDS deaths, have among the lowest pupil:teacher ratios in the country (i.e., appear to be able to attract teachers more readily than other regions). Impact on Quality and Efficiency of Education System 3.51 Illness and deaths of teachers and students from AIDS will have a deleterious effect on the quality of education and the efficiency of the system, where efficiency can be defined as the educational outcome that is achieved for the money spent. The loss of experienced teachers cannot be quantified, and may need to be compensated for by special measures, like better teacher guides, more in-service support and training, and, a system that has been tried elsewhere, "mentor" teachers who are paid extra to help newly-trained teachers. Illness of teachers will reduce teaching time, while absenteeism and drop-out rates among students are likely to increase because of illness or because students are needed at home to help look after sick family members. Given traditional roles, this is likely to affect girls particularly. TABLE m.9 - Prolected Teacher Needs for 34:1 Pupil:Teacher Ratio (Rounded to nearest 100) Primary School Teachers Needed 200R (68X enrollment of 7-14 age cohort) - no AIDS 153,500 213,700 266,500 - 452 monogamy 152,300 207,000 248,700 - 15X monogamy 148,200 188,000 208.400 Lost Investments in Education 3.52 Investments in education by the Government and famlies are considerable. Per year, it costs the taxpayer TSh 2,350 for every child in primary school, TSh 47,000 for every student in govermment secondary schools, TSh 140,000 for each trainee teacher, and TSh 800,000 for each university student (1990/91 budget figures). At these expenditure levels, aZ ' Based on data on actual expenditures and student numbers in 42 Teacher Training Colleges in 1990/91 collected from the Ministry of Education and Culture by the Education Sector Report team. - 70 - more than TSh 2.6 million has been invested in someone who has spent 8 years in primary school, 4 years in secondary school, and 3 (or more) years at the university, and this is without taking into account the costs of uniforms, fees, etc. borne by the family, or of the opportunity cost in foregone earnings of the years spent in secondary and tertiary education. Thus the early death of a well-educated person is not only a terrible loss to family and friends, but an expensive loss of investment in education for society at large. A Lower Demand for Education? 3.53 Not only is the internal efficiency of the educational system likely to suffer, raising the cost of educating each graduate, but AIDS deaths mean that large investments in education are lost. This Is likely to lower the returns to education realized by graduates, their families, employers and communities. Lower private returns to educational outlays may reduce the demand for education. Families may be less prepared to release children from household and farm chores to attend school, or to allocate household expenditures to education rather than other items. Willingness and ability to pay for education may decline if the quality of schooling deteriorates, and illness and death make additional claims on a reduced household income.v To the extent that this affects girls particularly, this could threaten the gains in family nutrition and health and lower fertility that are associated with higher female educational levels. 3. Impacts on Other Sectors 3.54 While the impact of AIDS on the health sector is undeniable, its impact on other sectors is less clear. The slower growth rate of the population under a "with-AIDS" scenario suggests that expenditures in other sectors may not need to rise as rapidly as would otherwise be the case. To the extent, for example, that Government expenditures in agriculture, in transport and communications, and in energy are linked with population growth - as they are in education - then absolute expenditures in these sectors may grow less rapidly than in an AIDS-free economy, all other things being equal. However, since the economy itself will be growing less rapidly with AIDS, as discussed above, this diminished expenditurs demand will not offer fiscal relief, as such. 3.55 What is clear is that all sectors will lose substantial numbers of trained people to the epidemic. How badly any given sector is affected will depend upon how easily those people can be replaced. For example, HIV infection is spreading very rapidly among truck drivers on the major transport routes, suggesting that this segment of the work force will lose a disproportionately large number of people. Replacements, however, should not prove too difficult. On the other hand, other sectors - such as energy and health, for example - already employ the relatively few highly-skilled and experienced people in the country with the specialized skills on which the sectors depend. The loss of only some of these people could effect the efficiency and productivity of these sectors; replacing them could take many years, suggesting that an even greater effort will be needed in building Tanzania's human resource capacity. V A small study of 23 households in Mexico in which someone had AIDS found that over 50% of household expenditures were on the AIDS patient and that household income decreased by 18% on average. - 71 - C. IMPACT ON HOUSEHOLDS AND COMMUNITIES 3.56 Economic literature suggests that deaths, particularly adult deaths, impose numerous economic impacts at the household level.X As illustrated in Table III. 10, these impacts can be classified by their sphere of economic influence and by the time they are felt by the household in relation to the time of death. Such impacts occur over an extensive period, beginning at the onset of the illness and continuing until well after an individual's death. They are felt not only by the household of the person dying but also by other households in the community. For this reason, households experience some impacts even before, or without, an illness and death of their own members. 3.57 BecausF illness and death are recognized as an inevitable part of the human condition entailing uoth psychic and monetary costs, households take certain precautionary actions in an attempt to minimize the negative consequences of these future events. These precautions include, for example, investing in preventive health care, setting aside a portion of income to meet possible future needs (precautionary saving) and/or purchasing insurance. They may also assist other households currently experiencing the loss of a loved one, with the expectation that this support will be reciprocated in their own time of need. These precautions affect current consumption and investment of the household. But expectation of future losses of family members also affect decisions with regard to family size (number of children, adoption of extended family members, etc.) and how or where the household supports itself (where the family resides, how many members work, etc.). 3.58 The impacts are, of course, more acute, once the household is itself experiencing illness of one or more of its members. If the ill person is a working adult, his or her labor productivity may be reduced and others may have to take on a larger workload if household income is not to fall. A sick mother may mean that children are not as well cared for as normal, with negative consequences for their health and well being. A sick child also impacts on the household, making additional demands on productive family members, with the possibility that overall output is reduced. Household consumption and investment are affected to the extent that treatment and care costs cannot be met from current savings. These changes, if extreme, can affect the long-term welfare of the household. For example, household assets may be sold, leaving the family without the means to support itself in the future. These negative impacts may, of course, be moderated by transfers from outside the household; that is, from extended family, friends, or the community at large. LI The following section draws heavily from Over, M., M. Ainsworth, P. Mujinja, G. Koda, G. Lwihula and I. Semali. "The Economic Impact of Fatal Adult Illness in Sub-Saharan Africa: A Research Proposal," World Bank and University of Dar es Salaam, November 1989. - 72 - TABLE 111.10 - The Economic Immact of Adult FAtal Illness on the Housghold T y p o f _ _ _ _ _ _ _ _ _ _ __ _ _ _ _ _ Economic Inpact - o (A) C ) tC) (D) Prior to an Dur ng an Inuediatol y Long-torm Iltness Illness after death Impact of death (1) Organization of Reduced pro- Lost output of Reatlocation of Production and economic activity ductivity of deceased membor land and labor Earnings affected person Choice of area of Reallocation of _______ _ _ _________ residence labor - _-_ __ (2) Insurance Medical cost of Funeral costs Changes In typ Consumption and treatment and mount of Investment Preventive health Receipt of consurption and care Dissavings transfers investment Precautionary Changes In Payment of legal savings consumption and fees investment Transfers to other households Receipt of transfers _ (3) Household size Change in Loss of nmbeor Dissolutfon or Household Health allocation of time reconstftution and Composition Fertility to health of household maintenance Migration Poor health of l _ - ~~~~~~~~~~~~~~~~~~~survivor (4) Disutilfty of ill Grief of I Psvchic Costs indivfduat survivor_ _ _ Source: M. Over, M. Ainsworth, P. Mujinja, G. Koda, G. Lwihula, I. Semali, "The Economic Impact of Fatal Adult Illness in Sub-Saharan Africa: A Research Proposal", World Bank and University of Dar es Salaam, Table 4, p. 15, November 1989 3.59 The impacts of the death itself are also important. The primary economic impact, if an adult has died, is the permanent loss of output or income which that person contributed to the household. If that person was the primary or sole income earner, the very survival of other family members may be in jeopardy. If that person was the primary caregiver - for example, the mother - the welfare of other members, particularly of young children, may be severely affected. Throughout the developing world, the loss of a mother has been found to lessen a child's chance of survival. The psychic costs of losing a parent or spouse are, of course, enormous, and can have economic implications if the loss permanently diminishes the ability of other family members to cope. Although a one-time cost, the expenses associated with burial of the deceased are substantial in Tanzania, and may further erode the household resources. 3.60 The impact of a death on a household over the longer term can be moderate or severe, depending upon the resources available to it. If savings have not been depleted, if sufficient able-bodied adults remain to maintain household output, if past investments have been possible in human and other capital, and if the emotiovsl health of remaining family members is intact, the household may not face a permanent diminishment of its earning capacity or overall welfare. However, if savings are gone and/or equity reduced, if dependents far outnumber healthy adults, if the family was living on the edge of survival to begin with, then there is likely to be a permanent reduction in the household's income and welfare. Less able-bodied labor within the household may mean less food grown, or a switch to less labor-intensive production, which is likely to make it more difficult to meet the -73 - nutritional needs of the family. Children may be pulled out of school to help with farming or household chores. Families may be broken up. 3.61 Although little hard data is currently available on the extent to which AIDS has permanent.y decreased the welfare and productivity of affected households, anecdotal evidence suggests that all of the above impacts have been felt by Tanzanian households experiencing AIDS, and the communities in which these households are found. The traditional coping mechanisms have, of course, gone a long way to lessen the impact of AIDS deaths, with extended family and community members assisting to meet the costs of care and burial of AIDS victims, taking in orphans, and providing emotional and material support to those who need it. However, because of the stigma attached to this disease, the support has not always been as great as it might have otherwise been, with women being particularly vulnerable because of their relatively weak economic and legal status. Women's limited legal rights under traditional inheritance laws make her particularly vulnerable if widowed because of AIDS. Her in-laws, on whose largesse her survival may depend, may blame her for her husband's death, and remarriage may be difficult if she is believed to be infected. 3.62 There is some evidence that traditional coping mechanisms are breaking down in areas where the AIDS epidemic has been particularly severe. All able-bodied adults in some families have died, with some households comprised only of grandparents and grandchildren. Non-governmental agencies are assisting government to provide support for severely-affected households, but finding affordable and equitable interventions which can actually make a difference has proved difficult. Research is currently underway in Tanzania which should shed additional light on the extent to which AIDS deaths are affecting household production and consumption, and whether the impacts of AIDS deaths differs significantly from those of other adult deaths. - 74 " IV. PREVENTING FUTURE HV INFECTION 4.01 As discussed in Chapter II, there are four principal routes of transmission of HIV in Tanzania: heterosexual, perinatal (mother t3 child), blood transfusions, and contaminated health care equipment. Heterosexual transmission of HIV, the route of the vast majority of HIV infections in Tanzania, can be diminished by: (i) reducing the number of sexual partners of sexually active people (and thereby reducing the probability of contact between HIV- infected and non-infected persons) through effective IEC; and (ii) reducing the likelihood that sexual contact between HIV-infected and non-infected persons will result in transmission, by - tor example - reducing other STDs (which heighten susceptibility to HIV infection and infectiousness of HIV-infected persons) or promoting the use of condoms. Perinatal 1ansmisiekl Is probably a consequence of heterosexual transmission and !'ierefore these approaches will also be most effective in decreasing such transmission. The only other option for reducing perinatal transmission is through screening and counselling of HIV-infected women, both those who are already pregnant and those of child-bearing age who could become pregnant. Transmission caused by blood transfusion can be prevented either by reducing blood transfusions or by screening blood for HIV antibodies prior to transfusion. Frannsmission from contaminated needles may be reduced through increased use of disposable needles, proper sterilization, use of protective clothing, and education of users and providers. 4.02 There are numerous ways to implement the options described above, by using different comununication channels, messages, types of facilities or mixes of personnel and targeting strategies. But while the options are numerous, the resources for program implementation are limited. Decision makers need to analyze and compare options to select a package of programs which will allow Tanzania to realize maximum prevention benefits within existing budget constraints. This part of the report presents a general framework for analyzing alternative prevention strategies, focusing mainly on the quantification of costs and effects. Five },;es of strategies are then discussed and analyzed in detail: STD prevention and control, condom promotion, information, education and communication strategies (IEC), blood screening, and reduction of blood transfusions. Possible development of a vaccine is also discussed. The discussion shows clearly that certain prevertion strategies are more cost- effective than others, and makes the case for focusing limited resoure's on these more cost- effective strategies. A. COSTS AND EFFECTS OF SELECTED INTERVENTIONS 1. STD Prevention and Control 4.03 Conventional sexually transmitted diseases, including syphilis, chancroid, trichomoniasis and chlamydia, increase the probability of transmission of HIV because the genital ulcers associated with them facilitate transmission of the virus.1Y The existence of genital ulcers increases both the infectiousnesv of the HIV-infected individual, and the susceptibility of their contacts to HIV. The presence of another STD in either partner can increase the probability of sexual transmission 3 to 10 times as described previously in 1' Bjorkman 1991. - 75 - Chapter II of this reportY In Zaire, even non-ulcerative cases of gonorrhea and chiamydia have been shown to enhance HIV transmission probability.2' 4.04 The prevalence of STDs presenting with genital ulcers appears to be higher in Africa than in North America and Europe. For this reason, some experts believe that STDs may be the key factor in explaining why heterosexual transmission of HIV is so much more frequent in Africa, compared with more industrialized countries.1' 4.05 Treatment of STDs and the careful management of STD patients as part of a comprehensive STD prevention and control program is an important AIDS prevention strategy, one which also provides significant other health benefits, as discussed below. An effective STD prevention and control program often includes the promotion of condoms. Effectiveness of STD Treatment 4.06 Health benefits from STD treatment accrue to the STD patients themselves and to their future partners. Benefits accruing to STD patients are called "static" benefits. Benefits gained by avoiding harm to future partners are called "dynamic" benefits, or "secondary prevention" benefits. 4.07 Both static and dynamic benefits include the health benefits of treatment or prevention of the classic STDs themselves (which can cause chronic disablement or death in a small percentage of cases), as well as the healthy life-years saved by avoiding cases of HIV infection. HIV infection is averted because the persons treated for classic STDs no longer have higher susceptibility to HIV infection. In addition, STD patients who are already infected with HIV are less likely to infect their partners once their genital ulcers have been healed through treatment. 1Piot and Over (1990) estimate a five- to ten-fold increase in risk, while Bjorkman uses a more conservative estimate of a three-fold ircrease. Z Laga, M. (1990) cited in Over, M. (February, 1991). 4' Piot and Over, op. cite., p. 7. ' Chin 1989. - 76 - 4.08 The different types of benefits from STD treatment are the following: BENEFITS OF STD TREATMENT'V Static Benefits 1. Benefits to HIV-positive STD patient o Life years gained by treating classic STD itself 2. Benefits to HIV-negative STD patient o Life years gained by treating classic STD itself o Life years gained by avoiding case of H:IV infection in STD patient through treatment of classic STD (reduces susceptibility to HIV infection) Dynamic Benerits 1. Benefits to HIV-positive partners of STD patients 9 Life years gained by avoiding case of classic STD 2. Benefits to HIV-negative partners of STD patients o Life years gained by avoiding case of classic STD o Life years gained by avoiding case of HIV infection by treatment of classic STD in partner who is HIV positive (reduces partner's infectivity) or by avoiding the partner's HIV infection in the first place through STD treatment (see #2 Static Benefit). 4.09 Estimation of benefits is further complicated by the variable risks of infection faced by different groups of STD patients, depending on behavioral factors such as frequency of sexual contact and rate of sexual partner change. Calculation of the effectiveness of STD treatment programs is therefore difficult. 4.10 Static Benefits. Bjorkman has developed a model which estimates the static benefits of STD treatment, in terms of cases of HIV infection avoided. The model assumes that about 70% of STD patients are males (with an average of 25 prostitute contacts per year); 10% are female prostitutes (with an average of 1,200 male contacts per year, or 48 times the sexually activity of male STD patients); and 20% are non-prostitute women (mostly wives of male STD patients). Bjorkman's calculations include other assumptions based on available Tanzanian data (e.g. prevalence rates by type of STD patient, efficiency of antibiotic treatment, etc.). He concludes that 14 STD treatments may be required to prevent one HIV transmission. Total benefits are underestimated, since the dynamic benefits of STD treatment (i.e. benefits of cases avoided in partners of STD patients) have not been included in the model.2" 0' Derived from Over February 1991. Z' Bjorkman 1991, p. 20. - 77 - 4.11 Over estimates the benefits of STD treatment in terms of years of healthy life gained (see Figure IV. 1). Over shows that averting one case of HIV infection saves almost 20 years of healthy life (discounted at 3 percent), ranking fourth among the twenty diseases considered. When weighted by productivity before discounting, averting HIV infection still saves 16 years of healthy life, rising from fourth to first place. Static benefits of treating the classic STDs themselves are also included in Figure IV. 1: these benefits range from 3 years of productive healthy life saved (syphilis) to less than one year (chlamydia, gonorrhea and chancroid). FIGURE IV.1 - Static Benefit of Preventing a Case of STD and Other Diseases In Sub-Saharan Africa Sickle Cell ------- Tetanus (Neo.) Birth Injury HIV Severe Malnut. Cirrhosis -- ---- Pneum. (Child s) Congenital malf. Cerebrovas. Tuberculosis SYPHILIS Measles Malaria Prematurity Injuries Pneum. (Adult) GastroenterIt is CHLAMYD IA, F CHLAMYDIA, M GONORRHEA, M Dscntc HLYs O22Dscntd 'rdctv HLY GONORRHEA, F__ _____ CHANCPOID _ ___ 0 5 10 15 20 25 30 Years of Life Gained Per Case Averted SiF: Uver, February 1991, Figure 1. 4.12 Dynamic Benefits: Targeting "Core" Groups. Over's model provides estimates of the dynamic banefits of STD treatment (i.e. cases avoided in partners of STD patients), as shown in Tables IV. I and IV.2. Table IV. 1 shows these benefits in cases averted, while Table IV.2 measures healthy life years saved. Table IV.2 also combines dynamic and static benefits, to compute the total benefits in terms of healthy life years saved. These tables consider diseases one at a time, and do not yet estimate how classic STDs influence HIV transmission. The interaction among STDs is discussed below. - 78 - TABLE IV.1 - Dynamic Effects on STD Transmission of Preventing IQO STD C:ases in Core and Noncore Groups* Discounted New Cases Averted Over Ten Years Byv Targeting Targeting Ratio: Core Disease the Core the Non-Core to Non-Core Gonorrhea 4,278 426 10.0 Chlamydia 4,096 423 9.7 Syphilis 4,132 422 9.8 Chancroid 810 83 9.8 NIV without ulcers 1,744 180 9.7 HIV with ulcers 2,106 201 10.5 *Te new cases averted over ten years are the sum of the savings in both the core and non-core group of an initial preventive or curative policy applied to only one group. The streams of saved cases are discounted at an annual rate of 3%. Source: Over, February 1991, Table III. TABLE IV.2 - Discounted Healthy Life Years Saved per Case Prevefnted or Cured When Epidemics are Independent: Core vs. Noncore Static Dynamic Benefit** Total Benefit+ Disease/Gender Benefit* Core Noncore Core Noncore Chancroid 0.2 1.6 0.2 1.8 0.4 Chlamydia/F 1.3 53.3 5.5 54.6 6.8 /M 0.8 32.8 3.4 33.6 4.2 Gonorrhea/F 0.5 21.4 2.1 21.9 2.6 /M 0.7 30.0 3.0 30.7 3.7 HIV w/o Ulcers 20.0 348.0 36.0 368.0 56.0 HIV w. Ulcers 20.0 422.0 40.0 442.0 60.0 Syphilis 5.0 206.0 21.1 211.5 26.1 *Static Benefit is the benefit to only the cured or protected individual from curing or preventing a case of an STD. The figures are from Table I in source document. **Dynamic Benefit is the benefit to people other than the cured or protected individual from curing or preventing a single case of an STD. It is computed by dividing the figures from Table IV.1 by 100 and multiplying by the static benefit per case averted. +Total Benefit is the benefit of a single cure or prevention to both the individual and to the people he or she would have infected. It is computed by adding the static benefit to the dynamic benefit for the Core or the Noncore respectively. Sou: Over, February 1991, Table IV. - 79 - 4.13 Over calculates that by averting one case of HIV infection in a "core" group of ST) atients, 17 to 21 additional cases of HIV infection are avert over ten years. The "core" group is defined as non-monogamous and highly sexually active, with a higher baseline STD prevalence.Y In a non-core group, dynamic benefits are much smaller: only about two additional cases are averted in partners for every one case of HIV infection averted in an STD patient. 4.14 The core versus non-core distinction illustrates clearly that an STD treatment strategy which targets highly sexually active groups such as prostitutes may be ten times more effective than one that targets less sexually active groups. This differential impact is depicted graphically in Figure IV.2, which shows the dynamic benefits of a one-time intervention FIGURE IV,2 - Dyn_mic Benerit of Curing or Preventing 100 Cases of Gonorrhea in the CQre vs. Non Core Groups = 400 m 300 Effect of Curing 100 Core Cases LU C 200 Effect of Curing ,,D 100 Noncore Cases 0) 3 0 123 Year of Simulated Epidemic Source: Over, February 1991, Figure 2. F Over defines the two population groups by level of sexual activity and STD prevalence rates. The core group is five times as sexually active as the non-core group, with a starting prevalence of 20% for STDs (including HIV), compared to 1% prevalence in the non-core. Over's model is similar to Bjorkman's, with the exception of level of sexual activity (Bjorkman assumes that prostitutes, his core group, are 48 times as sexually active as their clients, the non-core group.) - 80 - designed to "cure" 100 cases of gonorrhea in the core and the non-core.2' Figure IV.2 demonstrates the major impact that program design characteristics, especially audience targeting strategy, can have on overall program effectiveness. 4.15 Interaction among STDs. Over models the interaction between classic STDs and HIV transmission, in order to determine the number of STD treatments required to avoid one HIV transmission. Unlike Bjorkman, Over includes dynamic benefits in his calculations, using a time frame of ten years and discounting future benefits by 3 percent. His figures therefore show a much more dramatic effect of STD treatment in reducing HIV transmission; Over calculates that for every one classic STD case treated or prevented among core group members, more than one case of HIV infection can be averted. As shown in Table IV.3, the number of cases of HIV averted ranges from 3 to 12, depending on the classic STD treated. This translates to between .08 (syphilis) and .36 (chancroid) STDs treated per case of HIV transmission prevented. Bjorkman's calculations, including only static benefits, showed that 14 STD treatments were needed to prevent one HIV transmission. 4.16 In the non-core group, low HIV prevalence means that more STDs must be treated or prevented in order to avoid a case of HIV transmission: the numbers range from .92 (syphilis) to 5.75 (chancroid) cases treated or prevented to avoid a single case of HIV transmission. The number of treatments required is still much lower than Bjorkman's estimate, however, showing the major influence of including dynamic benefits in the estimation model. TABLE IV.3 - Dynamic Effects on EIV Transmission of Preventing 100 SID Cases in Core vs. Noncore Groups Discounted New Cases Averted Over Ten Years By: Targeting Targeting Ratio: Core Disease the Core the Non-Core to Non-Core Gonorrhea 425.1 36.2 11.7 Chlamydia 355.8 30.3 11.7 Syphitis 1,207.8 109.1 11.1 Chancroid 275.9 17.4 15.9 * The new cases averted over ten years are the sum of the averted cases of HIV infection in both the core and non-core group of an initial prevention or cure of 100 cases of the classic STD in only one of these groups. In addition to these health benefits, saving 100 cases of classic STD also reduces future cases of that STD in the amounts given in Table IV. 1. The streams of saved cases are discounted at an annual rate of 3%. Source: Over, February 1991, Table V. 2' Figure IV.2 shows that while the one-time intervention in the core group produces greater short-run benefits, over the long run the epidemic returns to its previous equilibrium prevalence rate. This is because neither the core nor the non-core intervention produced sustainable behavioral change which would lead to a permanent reduction in the prevalence rate. The core group intervention does po$spone more ill health than the non-core intervention, however, and therefore achieves greater benefits. - 81 - 4.17 Table IV.4 and Figure IV.3 combine the static and dynamic benefits of STD treatment and HIV prevention, showing the dynamic benefits due to the interaction between classic STDs and HIV transmission in a separate column. Benefits are presented in discounted healthy life years saved. 4.18 The data show that the largest health effects are achieved by avoiding syphilis in the core group: By saving 453.1 healthy-life-years per case cured or prevented, this intervention has an even greater impact on health than would the direct prevention of a case of HIV infection. (Over 1991, p. 16) In most cases, the beneficial impacts of classic STD treatment are also enhanced two- to five- fold in the presence of an HIV epidemic; in the case of chancroid, STD treatment benefits increase 10- to 30-fold when the effects on HIV transmission are considered. TABLE IV.4 - Discounted Healthy Life Years Saved Mer Case Prevented or Cured When STDs Affect HIV Transmission: Core vs. Noncore Static and Dynamic Dynamic HLYs Total HLYs HLYs per Case from per Case from Saved per Case Disease/Gender Classic STDs ontv* Averted NIV Onlv** of Classic STD+ Core Non-Core Core Non-Core Core Non-Core Chancroid 1.8 0.4 55.2 3.5 57.0 3.8 ChLamydia/female 54.6 6.8 71.2 6.1 125.8 12.9 /male 33.6 4.2 71.2 6.1 104.8 10.2 Gonorrhea/female 21.9 2.6 85.0 7.2 106.9 9.9 /imale 30.7 3.7 85.0 7.2 115.7 10.9 HIV without ulcers 368.0 56.0 368.0 56.0 HIV with ulcers 442.0 60.0 442.0 60.0 Syphilis 211.5 26.1 241.6 21.8 453.1 47.9 * First two columns are from last two columns of Table IV.2. ** Columns 3 and 4 are the additional benefits of preventing or curing a case of a classic STD due to the indirect prevention of HIV infection. They are calculated by dividing the cases of HIV averted in Table IV.3 by 100 and multiplying by 20, the static benefit per case averted (see Table II of Over paper). + Columns 5 and 6 are the total health benefit per case of STD cured or prevented and are the sum of the benefits in previous columns. Source: Over, February 1991, Table VI. - 82 - FIGURE IV.3 - Total Health Benefit of Averting a Case of STD When SIDs Exacerbate HIV Transmission: Core vs. Noncore 0 100 200 300 400 500 CHANCRO ID CHLAMYDIA - GONORRHEA SYPHILIS - _= 0 100 200 300 400 500 Healthy Life Years Saved Pet Case Avertd [ From STD: Core 3 From HIV: Core |E From STD: Noncore EM From HIV: Noncore File: STD-CE.WK1 *Sorcver, February 1991, Figure 6. 4.19 Average Cost per STD Treatment. Using an average clinic cost per hour in Tanzania of $2.20 (1990 dollars), Over estimates the average cost per treatment of an STD (including drugs, but excluding diagnostic procedures).2 Treatment costs range from $1.04 for chancroid to $2.39 for chlamydia in female patients, as illustrated in Table IV.5. Higher clinic costs prevailing in areas such as Dar es Salaam would raise the cost per treatment by 25-35%. TABLE IV.5 - Average Cost per Treatment of an STD Case Disease sex Drug Cost Clinic Time TotaL Cost CUSS) (minutes) (USS) Chancroid M/F 0.67 10 1.04 Chlamydis M 0.92 10 1.23 Chlamydia F 1.84 15 2.39 Gonorrhea F 1.00 12 1.48 Gonorrhea N 1.20 10 1.60 Syphilis M/F 0.80 15 1.40 Source: Over, February 1991, Table ViII. ' Clinic hour cost estimates are from Scofield (n.d.), as cited in Over, February 1991. They range from a low of $0.80 in PHC clinics to a high of $4.52 in Dar es Salaam. - 83 - 4.20 Bjorkman calculates a single average cost of $1.93 per STD, using salary data from Kenya along with somewhat different assumptions about clinic time needed for treatment. 4.21 Cost per Effectively Treated STD Case. But average treatment cost is not of primary interest in performing cost-effectiveness analysis. Of more importance is the cost per effectively treated STD case. Cost per effectively treated STD case will be higher than the average treatment cost in most cases due to a) costs of diagnostic procedures; b) cost of presumptive treatment or treatment of inaccurately diagnosed cases who turn out not to have the STD treated; and c) problems with the efficacy of drug therapy and degree of patient compliance with treatment. 4.22 Table IV.6 shows ranges for the cost per effectively treated STD under various assumptions about the type of diagnostic procedures used and the true STD prevalence rate in the population. The minimum cost per effectively treated case in a population group with 25% prevalence (i.e. a core group) ranges from $4 (gonorrhea in male patients) to $11 (chlamydia in female patients), although costs can be as high as $70 per case if complete diagnostic testing is included. TABLE IV.6 - Minimum Cost per Effectively Treated Cases of SFDs: Sensitivity to Prevalence Rate and Diagnostic Procedure Treatment Effective- Prevalence Rate ness DISEASE ____25_____ Sex Dxiye Percentage 1 25X 5X Chancroid 90 II- I_ r_ B Clinical _ 109 22 11 5 3i 2 1 B Cutture 823 166 83 34 18 12 9 B Treat Everyone 115 23 12 5 2 2 1 Minimum Cost: 109 22 11 5 2 2 1 Chlamydia 90 X - - - F ClinicaL 1 258 245 119 43 17 9 5 F Culture 1_698 342 172 70 37 25 20 F Treat Everyone 266 3 27 11 5 4 3 Minimum Cost: 266 53 7 11 5 4 3 N Clinical 88 18 10 5 3 2 1 N Culture 1,596 320 161 65 33 23 17 N Treat Everyone 143 29 14 6 3 2 1 |Minimnum Cost: 88 1 10 5 3 2* 1 Gonorrhea 95 F Clinical _ 606 119 58 21 9 5 3 F Culture 643 130 66 127 14 _ F Microscopy 349 71 36 15 8 6 4 F Treat Everyone ___ 152 30 15 6 _ 3 2 2 Minimum Cost: ._ ._._ 152 30 15 6 3 2 2 N Clinical 66 14 8 4 3 2 2 N Culture 576 117 59 25 13 19 * 7 N Microscopy 155 32 17 8 5 4 3 N Treat Everyone 165 33 16 7 3 2 2 Minimum Cost: 66 14 8 4 3 2 2 syphilis 99 - B Treat Everyone 136 27 14 5 3 2 1 BSerology 187 39 20 9 5 4 3 Minimum Cost: 136 27 14 5 3 2 1 Source: Over, February 1991, Tabie - 84 - 4.23 It is striking to note that it is never cost effective tQ seek laboratory confirmation prior to treatment. and in almost all cases where ST) prevalence is higher than 25 percent, which might reasonably be expected among those seeking treatment for STDs, presumptive treatment is called for. Cost-Effectiveness Estimates 4.24 Estimates of the cost-effectiveness of STD treatment per case of HIV averted, where treatment is targeted to core groups of highly sexually active adults, range from about $3.0 to less than 50 cents. depending on the classic STD being treated. Table IV.7 provides cost- effectiveness ratios for the core and non-core groups, according to estimated STD prevalence in the population. TABLE IV.7 - Cost per Discounted Case of HIV Averted by STD Treatment: Sensitivity to Prevalence Rate and Core vs. Noncore Strateav (1990 US$) Discounlted Prevalence Rate Case of "IV - - - IDISEASE Cases of HIV tX I _Averted 10% X 25X% 5% 75% 100% IChancroid_ Core Group 2.8 38.93 7.86 3.93 1.79 0.71 0.71 0.36 Noncore Group 0.2 545.00 1110.00 T55.00 25.00 1T0.00 110.00 o 0.26 ChtamYdia, F Core Group 3.6 73.89 14.72 7.50 3.06 1.39 1.11 0.83 Noncore Group 0.3 886.67 176.67 90.00 36.67 16.67 13.33 0.08 l Chlamydia, M ___. Core Group 3.6 24.44 5.00 2.78 1.39 0.83 0.56 0.28 Noncore Group 0.3 293.33 60.00 33.33 16.67 6.67 0.10 l Gonorrhea, F _______ Core Group 4.3 35.35 6.98 3.49 1.40 0.70 0.47 0.47 Noncore Group 0.4 380.00 75.00 37.50 15.00 7.50 5T.00o- o.1T Gonorrhea, M _ Core Group 4.3 15.35 3.26 1.86 0.93 0.70 0.47 0.47 Noncore Group 0.4 165.00 35.00 20.00 10.00 7.50 5.00 0.09 Syphilis _ Core Group 12.1 11.24 2.23 1.16 0.41 0.25 0.17 0.08 Noncore Group 1.1 123.64 24.55 112.73 4.55 2.7 1.82 0.02 8 Note: Cases of HIV averted are from Table IV.3. Dollars per discounted case averted are calculated by dividing the minimum cost from Table IV.6 by the number of HIV cases averted from the first column of this table. Source: Over, February 1991, Table XI. 4.25 As described by Over, the C/E ratios exhibit a sensitivity to the level of sexual activity of the treated person and to the prevalence rate of the STD in the population being treated. - 85 - The cost of prevenutsig a case of HIV through STD control is extremely small when the STD program is aimed at a core group with high STD prevalence rates. For example, when prevalence rates are 25% or above, a case of HIV infection can be prevented for less than two dollars by targeting the program at the core group and treating everyone in that group for syphilis, chancroid and gonorrhea. (Over 1991, p. 25). 4.26 C/E ratios for non-core groups are less favorable, ranging from $124 to $887 per case of HIV infection averted (assuming a population STD prevalence rate of one percent). 4.27 Over's core-group estimates compare favorably with the cost-effectiveness ratio from the STD control program in Kenya ($6.00), discussed in Section 2 of this Chapter. The Kenyan program, -'hich also targeted a core group (prostitutes), may have incurred greater costs by including condom distribution and health education activities. 4.28 Table IV.8 presents the C/E ratios in terms of cost per discounted healthy life year saved, a metric which is more helpful if STD programs are to be compared to other health interventions. This table includes all health benefits of STD treatment, not just the HIV infection prevention benefits. The cost per healthy life year saved is less than $0.10 when STD treatment is targeted to core groups with STD prevalence rates above 25%. TABLE IV.8 - Cost per Discounted Healthy Life Saved by STD Treatment In the Presence of an MIy Egidemic: Sensitivity of Prevalence Rate and Core vs. Noncore Strategv (1990 USS) Healthy Life Prevalence Rate Years Saved DISEASE Per 1- 5% 10% -%X 75X t O X Efectively 1% 5 0 5 j *-~10 Treated Case Chancro d Core Group 57.0 1.91 0.39 0.19 0.09 0.04 0.04 0.02 lNoncore Group 3.8 28.68 5.79 2.89 1.32 0.53 0.53 0.26 Chlamdia __ ___ ________ ________ ___ Chlanwdia, F Core Group 125.8 2.11 0.42 0.21 0.09 0.04 0.03 0.02 Noncore Group 12.9 1 ~7TT20.62 1 T1 2.09 0.85 0.39 0.31 0.08 ChLamydia, M I Core Group 104.8 0.84 0.17 0.10 0.05 0.03 0.02 0.01 Noncore Group 10.2 8.63 1.76 0.98 0.49 0.29 2.20 0.1 Gonorrhea, F Core Group 106.9 1.42 0.28 0.14 0.06 0.03 0.02 0.02 Noncore Group 9.9 15.35 3.03 1.52 0.61 0.30 0.20 0.10 Gonorrhea. M _______ Core Group 115.7 0.57 0.12 0.07 0.03 0.03 0.02 0.02 Noncore Group 10.9 6.06 1.28 0.73 0.37 0.28 0.18 0.09 Syphilis I_I_I_I Core Group 453.1 0.30 0.06 0.03 0.401 0.01 0.00 0.080 Noncore Group 47.9 2.84 0.56 0.Z9 0.10 0.06 I r4 0.02 cRen Healthy life years saved from Table IV.4. DolLars per heatthy life year saved are calculated by dividing the minimun cost from Table IV.6 by the number of healthy life years saved from the first coLuffn of this table. Source: Over, February 1991. Table XII. - 86 - 4.29 Bjorkman estimates that the cost of preventing one case of HIV infection is approximnately $27.00 ($1.93 per STD treated x 14 STDs treated per case of HIV averted). Again. this cost-effectiveness ratio would be lowered if dynamic benefits were included in the analysis. Implications for Policy DecisiQns 4.30 Over makes a strong case for STD treatment programs in the conclusion to his paper: Even in the absence of an HIV epidemic, a treatment program for sexually transmitted diseases is a highly cost-effective government health care intervention when targeted to core groups. However the presence of an HIV epidemic multiplies the benefits of STM control several times without altering its costs... .If the cost of STD control per effectively treated case is unaffected by an HIV epidemic, as we assumed above, the cost-effectiveness of an STD treatment program in comparison to other government health care interventions is improved by [a factor of two (for syphilis and chiamydia) to thirty (for chancroid) times]. The argument for implementing such a program is extremely strong, both in absolute terms and in comparison with other health care programs. 4.31 Several factors can influence the cost-effectiveness of a particular STD treatment strategy. First, a strategy which targets "core" groups of individuals engaging in high-risk behavior (i.e. frequent sexual contacts and high rate of partner change) can achieve many times the benefits of a strategy which does not target "core" groups. To the policy decision maker, this means that clinics might be better placed in areas where urban residents and prostitutes can be served. 4.32 Secondly, confirmatory diagnostic testing is not essential to patient care for all STDs. Presumptive treatment of cases which exhibit certain clinical symptoms can reduce costs significantly, thus improving the cost-effectiveness of treatment. The cost savings from the elimination of some diagnostic testing could then be used to expand STD treatment to more patients, thus avoiding more cases of HIV infection. The advantages and disadvantages to adopting a po!i y of presumptive treatment should be discussed by clinicians and policy analysts, weighing cost savings against possible negative outcomes from improper diagnosis and treatment.1' 4.33 Finally, in order for an STD treatment strategy to have a lasting effect in reducing disease prevalence rates, the program must be sustainable. A one-time intervention achieves some health benefits by postponing cases, but does not reduce the total number of cases who will be infected in the long run. STD treatment should therefore be combined with other interventions like condom social marketing and health education in order to produce sustainable behavioral change, which, in turn, will help achieve permanent reductions in HIV prevalence rates. w' See pages 19-22 and Table IX in Over, February 1991, for a more complete discussion of the costs of screening (diagnostic testing) versus presumptive treatment (treatment without complete diagnostic testing). - 87 - 4.34 Areas where policy makers can be assisted by further research include &IU&W otimation and analysis of the feasibility and effectiveness of ptartner notification Drogram. Regarding demand for treatment, Over's paper assumes that all persons suffering from STDs will seek treatment. This assumption will need to be validated or modified on the basis of field experience, since people may in fact avoid seeking medical attention for STDs due to shame or a belief that the STD will go away without treatment. 4.35 If evidence emerges that many people do not seek treatment for known STDs, partner notification programs should be explored as a way to provide encouragement to infected individuals to obtain treatment. Partner notification is also important because some people may be infected with an STD but remain asymptomatic and thus unaware of the need for treatment. Partners of individuals with an STD have a much higher probability of being infected than an average individual. Once government treatment programs are operating effectively to meet existing demand for STD treatment, outreach programs which seek out and treat sexual partners may be a cost-effective means to expand AIDS prevention activities even further. 2. Condom Promotion L2 4.36 The proper and regular use of condoms during sexual intercourse between an HIV- infected person and a non-infected person has been found to reduce the probability of transmitting the virus to almost zero. It has also been suggested that condom use by those already infected with HIV may delay the onset of the disease state (AIDS) since the rapidity of the progression of the disease is likely to be increased with each new infection. Since sexual transmission accounts for about 80 percent of HIV infections in Tanzania, promotion of condom use - especially in conjunction with an effective STD case management program - is an important means of preventing further spread of AIDS. Condom use prevents AIDS by stopping the transmission of HIV itaelf and by preventing other classic STDs, which increase the likelihood of HIV transmission. There are two parts to any strategy for condom promotion: 1) increasing availability of condoms by reinforcing existing logistics systems and stepping up commodity procurement, and 2) increasing use of condoms through IEC activities. Addressing either the supply side or demand side issues alone will not provide an effective intervention.1' 4.37 As with the IEC activities described later in this chapter, the effectiveness of condom promotion is increased by targeting to high risk groups. Distribution efforts should therefore target strategic locations such as bars, night-clubs, hotels, markets, pharmacies, and health facilities, while IEC efforts should target strategic audiences including barmaids, students, military personnel, truck drivers, and STD patients. W' This section is drawn from interviews with Suzanne Thomas, former WHO condom logistics consultant to Tanzania; Paula Tavrow, former USAID Health, Population and Nutrition Officer in Tanzania; Steven Forsythe and Nancy Lamson, AIDSTECH Project; and various documents footnoted in text. LI' Suzanne Thompson, personal communication, Jan. 16, 1991. - 88 - Availlit 4.38 Until very recently, condom supply within Tanzania has been very limited, and has been noted as a factor limiting the success of the Government's family planning program. Currently, however, a large supply of condoms are available for distribution, free of charge, specifically to combat the spread of AIDS. In addition to the condoms received by the Family Planning Association of Tanzania (UMATI), which totalled about 5 million in 1990, NACP has received about 50 million condoms from USAID, . 'lout 37 million of which have been distributed throughout the country. USAID and other donors are also involved in smaller scale programs to promote condom distribution to groups at high risk of acquiring AIDS. For example, about 130,000 to 150,000 condoms per month are distributed through a pe'r education/condom promotion program targeting six truck stops along the Dar es Salaam- Mbeya road, managed by the USAID-funded AIDSTECH project. USAID/Dar es Salaam has also requested assistance from the USAID-funded AIDSCOM project, which has initiated a condom social marketing project with a budget of $400,000 for two years (excluding the cost of the condoms). 4.39 Condoms procured by the Swedish International Development Authority (SIDA) have been distributed in selected regions targeted at bars, restaurants, truck stops and clinics. Other donors (including the GTZ, the German technical assistance organization) have been involved in condom distribution as well, although exact quantities and program coverage are not known. Finally, a major AIDS intervention and research project ($8 million) to be funded by the Norwegians in the Kilimanjaro and Arusha Regions is in the planning phase and may include targeted condom distribution to truck drivers and other high risk groups.& 4.40 An informal survey by WHO indicated that the high number of condoms procured for Tanzania to date is not excessive. However, concerns are being raised that there may be some leakage of condoms to private sector pharmacies and across the border to Uganda, although this dispersion has been difficult to quantify.l' IUs 4.41 Again, WHO has noted significant levels of use among certain population groups in Tanzania, although quantitative data do not exist to verify use patterns. One small-scale targeted intervention geared toward prostitutes reported high compliance from clients, while the Tanzanian Armed Forces have also reported high use.'7' Of the 3 million condoms MY Paula Tavrow, personal communication, Jan. 26, 1991. t' Thomas, op. cite. -6 Thomas, op. cite. -7' Tbe targeted prostitute intervention was developed by Dr. Nkya, a Tanzanian doctor. The intervention reportedly achieved 100% compliance since the prostitutes knew each other well (they were from the same ethnic group) and worked together in a house, which made it easier for them to enforce rules about condom use. This project is an interesting model and should be evaluated and possibly expanded on a pilot basis to other areas. - 89 - procured for Tanzania by USAID in 1988, about half were distributed to military personnelA. 4.42 The survey on sexual practices in Tanzania, described in Section 2, below, contained several questions about condom use. The results indicate that most adults have knowledge about condoms, their use, and effectiveness in preventing conception and the spread of STDs, including AIDS. However, only a small proportion of those surveyed (less than three percent) reported using condoms regularly. These results, when fully analyzed, will be used to help develop more effective IEC messages and appropriate distribution channels. One important aspect to be investigated is anecdotal evidence of high rates of condom breakage, which may indicate improper use and could be corrected through education and training activities.'9 4.43 Use of condoms in Tanzania is considerably lower than in a number of other African countries. Comparative information from Kinshasa, Zaire, for example, shows relatively high use of condoms by men and women during extramarital contacts (24 and 12 percent, respectively). Half of male condom users reported that they used condoms "always" or "most of the time" with extramarital partners. Condom use was less frequent with spouses: 8 percent of married men and 7 percent of married women reported current condom use with spouse, but less than 2 percent of men aiid women reported using condoms "always" or "most of the time" with their spouse.2' 4.44 IEC activities are needed to promote condom use and to dispel mistaken notions about condoms which may discourage use for AIDS prevention (e.g. that condoms break during intercourse, decrease sexual pleasure, and stay in the woman's vagina after sexual relations).2W 4.45 One condom promotion and STD treatment program in Kenya has reported total annual operating costs of $70,000, including facilities, staff salaries, laboratory investigation, and therapeutic drugs.2' Given the target aud-ence for that program of 1,000 prostitutes, this amounts to $70 per prostitute per year. Unfortunately, it is not possible to separate the condom distribution program costs from STD treatment costs. 4.46 Three other condom promotion programs started with assistance from the USAID- financed AIDSTECH project in the Dominican Republic and Mexico have operating costs ranging from $20,000 to $70,000. The size of the target audiences is not known, but total LI Smith 1990, p. 3. w9' Thomas, op. cite. High breakage rates were also observed during the initial stages of a prostitute intervention study in Cameroon, as recorded in Vian 1988. V Bertrand, et. al. 1991, p. 55-56 and Table 4. 2' Bertrand, op. cite. ; Moses, et. al. 1990. - 90 - condoms distributed is between 78,000 and 180,000, or $0.35 to $0.55 per condom distributed.>' The condoms themselves account for between 8-13 percent of the total program cost. 4.47 Condom costs vary depending on procurement source, packaging material and shape, and whether Independent quality testing or shipment to port of entry are included. WHO- procured condoms cost $.024 each, while USAID procured condoms can cost between $.043 and $.045 each, or almost double.m' 4.48 The AIDSCOM condom social marketing project in operation in Tanzania is budgeted at $400,000 for two years, excluding the cost of condoms. Much of this budget is capital investmnent expenditures, however, including -inding needed to develop advertising capability in Tanzania. It is expected that recurrent costs will be lower. The distributer being used in the AIDSCOM project is the same one used to distribute Coca-Cola in the countryAl' 4.49 Cost savings for condom distribution programs are possible by sharing transportation, staff, and storage facilities for condoms with storage and distribution channels used for other commodities and supplies. Early in Tanzania's experience with condom programs, efforts were made to separate condoms from other commodities, which meant that program costs were relatively high. Alternative proposals for distribution were developed subsequently, relying on public transportation and other means to reduce costs. The NACP is now collaborating with the Expanded Program for Immunizations (EPI) to borrow EPI trucks for condom deliveries, which will substantially contain transport costs. Donors are also collaborating on the construction of a new warehouse for use by all PHC programs, including the NACP, which will reduce the rental payments now being incurred to obtain adequate storage space for condoms.2- Effectiveness of Condom Promotion and Distribution Programs 4.50 As with the IEC and counselling activities described above, the outcome of interest in condom distribution programs is cases of HIV infection averted. Measuring cases of HIV infection averted is made difficult by many factors, including different levels of infection among prostitutes and clients prior to the intervention, different levels of frequency of contact, and variability in proper use and efficacy of the condoms being promoted. Bjorkman I' Steven Forsythe, AIDSTECH Health Care Finance Research Associate, personal communication, January 17, 1991. 2' Prices given by Suzanne Thomas and Steven Forsythe, personal communication, January 1991. According to Thomas, the WHO price ($0.024) included plain silicon lubricant, square package, aluminum laminant packaging, independent quality sampling at the producer before acceptance, and shipment to port of entry. The AID price ($0.043) included plain lubricant, square package, plastic laminant, and no independent quality sampling or shipping costs. Discounts are provided on larger shipments. The second AID price ($0.045) was provided by AIDSTECH. LI Tavrow, op. cite. 2' Nyamuryekung'e 1990, p. 3. - 91 - (1991) provides some estimates of risk of sexual transmission and the effects of various parameters and assumptions. He concludes that if men use condoms in 40 percent of contacts with prostitutes, risk of sexual transmiission of HIV could be reduced by a third.22' 4.51 Only one study to date has attempted to measure the effectiveness of condom distribution by calculating the number of cases of HIV infection averted: in Nairobi, Kenya, researchers analyzed the cost-effectiveness of an STD/AIDS control programme for high frequency STD transmitters.' The intervention included STD treatment, general health education and counselling in addition to condom promotion, and was aimed at a cohort of 1,000 mostly HIV-infected commercial sex workers. The study team estimated that about 12,000 HIV infections per year were averted through the program, which had operating costs of $70,000, as noted earlier. This included infections averted among clients as well as contacts of clients. The cost per HIV infection averted was therefore about $6.00. 4.52 Over22' has taken a somewhat different approach in attempting to calculate the cost- effectiveness of a condom social marketing program in Tanzania in preventing HIV transmission. Using descriptive information provided by AIDSTECH staff, he has calculated the cost per couple-year of protection under the AIDSTECH condom distribution program in Tanzania. The results, shown in Table IV.9, indicate that a year's protection for someone in the "core" population costs about $456, while the costs for a "non-core" person are only a tenth of this amount. 4.53 Since the Tanzania program does not charge clients for the condoms provided, these costs - amounting to about $46 per couple-year of protection if targeted to the population at large - are met entirely by Government (and donor) subsidy. For comparison, the estimated costs of a couple-year of protection provided under condom social marketing efforts as part of family planning programs in Honduras and Bangladesh are about $15 and $6.50, respectively, of which about $4.60 and $5.90, respectively, were subsidized by the governments concerned. Thus, it should be possible to bring down the costs in Tanzania as experience is gained. 4.54 When a sexually active individual (male or female) insists that a condom be used on every sexual encounter, the probability of HIV infection on those encounters is greatly reduced. Fewer infections in the population means that there is less chance for other people to become infected - even if they do not use condoms. Therefore, the total benefit of expenditures to subsidize condom sales to a single individual extend far beyond that individual and include the discounted sum of all averted infections per time period. Furthermore, the condom user partly protects people who do not use condoms, because those people are less likely to encounter an infected partner. 2' Bjorkman 1990, p. 19. ;' Moses, et. al. 1990. V' Over 1991. - 92 - TABLE IV.9 - Average Cost per Year of Protection for a Condom Social Marketing Program in Tanzania Location 6 tfuck stops and 2 ttucking companies in Tanzania Start Date 1989 Baseline Information Half of drivers had more than 50 lifetime sexual partners. Condoms used consistently with casual partners by 42% of bar girls and CSWs and by 37% of drivers Accomplishments Each truck stop is serviced by an average of 200 CSWa who use 20,000 condoms per month or an average of 100 per CSW per month or 4 per night. Total Cost Per Year $100,000 budget plus cost of condoms provided free by National AilDS Control Program and amortization of vehicles provided free by AMREP. Perhaps $750,000. Total Condom Sales per 2 million Year Sales Price Per Condom No charge to consumer. Cost Per Core US$0.38 per condom * 4 contact per Couple night * 25 nights per month * 12 Year of months - USS 456 Protection (USS) Noncore US$0.3 S per condom * 10 contacts per month * 12 months = USS 45.60 Contractor Family Health International/AIDSTECH Source: Over, April 1991, Table I. 4.55 As in the calculation of benefits attributable to STD treatment, a distinction is made between benefits derived from reaching members of the "core" population as compared to the "non-core" population with a condom social marketing program.A' Using the same characteriza:ion of the path of an HIV epidemic over ten years as was used in the above Section on STDs,3" two alternative scenarios have been simulated for an AIDS epidemic in which 100 people use a condom on all sexual encounters during the first year of an epidemic and subsequently cease using condoms. The difference between the two scenarios lies solely in the identity of the condom users: in the first scenario all 100 condom users are in the core and in the second all 100 are in the non-core. The total number of new HIV infections over 2' The core group is assumed to be ten times more sexually active than the non-core, but fifty times less numerous in the population. wV For a full description of this model, see M. Over, "Costs and Effects of STD Treatment as a Strategy for AIDS Prevention in Tanzania," 1991. - 93 - ten years between the base scenario and each of the two condom use scenarios are then compared. The results are presented in Table IV. 10. TABLE IV.10 - HIV Infections Averted Over Ten Years When 100 People Use Condoms for All Contacts for One Year If 100 Users Are If 100 Users Are From Core Group From Non-core Group HIV without ulcers 189.8 2.4 HIV with ulcers 702.9 11.2 Sou Over, April 1991, Table II. 4.56 Importantly, one finds that the number of averted HIV infections is 60-80 times larger if the 100 individuals are in the core group than if they are in the non-core group, underlining the importance of targeting the intervention at individuals in the core group. 4.57 Table IV.9 showed that protecting a core individual for a year via a subsidized condom social marketing program is also more expensive than protecting a non-core person. Is the extra effectiveness of a core strategy worth the extra expense? To examine this issue, Table IV. 11 presents calculations of the cost-effectiveness of core and non-core targeted social marketing campaigns for averting HIV transmission. Costs per couple-year of protection have been set at $40 for a non-core person and $400 for a core person, roughly approximating the costs obtained in Table IV.9. Despite the ten time greater cost of protecting a core person, such a targeted program proves to be from six to eight times more cost-effective in averting future HIV infections than would a non-targeted campaign. TABLE IV.11 - Cost-Effectiveness of Social Marketing Campaign to Induce 100 People to Use Condoms for All Contacts for One Year Dollar Per Discounted Case Averted If All 100 People Using Condoms Are In The: Core Group Non-Core Group HIV without ulcers 210.75 1,666.67 HIV with ulcers 56.91 357.14 Soru: Over, April 1991, Table m. 4.58 At these assumed costs of a year's protection, social marketing is not competitive with STD treatment campaigns or blood screening for the prevention of HIV infection in Tanzania. However, suppose the costs achieved in the Honduran and Bangladeshi family planning programs are eventually achievable in social marketing campaigns in Tanzania. This would argue for reducing the cost per year of protection by a factor of ten to around $4 budgetary - 94 - cost per year of protection in the noncore and $40 in the core. A cost reduction this large would improve the cost-effectiveness of social marketing so that averting a case of HIV infection through social marketing of condoms will cost between $5.69 and $21, depending on the HIV prevalence in the population. Under these cost conditions, social marketing of condoms would be competitive with blood screening, but still three to five times less cost- effective than STD treatment for preventing a case of HIV infection.-W 4.59 Based on this analysis, it is recommended that STD treatment and blood screening be given higher priority than publicly-subsidized condom social marketing for HIV prevention. The promotion of condoms as part of an STD prevention and control program should prove to be more cost-effective than assumed in this analysis, because the existing health infrastructure can be utilized and those reached are likely to be at high risk for HIV infection. Subsidized social marketing of condoms should be tightly targeted at the core group when it is ultimately developed. The techniques used in the Tanzanian truck driver campaign are particularly promising in this regard, because they apparently achieve high coverage of the target group and, by employing peer counselling, they avoid the danger of stigmatizing that group. More evaluation of that experience would be useful. 4.60 Evaluation of the AIDSCOM project experience will also be important because it will give a sense of the exient to which people are willing to pay for condoms. If users are willing to pay for condoms, and since donors are able to obtain condoms from manufacturers at less than $0.05 apiece, it should be possible to offer them to the public at prices they can afford without requiring great subsidy from Government. Given the projected needs in the years ahead, at least partial cost recovery for condoms should be seriously considered. 3. IEC and Counselling Activities LI' 4.61 For the purposes of this study, the cost/effectiveness of four IEC interventions and one counselling intervention in Tanzania were estimated, using the model developed by Foote (1990). The results are reported in detail in Heald (1990). The interventions studied were major components of the National AIDS Control Program's IEC campaign, and were designed to promote behavioral changes that would decrease heterosexual and Derinatal transmission of HIV. In addition, the interventions were targeted nationally to those considered to be at risk of infection (i.e., women and men in their reproductive years, particularly school-age youth, college students, military personnel, prison inmates, women of child-bearing age, and families of those already infected). The interventions studied included: 2' See Table XI on page 25 of Over (op. cit., 1991). For a prevalence rate of 25%, treating the classic STDs averts a case of HIV infection for an average cost of US $1.50. i' This section draws heavily on the background paper by Heald 1991. - 95 - IEC INTERVENTIONS o Cartoons (created in Swahili for commercial print media which had appeared in two national newspapers) o Newspaper "News Flashes" (created in Swahili and English for commercial print media which appeared in three national newspapers) o Radio Spots (announcements broadcast over Radio Tanzania during intensive broadcast periods) o Posters, Calendars (with brief messages, distributed nationwide) COUNSELLING INTERVENTION o Counselling and Social Support to families of AIDS victims (provided by trained health professionals, usually within medical settings) 4.62 The IEC interventions all involved mass media for reaching large population groups. In contrast, the counselling program provides an example of a means for reaching a more limited audience, with a message designed specifically for them. Each of these interventions has different cost and effectiveness implications. cosu 4.63 As shown in Table IV. 12, IEC costs (excluding media and materials) included both investment and recurrent costs for the IEC unit of the NACP over a set project period.-; To these were added specific service activity expenses for media and materials. Costs were also "standardized" by holding constant the number of unique messages provided within a given time period. In this case, each intervention was costed assuming it was intended to convey 13 unique messages during a 90 day period.-' i' Heald notes that discount factors were not applied to correct for differences in timing of expenditures or program effects. Costs in Table IV. 12 were derived from historical and budgeted NACP expenditure data and MOH data on building costs. Cost data included salaries, building/office, utilities/communications, equipment/furniture, supplies and training, even if some of these costs did not appear in IEC budgets. Media and materials costs were calculated separately using commercial communications media charge data, and are contained in Appendix D. W Heald 1991, p. 16. - 96 - IV.12 - NA IE Cemts NATIONAL AIDS CONTROL PROGRAMME INFORMATION, EDUCATION AND COMMUNCATI COSTS (TOTAL AND DISTRIBUTED ACROSS MAJOR IEC ACTIVITIES / MEDIA) PERCENT TIME I COSTS DISTRiUlED ACROSS MAJOR EC ACTIVITIES I MEDIA (90 DAY C4ampaW) Recurrent Cost S8Y 6106.022 126A0 $50 $1,326 62.661 63.976 $18.024 SUkd*gfto 258 64 b 1 3 e to 44 " 6.5fS94 1.649 a 33 82 166 247 1.121 Equ~ewnt*urdlur. 53,906 13.476 d 270 674 1.348 2,021 9,164 8&6pIlss 13.420 3,356 d 67 160 33 603 2,281 Tra_n 71a.60 17.960 d 369 896 1.795 2.693 12.206 TranspotI 1,760 440 * 9 22 44 6s 290 &Mbow 4213.759 6440 $1'269 3,172 6.344 $9.516 $43.139 hN_b7tmaw Cest &0d000fc 661"44 $161 a 63 so 616 824 $109 ,lusortalo 3200 800 * $10 640 680 $120 S44 8-01Wm 63844 6961 619 848 69 $144 663 TOTALESTIMATEDCOSTS 0 27.03 664,401 $1.218 63=0 6.440 WM969 $43.7 a F res ba en 199091 salaries paid to 2.25 pr c l, .1 crar.4 _Ame s staf,44 pWort staW and 1 drior. Suc dMW I Hl. Nanl AIDS ConWt ProrameACP)saff.and the 'OVW ADS Corol AcM Up to SeptmWbe1990. p.47 * EsmatedtoeMnlu He tobe 2%o stmt b ld oflecososperyr. Sourcs Mbdinot Health. Planning OMIlon. B9sd an te 1900 MACP managsemw _opoal cs efstma mulOd b .157(IEC propoftn di th tl NACP 1990 budget. uding salaris). Sour: 'OvW of te AIDS Conrol Activities Up to Sept r 1990.' p. 47. - d on t 1990 IECbudget urce 'O vie theASCorSConrol ActMtlesptoSeptee19S0,' p. 34. PRecet fgur d on aJu etmatd fUl cos wr vehicl) of 6NMed WManenac cost d 960 0' 2 vshlcews) invemat fgue bsed n eimed COt o two vehe, t 66.000 seh. anorized Om as f bkvstens 9buIg m oostsbaedon3dl avg0S20P a rseahat635 r ar, b mate.a zd oer 20 e Sourw diniUV of Health. Plannino fWMkn Source: Heald, 1991, Table 2. - 97 - 4.64 From Table IV. 12, the following estimates of total cost by intervention are obtained: o Cartoons $ 1,719 o Newspapers $ 6,666 o Radio $ 15,185 o Posters/Calendars $ 82,038 o NACP Counselling $123,399 Effects 4.65 Effectiveness is more difficult to quantify. The effects presented here were based on estimates made by members of the National AIDS Control Program's IEC Subcommittee, guided by an IEC specialist. Together, they attempted to estimate each intervention's efficacy (how effective the strategy is at changing behavior to reduce risk), coverag (estimated number of people reached by the intervention), and ih (frequency of risk behavior and probability of transmission among the population being targeted). Total effect - the number of cases of HIV infection averted - was assumed to be a function of these three variables. Absolute estimates were thought to contain such a high degree of uncertainty that effects are presented only in relative terms, as discussed below. Relative Costs, Effects and Cost-Effectiveness Ratios 4.66 Estimates of costs and effects are combined and presented as relative costs, effects, and cost-effectiveness (C/E) ratios in Tables IV. 13, IV. 14 and IV. 15. Relative rather than absolute numbers are presented due to the high degree of uncertainty in assumptions regarding the costs and effectiveness of specific interventions. The cost-effectiveness of the cartoon IEC intervention was used as the baseline against which the other C/E ratios are compared. Table IV. 13 presents "intermediate" estimates of cost/effectiveness ratios. Sensitivity analysis was also performed, as illustrated in Tables IV. 14 and IV. 15 which present "optimistic" and "pessimistic" estimates, rvspectively. 4.67 On the cost side, Table IV. 13 shows that cartoons (the baseline) are the cheapest intervention. "News flashes" cost about four times as much and radio spots almost nine times more. On the very expensive side are posters/calendars (48 times higher cost) and counselling (72 times higher). 4.68 Looking at effectiveness, cartoons are also the least effective intervention. Posters/calendars are most effective (50 times as effective as cartoons), followed by radio spots (25 times as effective) and counselling (13 times higher). "News flashes" are about 3 times more effective than cartoons. - 98 - TABLE IV.13 - Estimated Relatie Cost/Efm etiveness of Alternati_e IEC Interventions in Avein New HIV Infections (Inmdiate Estimatle) TOTAL ES1MA1 . AB a E COST_ g00oyC Amplgn 1.0 L i1 7.0 45.2 70.4 E01MATED HIV PUWVENTd ABA1I OUTCOMES Prwenton EffICy 1.0 1.8 0.5 1.8 14.6 AudlmCoWOt 1.0 0.7 16. 17.4 0.1 Srtlc Audec SIXe 1.0 0.6 9.5 .5 0.0 IEC PRAch 1.0 1.2 1.8 1.8 4.2 IEC FtequenCy 1.0 0.9 1.4 1.3 1.7 Audnce Risk Factor 1.0 2.0 1.6 1.6 8.0 EtnadOros mpveeIons (000) 1.0 0.7 29.3 2Z2 0.2 Ewlmed Aveted Ca 1.0 2.7 25.2 50.4 12.6 ESTIMATE COST I OUTCOME FAOS CosIGrome lnprelon (000 - CPM 1.0 1 0.3 2.1 399.6 Cost/Averted cam 1.0 0.8 0.3 0.9 5.6 Sourc: Helid 1991, Table 3 TABLE IV.14 - Estimated Relative Cost/-Effectiveness ot Alternative M3C Interventions In Averting New HE Infections (Qgtimistic Estimate TOTAL ET W eu ROAM 006081 o0oevCamuPnp 1.0 I 21 7.0 45.2 70.4 ElBMA1 HIV PFIBEENT 1ED OUTCOMES PreVen0o EfICaNCy 1.0 2aS 1.0 24 15.7 AudIceCo0Vreg 1.0 0.7 14.7 15.7 0.1 Strate0I Audiene Sin 1.0 0.6 9.5 9.5 0.0 IECPech 1.0 1.2 1.5 1.7 &1 ISC Frsquen 1.0 1.0 1.4 1.5 2.5 Audlen. FSCtrs 1.0 20 1.6 1.6 1Q0 Es o"atd Ooe Imressn (000) 1.0 0.7 20.5 3.6 0.2 Estmated Averted Ca 1.0 3.6 22.9 59.5 12.4 E1EAw cooS OuTCOMERATIOS CosIaros impreIo n (000)- CPM 1.0 .0 0.4 20 368.8 I CoeAvnledCawf 1.0 0.6 0.3 0.6 5.7 u Heald 1991, Table 4. -99 - TABLE IV.15 - EstiMated gelative Cost/Etetiveyneua of Alternativ IEC Int=entions In Averting New HEY Infections flfinistic Eutimates TOTAL EST3TWED REATE COST 90O DyCaniagn 1.0. 2.1 7.0 45.2 70.4 ES11^ED HIV IPIVENlN REATED OUTCOMES Ptnllon Eflcacy 1.0 1.0 0.9 1.2 13 Audierce Cowwea 1.0 0.8 21.4 20.8 0.2 Strategic Audiec Sin 1.0 0.6 9.5 9.5 0.0 IEC Reach 1.0 1.3 2.2 .2 6. IEC Frequency 1.0 09 1.4 1.0 0.6 Audinco Risk Fators 1.0 20 1.6 1.6 6.0 Esilmated Grow iprelon (OO0 1.0 0? 29.4 20.8 0.1 EsbMatd Avwted Casa 1.0 1.6 30.8 39.7 13.7 EITUATWE COST I OUTOME RATIOS COwross IfmprssI (000) - CPM 1.0 3.1 0.2 22 673 CcsUAVtlAted casm 1.0 1.3 0.2 1.1 5.1 Sourc: Heald 1991, Table 5. 4.69 Effectiveness is broken down by the three contributing factors of efficacy, coverage and risk. Examining efficacy, radio spots are seen as least efficacious interventions, exhibiting about 90 percent the efficacy of cartoons. Posters/calendars and 'news flashes' are both about 80 percent more efficacious than cartoons, while counselling is estimated to have about 15 times the efficacy of the baseline intervention. 4.70 Secondly, looking at coverage, counselling services have the lowest estimates (as might be expected) showing only 10 percent the coverage of cartoons. "News flashes" are also slighdy less effective than the baseline (70 percent), while posters/calendars and radio spots have roughly 17 times the coverage of cartoons. The high coverage of radio interventions is also notable in Zaire, where more adults surveyed in Kinshasa were reached by radio than by any other medium (96 percent of men and 83 percent of women). 4.71 Finally, rhk factors do not show as much variability as do efficacy and coverage factors. Counselling services seem to reach audiences most at risk (8 times the baseline), while the other interventions (radio spots, poster/calendars, and 'news flashes") all reach audiences between 1.5 and 2 times more at risk than the audiences reached by cartoons. -V Bertrand, et. al. 1991, p. 54. -100- 4.72 In this exercise, the relative cost-effectiveness ratios obtained indicate that radio spots are the most cost-effective intervention strategy, averting cases at less than half the cost per case of the baseline intervention. The next most cost-effective strategy, posters/calendars, is slightly better than cartoons in cost effectiveness. "News flashes" are only slightly less cost- effective than cartoons, while counselling is five times less cost-effective than the baseline and Is the least cost-effective strategy overall. 4.73 The sensitivity analysis (Tables IV. 14 and IV. 15) shows that the relative C/E ratios remain stable, changing only slightly under the more pessimistic assumptions (posters/calendars become slightly more cost-effective than cartoons). Implications for Policy Decisions 4.74 The purpose of cost-effectiveness analysis is to show policy makers where they can achieve the most benefits for the least cost, and how they might best focus efforts to minimize costs or increase benefits in pursuing specific program objectives. Table IV. 16 summarizes the relative costs, effects, and C/E ratios for the MEC and counselling interventions considered, highlighting the most favorable (blocked rectangles) and least favorable (shaded rectangles) strategies according to each measure. TABLE IV.16 - Summary of Relative Costs. Effects and CIE Ratios, JEC and Counselling Avwted CosetAvated ctites I Medla Cost Effacy Coverage Risk HIV Cases HIV C Cartoons 1.0 1.0 1.0 1.0 'News Flashes" 2.1 1.8 0.7 2.0 2.7 0.8 Radio Spots 7.0 16.8 1.6 25.2 0.3 Posters / Calendars 45.2 1.8 17.4 1.6 [@ ] 0.9 NACP Counselors 114.6 6 l - 1 12.6 ue: Heald 1991, p. 24. 4.75 This analysis highlights several points. First, it shows that there is no clear "winner," i.e. an intervention that is vastly more cost-effective than others. Although radio spots have the lowest C/E ratio, three other strategies examined (posters/calendars, cartoons and "news flashes") have cost-effectiveness ratios in a similar range. Given this similarity, programmers - 101 - may wish to combine several of these strategies using different nmedia.22' Research has shown that a mix of medias can actually increase effectiveness of messages.'w 4.76 Secondly, the analysis suggests that counselling of the type being carried out is much less cost-effective than the other prevention strategies. The policy conclusion is that person- to-person counselling as a preventive strategy, because of its high cost, should be targeted to very high risk groups; that is, not just those who are at risk of acquiring AIDS such as families of AIDS patients but those who are also at particular risk of spreading the virus once infected. 4.77 AIDS prevention resources should not be wholly or mostly concentrated on counselling, but should emphasize more cost-effective strategies. This does not mean that counselling should receive no resources at all, however. The counselling program is a coping strategy as well as a prevention activity, and therefore has some benefits which have not been taken into account in the C/E analysis. In addition, equity considerations might require that counselling services be used to provide some AIDS-related messages which people need to know, but which cannot easily be communicated through any other media. Counselling can also be made more cost-effective through programmatic changes, as discussed below. 4.78 Finally, and perhaps most importantly, the analysis highlights many ways in which policy makers can increase the cost-effectiveness of particular strategies by attempting to boost effectiveness and/or reduce costs. 4.79 On the effectiveness side, the most significant factor seems to be coverage. For example, Table IV. 16 shows that although counselling has the highest efficacy and targets individuals most at risk, its low coverage and high costs offset the benefits, causing the cost per case averted to be relatively expensive. To improve the C/E ratio of this intervention, policy makers should therefore concentrate on improving coverage and reducing cost. Heald estimates that the cost-effectiveness of counselling programs could be made about equal to that of the other IEC interventions by modifying the program strategy to include group counselling sessions instead of individual sessions. While group sessions are not always feasible, the enormous impact of the increase in coverage is evident.22' 4.80 The next most significant factor influencing effectiveness is the degree to which the target audience is at risk of infection. The more at risk the audience is, the greater the effect the intervention will have. Cost-effectiveness of all interventions can be improved by focusing resources more on populations at higher-risk. Heald suggests that an IEC activity targeted more (though not exclusively) in the Dar es Salaam, Kagera and Mbeya regions can 3"/ Because of the many assumptions used in the analysis, further data collection and analysis may alter the results and show that indeed one or two interventions are much more cost- effective than the rest, and should be emphasized at the expense of the others. But for now, the four EEC interventions are not very different in terms of cost-effectiveness. LI Heald 1991, p. 28. 3' Heald 1991, p. 25-26. The total number of counselling sessions would remain stable at 33 sessions per counsellor. Only the attendance per session would be increased from one (individual counselling) to five persons in order to achieve this level of impact. - 102 - be twice as effective as one targeted to the balance of mainland Tanzania. Costs can be reduced substantially by targeting as well.' Methods of targeting include regional newspaper "runs" (allowing the NACP to place more cartoons and "news flashes" in papers distributed in high risk areas), and radio audience segmentation (to allow radio spots to be targeted during times when more at-risk populations might be listening, for example during sports shows). 4.81 Finally, effectiveness can be increased by enhancing the efficacy of the intervention (i.e. the proportion of the audience who will respond to the intervention by changing their behavior so as to reduce risk of infection). Efficacy is increased by providing messages which have been thoroughly researched so that they are adapted to audience values, beliefs and concerns. The targeting strategies discussed above, including regional newspaper "runs" and targeted broadcasts, can also improve efficacy since messages can be tailored to the audience being approached. Heald estimates that overall effectiveness can be increased 30-60 percent by providing "targeted messages that are more believable, that are ultimately shared among friends, and that are received repeatedly and across different channels. "41 4.82 On the cost side, reductions can be achieved by substituting lower paid staff where possible, or by relying more on community volunteers (although there can be an economic cost to involving community members if they reduce other productive labor activities in order to participate). Cost savings can also be gained by sharing equipment, buildings, and storage space with other programs or projects, and by contracting for goods and services through other suppliers who are able to produce more efficiently. 4.83 Because the above analysis only attempted to explore the relative cost-effectiveness of NACP IEC interventions, it shed little light on the extent to which their interventions have indeed been effective in bringing about the necessary behavior change to make an impact on the course of the epidemic. The preliminary results of a large, population-based survey of knowledge, attitudes, beliefs and practices (KABP) associated with HIV infection and AIDS in Tanzania, sponsored by the WHO's Global Program on AIDS, have recently become availableA2' They indicate that a very high proportion of the population (97 percent of respondents) are aware of AIDS and, indeed, are knowledgeable about the most salient features of HIV transmission. There were differences among people's knowledge, however, when the data were analyzed by sex, age and place of residence. Rural women, and adolescents, were l1: well informed about the disease than other population groups. 4.84 An important fact which relatively few people (39 percent) knew was that one could be infected with the virus but not have any symptoms, although IEC had attempted to address this issue. This is, of course, key information for assessing one's own risk of acquiring the disease. Significant proportions of the population also held erroneous beliefs concerning dP Heald 1991, p. 27. The author suggests that while targeted print messages have higher unit cost, total costs are still much lower. 4' Ibid, p. 28. L'Muhondwa (1991). - 103 - transmission of the virus, such as its transmission via mosquitoes (33 percent), kissing (31 percent), and sharing eating utensils (26 percent), also addressed in IEC messages. Fortunately, only a very small proportion of those who were aware of AIDS believed (erroneously) that one can get AIDS by merely touching an infected person (13 percent). As the researchers notes, "this has spared this country of the discrimination and panic which would have ensued as infected people continue to live in the communities and to use public acilities...."4' Although source of information was not sysLtmatically asked, most respondents indicated that they had acquired their information through radio broadcasts. 4.85 A complementary, population-based study on "Partner Relations and Risk of HIV Infection in Tanzania"4' has also been carried out. Its results indicate that, although the IEC carried out to date has been successful in informing people about AIDS, it has not resulted in significant behavior modification among those at risk. A large proportion of both men and women admitted to having had multiple sexual partners outside of marriage over the previous year (20 and 15 percent, respectively). One in three adolescents is sexually active. Relatively few people had protected themselves (or their partners) by using condoms (3 percent with spouses, less than I percent with non-spouses), or admitted to having decreased their sexual activity as a result of AIDS. This was consistent with the finding that few of the respondents perceived themselves to be at risk for acquiring AIDS, despite their behavior and "knowledge" of the disease. These findings suggest that if "effectiveness" is measured in terms of new HIV cases averted, IEC efforts to date have not been very effective. 4.86 Chapter VI assesses the NACP's IEC capacity and makes specific recommendations for strengthening its effectiveness. 4, Blood Screening 4.87 Currently, the most commonly used test for the presence of HIV antibodies in the blood is the enzyme-linked immunosorbent assay (ELISA). The average variable cost of an ELISA test can be as low as one dollar in a well-run laboratory in a developing country which performs large numbers of such tests. Assuming another dollar per test for management overhead, the average total cost of such a test could be as low as two dollars. However, to obtain this low a cost requires good management, well-trained and well-managed technicians and large volume, conditions which are difficult to obtain in developing countries. Costs of performing an ELISA test in Tanzania are probably higher; they have been estimated to be as high as $10 per unit of blood screened in one capital city of Africa.' 4.88 As ELISA tests are not really feasible under many of the laboratory conditions prevailing in developing countries, including parts of Tanzania, much research has been directed toward developing less expensive screening tests which are easier and faster to implement. Several of these "rapid" assays are already available and are being used. The 4'Muhondwa (1991), p.9. -wLeshabari (1991). 'Over 1991. - 104- assays have been shown to give reliable results in less time than the conventional method of ELISA, while providing high levels of sensitivity and specificity as well.& The average variable cost is as much as four dollars per test, which becomes five dollars with the addition of a dollar for management overhead.4' Effectiveness 4.89 Over (1991) has developed a model for determining the cost-effectiveness of screening a unit of blood. According to this model, if one assumed that a perfect test to determine whether a unit of donated blood is infected with HIV costs two dollars per blood sample tested, and five percent of donors are known to be infected, it would require an average of twenty tests to find a single infected unit of blood. By eliminating this unit of blood and replacing it costlessly with an uninfected unit, a transfusion service has avoided transfusing a patient with HIV-infected blood at twenty times the cost of a blood test. If one further assumed that three-quarters of the people to receive blood transfusions are HIV-negative before the medical emergency and subsequently survive the medical problem that caused them to need the blood, the cost of averting an HIV infection through blood screening is four-thirds of twenty times the cost of a single test or $53 per HIV infection prevented. The general equation is: Cost per HIV Infection = Cost per Test/Prevalence Rate Averted Survival Rate of Transmission Recipients 4.90 A more complete model of the cost-effectiveness of blood screening would relax many of the assumptions made in the above analysis. For example, the tests currently available are not perfect, especially under field conditions. They generate false positives and false negatives. Furthermore, infected units of blood cannot be replaced costlessly, but cost as much as $5 each to replace. However, elaboration of these more complete models with plausible values for the complicating parameters raises the estimated cost per averted case of HIV infection by only a few percentage points. Therefore, for ease of exposition, the results of this extremely simple model are presented here.40 4.91 Figure IV.4 shows on a logarithmic scale the relationship between the cost per averted case of HIV infection and the prevalence rate graphed for two different values of the cost per test, two dollars and ten dollars. At a prevalence rate of five percent and a test cost of two dollars, the cost of averting a case of HIV infection is $53 as in the above example. Note the dramatic effect of prevalence rate on cost. At one extreme, when the prevalence rate among donors is as high as 40 percent, as it might be for the donors that an urban prostitute would recruit from among her co-workers in a high prevalence African city, screening can avert a case of HIV infectior for only seven dollars per case averted. At the other extreme, in areas of low prevalence, blood screening is a very expensive way to avert HIV infection, costing N'Bjorkman 1991, p. 29. -'Over 1991. -W Bertozzi (1991) presents two more complex models and sensitivity analysis to show that they differ little from one another. - 105 - more than $6,000 per case averted. The left-most data S100000 Cost Per HIV Infection Averted point on the graph is at a $33 333 seroprevalence rate of 4 per 333 10,000 or 0.04 percentA., All of these costs are 3 multiplied by five if the cost $1,000 7 per test is ten rather than 67 7 two dollars. These higher S100 ss3 3 costs are related to prevalence rates in the upper line in Figure IV4. Cost eF Tes S2 I Cost Per tes = 4.92 In any given country $1 1X H I 0.1 1I 10X 10 I XII 0.01% 0.1% 1% 10% 100% the possible sites for blood Proportion of Donors Infected with HIV transfusion all have different prevalence rates among their Fue 4 Cost per HIV Infection Averted by Blood Screening as a donors and different Function of Prevalence Rate and Test Cost laboratory conditions leading to different test costs. This means that the cost-effectiveness of blood screening for averting cases of HIV infection will vary a great deal from one of these sites to another. Screening should begin at those sites where screening is most cost-effective and only be developed in the least cost-effective sites if the country is unable to avert cases of HIV or to save healthy life years more cheaply in other ways. 4.93 Bjorkman has also examined the cost-effectiveness of blood screening as a way to reduce HIV transmission through blood transfusions. Bjorkman's model assumes a mix of ELISA and rapid assays for blood screening. 4.94 The model first estimates the number of blood units which would need to be screened in order to avoid one case of HIV infection. Bjorkman performs this calculation by referral level (e.g. regional hospital, district hospital, and peripheral levels), since test sensitivity is higher at more central levels due to better laboratory conditions. At the regional level, it is assumed that donations are only accepted and screened after prior elimination of high-risk donors through the administration of a questionnaire. 4.95 Bjorkman finds that between 52 and 111 units of blood must be screened in order to avoid ono, case of HIV infection, depending on the referral level. Less testing is needed at higher referral levels (52 screenings at the regional level and 54 at the district level) compared with the periphery (111 screenings). This is because at the higher levels, blood banking is feasible and affordable, allowing regular, reliable donors to be used with some frequency. At the peripheral level it is assumed that no blood donations come from regular donors. 49' This is the prevalence rate recently reported among blood donors in Delhi, India by Singh eL (1990). Singh, Y.N., A.N. Malaviya, S.P. Tripathy, K. Chaudhuri, S.D. Khare, A. Nanu, R. Bhasin, "Human immunodeficiency virus infection in the blood donors of Delhi, India," Journal of Acquired Immune Deficiency Syndrome, 1990; Vol. 3, No. 2, pp. 1524. - 106- 4.96 On the cost side, Bjorkman assumes that a mix of ELISA and rapid tests is used. He adds in costs for repeated tests, additional laboratory supplies, and lab technician salaries to obtain a total cost of about $1.50 per HIV screening test, somewhat lower than that assumed by Over. 4.97 In the Bjorkman model, which assumes HIV prevalence rates of 10.2 percent in urban adults and 4.7 percent in rural adults, the cost per HIV infection averted is: Regional and district level: $ 80 Peripheral level: $166 4.98 Both analyses suggest that blood screening is not as cost effective an intervention for averting HIV transmission as, for example, an STD control program. Even more cost effective is avoiding the need for blood transfusions in the first place, as analyzed below. 5. Reduction of Blood Transfusions 4.99 It has been estimated that about 100,000 blood transfusions are given annually in Tanzania. Women, and children under five, are more likely to require blood transfusions than men, and therefore to acquire HIV via this route, because of the high prevalence of anemia among them, due to inadequate nutrition and to the severity of malaria attacks. Between 40-60 percent of transfusions are due to severe anemia in children, while 10-40 percent are due to severe anemia in pregnant women, before or after birth. Blood transfusions always carry risks, both because of the diseases which can be transmitted via this route (HIV, malaria, syphilis, hepatitis B and C, etc.) and because of the impaired healing and increased risk of secondary infections when done after surgery. For this reason, replacement solutions (plasma expanders) are often preferable to transfusion of blood for urgent volume rep!acement following acute bleeding. For anemia, however, prevention is preferable to treatment. 4.100 Thus, a second - and often preferable - way to avoid transmission of HIV through blood products is to reduce the need for blood transfusions. Methods to reduce blood transfusions include a) more rigorous adherence to medical standards to avoid unnecessary transfusions2'; b) transfusion of replacement solutions (plasma expanders) instead of blood as a way to treat some cases of acute bleeding; and c) prevention of anemia due to malaria (a condition which often requires blood transfusion) in children and pregnant women. The next section discusses options b) and c), replacement solutions and prevention of anemia. Use of Replacement Solutions 4.101 Major blood loss (more than 30 percent of volume) can lead to shock, and requires urgent volume replacement, usually with blood. Blood is rarely needed in the initial stages of bleeding, however, where the key problem is not the loss of red cells but the loss of volume. In these cases, plasma extenders such as crystalloids and colloids can be used as replacement fluids instead of blood. Advantages of plasma extenders include their immediate availability e' For example, in their study of Zairian hospital patients in 1988, Jager et. al. (1990) estimated that 13% of transfusions to children under five and 21% of adult transfusions could have been prevented had blood transfusior guidelines been in effect and strictly followed. - 107 - and relatively low cost. They also avoid some of the problems associated with use of whole blood, including disease transmission and immunological incompatibility.a' 4.102 Bjorkman estimates that between 625 and 1,250 replacement units (500 ml) are required to prevent one HIV infection, depending on the referral level. These figures assume that the solutions are used to replace blood units which have already been screened for HIVWY Where blood has not been previously screened for HIV, fewer replacement units (100) are required to prevent one HIV infection. 4.103 The cost per replacement unit5' is estimated at $0.20 per bottle, yielding the following cost-effectiveness ratios (per case of HIV infection averted): Where solutions replace screened blood: Regional level: $250 District level: $125 Per:pheral level: $200 Where solutions replace unscreened blood: Peripheral level: $20 4.104 If blood screening facilities already exist, replacing screened blood with solutions is not a very cost-effective approach if the only objective is averting HIV infection. However, since there are sound medical reasons for using replacement solutions rather than screened blood in some cases, the real lesson from this analysis is that doing so is also an effective means of reducing HIV transmission. If blood screening has not yet been established in an area, then it may be more cost-effective for reducing HIV transmission to put resources into replacement solutions and averting anemia, depending upon HIV-prevalence of the population. Malaria Chemotherapy and Chemoprophylaxis 4.105 Severe anemia in pregnancy, often caused by malaria, contributes substantially to maternal mortality and can result in lower birthweight, thus increasing risk of infant mortality as well.5' Children are also at risk: chronic malaria or even one acute episode of malaria a' Bjorkman, p. 27. 2' Since screening tests such as ELISA or the rapid assays are not 100% accurate, some HIV infections will occur even if all blood donations are screened and found negative. The use of replacement fluids in conjunction with blood screening can further reduce the incidence of transfusion-related HIV infections. IF Cristalloids (physiological saline solutions) are costed, rather than colloides, since the former are more stable and much cheaper. ' Belsey and Royston 1988, p. 697. - 108 - with high parasitic density can cause severe anemia in children under five.5' Treatment for thiz iife-tlhreatening condition often requires blood transfusions to boost hemoglobin levels. Prevention of anemia, therefore, is a way to avoid blood transfusions, thus reducing risk of HIV infection. It is also an activity which will improve the health of mothers and children and reduce infant and maternal mortality. 4.106 Anemia can be caused by inadequate nutrition, malaria, or other infections. Since malaria is the major cause of severe anemia,-' prevention and/or treatment of malaria in children and pregnant women is likely to be the most effective anemia prevention strategy. Anemia can be prevented through malaria chemotherapy (treatment of malaria with pharmaceutical products) or chemoprophylaxis (prevention of malaria with similar pharmaceutical products taken regularly and in smaller doses). 4.107 Bjorkman has quantified the benefits of malaria treatment and prevention in terms of numbers of HIV infections avoided. In areas where blood screening programs are already in effect and therefore blood supply is relatively safe, Bjorkman estimates that 300 episodes of malaria in children would need to be treated in order to prevent one HIV transmission. On the prevention side, Bjorkman notes that 250 child-years of chemoprophylaxis are required to prevent one HIV transmission. Similarly, for pregnant women 300 pregnancies with chemoprophylaxis are required to avoid one HIV infection. In areas where blood is not yet being screened, one-tenth as many malaria treatments or units of chemoprophylaxis are needed to avoid one case of HIV infection. 4.108 Treatment costs, including drugs and microscopic diagnosis, amount to about $0.15 per malaria episode, while chemoprophylaxis in children (one year) and pregnant women (one pregnancy) is estimated to cost $0.75. 4.109 In areas where blood screening programs have not yet been initiated, the cost- effectiveness ratios of anemia prevention strategies are quite favorable, ranging from $4.50 per HIV infection averted for malaria treatment, to $19 - $22 per infection averted for the two prophylaxis interventions. Where blood is already being screened, anemia prevention strategies avert less cases of HIV infection, which in turn increases cost-effectiveness ratios ten-fold ($45 per infection averted for malaria treatment; $188 - $225 for the malaria prophylaxis strategies). Table IV. 17 below summarizes these findings: N Bjorkman 1991, p. 35. N Bjorkman 1991 (p. 40), citing McGregor, et. al. 1966 and Flemin, 1989. - 109- TABLE IV.17 Cmt,F.ffeetivenes oif Anemia Pr.ventinn lhrniagh Mslnrin Treatment and Prevention Areas Without Areas With Blood Screening Blood Screening Unit Interven. Interven. Cost Needed* C/E** Needed* C/E** Child Malaria Treatment (per episode) $0.15 30 $4.50 300 $45 Child Prophy- laxis (per year) $0.75 25 $19.00 250 $188 Pregnant women Prophylaxis (per pregnancy) $0.75 30 $22.50 300 $255 * "Interventions needed" measures the number of treatments or units of prophylaxis required to avert one case of HIV infection. Prophylaxis during pregnancy is for last trimester only. ** C/E ratio is the cost per case of HIV infection averted, and is calculated as unit cost times interventions needed. For example, the C/E ratio of child malaria treatment in areas without blood screening is $4.50, or $0.15 x 30. 6. Possible Vacdne Develo_ment 4.110 Stories break in the international and domestic press frequently about the possible future development of a vaccine which may protect people from the effects of the AIDS virus. Certainly, much effort is being devoted to the development of just such a vaccine, and number of different types of vaccine which show some promise are being tested around the world. However, those most knowledgeable about ongoing research in this area warn that, even if these efforts are successful, it is extremely unlikely that they will lead to an eradication of the disease, or even to affect the course of the epidemic in the near future. It is even possible that the epidemic could be worse with a vaccine than without one, if people continued to practice high risk behavior. The problems are numerous. 4.111 First is the fact that any vaccine which seemed promising would have to be tested for a number of years before one could be sufficienty assured of its efficacy or safety to recommend widescale use. Because of the nature of the HIV, it is conceivable that a vaccine which appeared safe and useful in the laboratory could, in fact, cause the disease, or cause someone to be immune to the benefits of another, better vaccine which might be developed subsequently. Second, many vaccines being studied are very unstable, meaning that - if they prove effective - would be unlikely to remain so except under ideal conditions of, say, temperature and time. Given the difficulty which many countries have had in maintaining a - 110- cold chain for the measles vaccine, it is probable that many people who were given such an unstable vaccine would not, in fact, be protected from the virus. Third, is the general problem of vaccinating large numbers of people. It is extremely unlikely, even if a 100 percent effective vaccine were developed, that everyone at risk of contracting the disease could be reached with an immunization program. Fourth, is the cost. Given experience with other vaccines and drugs, it would probably be years - after a vaccine was developed, tested and proven to be effective - before the costs of producing it on a large scale could be brought down to affordable levels. 4.112 Unfortunately, if a vaccine only reached 70 percent of those at risk of acquiring the disease, and was, say, only 80 percent effective (only protected 80 percent of those people who received it) - an extremely likely scenario - almost half of those at risk could still become infected. If high risk behavior continued because people mistakenly assumed they were protected, the epidemic could be even worse than if no vaccine were developed. This means that the only chance for stemming the epidemic in the foreseeable future is behavior change, with or without a vaccine, and the recommendations contained in this report remain valid even in a "with-vaccine" environment. - III - V. COPING WITH THE AIDS EPIDEMIC 5.01 Policy makers in Tanzania have many options for coping with and preventing further spread of the AIDS epidemic. Two types of decisions must be made: first, a balance must be struck between investing in coping versus prevention activities, and secondly choices must be made about how best to invest resources among alternative strategies for either type of intervention. In this chapter of the report coping strategies will be discussed. These include medical treatment for persons with AIDS and programs for assistance to survivors. A. MEICAL TREATMENT FOR PEOPLE WITH AIDS 5.02 Medical treatment options for persons with AIDS include a) treatment for the many opportunistic illnesses which strike HIV infected persons, b) treatment which attempts to hinder or kill the human inununodeficiency virus directly, and c) treatment for the immunodeficiency resulting from HIV infection which allows the AIDS-defining opportunistic infections to develop.1' At this time, the latter two options are not viable treatment strategies in Tanzania. Those drugs which are available are extremely expensive,2' of undemonstrated efficacy, difficult to obtain and administer, and accompanied by serious (and even fatal) side effects. Tanzania should nonetheless continue to monitor developments in antiretroviral therapy and immunotherapy to re-evaluate the feasibility of these options as results of clinical trials becomes known and prices of effective, nontoxic antiretroviral products come down. 1. Treatment Options for O__ortunistlc Illnesses 5.03 Treatment of opportunistic illnesses associated with AIDS can have several benefits. First, treatment can slow the progression from AIDS to death, in some cases adding several healthy months or even years to a patient's life.4' In addition, treatment of these illnesses can increase the quality of a patient's life in the last few months, offering relief from pain and discomfort and perhaps adding to productivity as well. Thirdly, treatment may reduce the infectivity of the patient, thus reducing the probability that the person will spread the virus to others, although this effect has yet to be provenY Finally, treating tuberculosis and other 1' Ellen C. Cooper, "Treatment Issues in AIDS," Chapter 9 in Corless and Pittman-Lindeman, eds. AIDS: Principles. Practices and Politics. Washington, D.C.: Hemisphere Publishing Co. 1988. 2' For example, a six-month course of Kemron (an, as yet unproven, anti-HIV drug developed in Kenya) costs TSh 100,000 ($500) and one of zidovudine (formerly called AZT) costs TSh 230,000 ($1,150). -' For example, zidovudine inhibits the proper functioning of the patient's bone marrow, thus causing severe anemia, which itself requires sophisticated treatment. (Cooper, op. cit.) # Over and Piot 1990, p. 50. Treatment benefits in terms of life prolongation are difficult to quantify with precision. E Bjorkman 1991, p. 48. - 112 - communicable illnesses can prevent the spread of those diseases among the non-HIV infected population.0' Defining Treatment Objectives 5.04 An important policy question is how the objectives of case management for AIDS should be defined. This question becomes relevant because AIDS is a fatal and incurable disease. Since AIDS cannot be cured, a treatment strategy which focuses on alleviating the symptman of opportunistic infections might be more cost-effective than one which focusses on treatine the causes of these infections. 5.05 Two criteria for selection of treatment options should therefore be considered in the case of AIDS patients: 1) the relative cost and effectiveness of a given treatment to increase the comfort of the AIDS patient; and 2) the relative cost and effectiveness of a given treatment in preventing the spread of a given communicable disease (for example, TB) from an AIDS patient to the community.2' Case management according to these objectives would focus on palliative treatment except where the opportunistic illness is contagious. This is said to be the treatment protocol for AIDS patients practiced in Tanzania. 5.06 At the same time, policy makers must remember that most patients will be treated without the provider or the patient knowing that the patient is infected with HIV or has clinical AIDSY It is therefore difficult to implement a treatment protocol for opportunistic infections in AIDS patients that would be different from treatment protocols for the same illnesses when appearing in non-HIV infected persons. 5.07 For this reason, the treatment protocols modeled by Pallangyo and Laing do not handle opportunistic illnesses in AIDS patients any differently than illnesses appearing in the non-HIV infected population. Nevertheless, because Tanzania has limited resources to spend on pharmaceuticals, Pallangyo and Laing follow the two treatment criteria described above and suggest palliative treatment except in cases where patients respond well to curative treatment (e.g. pneumonias and tuberculosis). Pallangyo and Laing emphasize that these treatment protocols were developed in consultation with a relatively small number of Tanzanian clinicians and involved a number of assumptions which need to be confirmed or modified through discussions with a larger audience of health professionals from throughout Tanzania. Defining AlteMatie Treatment Options 5.08 Alternative treatment options are defined according to two parameters: referral system and drug availability. As shown below, Pallangyo and Laing present two possible referral systems, and three possibilities with regard to drug availability in order to compare the cost ramifications: 9' Slutkin, et. al. 1988. 21 Dr. Jean-Louis Lamboray, World Bank, personal communication 1/22/91. t Pallangyo and Laing 1990, p. 1. - 113 - REFERRAL SYSTEMS2' o CURRENT REFERRAL SYSTEM: 20% hospital care, 18% primary care, and 62% home care. o DECENTRALIZED REFERRAL SYSTEM: 12% hospital care, 27% primary care, and 61 % home care. DRUG AVAILABILITY o 100% DRUGS: All drugs on Tanzanian treatment protocols are available for all patients. o 60% DRUGS: Limited drug availability, assuming that 60% of all drugs on Tanzanian protocols (except antitubercular drugs) are available for all patients. Antitubercular drugs are assumed to be in constant supply due to the strong supply system supporting this program. o 100% DRUGS PLUS WHO: All drugs on Tanzanian treatment protocols are available for all patients, plus I percent of patients have access to the additional drugs which also appear on the WHO-recommended AIDS treatment protocol. Note that many of the WHO recommended drugs are included in the Tanzanian treatment protocols already. Those that are not are currently excluded, in theory, either because they are not on the country's Essential Drug List, or because the drug is very expensive or is not recommended by Tanzanian medical authorities. Analysis of this option is included to demonstrate the financial ramifications of allowing even a small proportion of AIDS patients to have access to these drugs. 2. Episodes of Illness Per Case 5.09 All treatment options assume that all HIV infected patients will seek treatment for at least some of the opportunistic illnesses from which they are suffering. Table V. I records the estimated number of episodes of each opportunistic illness for which it is assumed that treatment will be sought over the lifetime of an average AIDS patient. Adult conditions are separated from pediatric conditions. The table shows that adult AIDS cases will seek treatment for almost 17 episodes of illness before death, while pediatric cases will be treated for approximately 6.5 episodesA' 2' Figures presented are for adult patients. Percentages for pediatric cases are: 19% hospital, 12% primary, and 69% home care under current referral system and 13% hospital, 14% primary, and 73% home care under the decentralized referral system. LI2 The assumption that HIV infected persons will seek care for this many episodes of illness may be incorrect. Policy makers may want to consider alternative assumptions, e.g. care is sought for 50% or 75% of episodes. To do this, the second column of Table V.3 (total expected episodes) should be multiplied by 0.50 or 0.75. The result is to decrease the unit cost per AIDS case by half or three-quarters. - 114- TABLE V.A: Average Number of Episodes of Onnortunistic Illness gm AIDS Case. Adults and Children Percent Nu.mber AIDS Cases Episodes Total with 5 I if s I Expected A. Adult Conditions Eisode Eis EIsodes* ChronIc dIarrheal dIsease 60 4 2.40 Oral & oesophageal thrush 100 10 10.00 Tuberculosfs 25 1 0.25 Septicaemia 35 2 0.70 Pneumonfa 50 1 0.50 Neurological conditions 20 1 0.20 Severe headache 5 3 0.15 Eczema (allergic skin condition) 15 3 0.45 Bacterial skfn infectfons 35 3 1.05 Prurigo (allergic skin condition) 35 3 1.05 Pain ** 100 N/A N/A Viral skin infections 10 1 0.10 Kaposi sarcoma 2 1 0.02 Total Adult Epfsodes 16.87 B. Pediatric Conditions Tuberculosis 35 1 0.35 Pneumonia 50 2 1.00 Oral & oesophageal thrush 50 4 2.00 Persistent or recurrent fever 75 1 0.75 Chronic dfarrheal disease 50 2 1.00 NIV-associated skin disease 35 4 1.40 Total Pediatric Episodes 6.50 Sot: P aangyo and Laing, 1990. Expected episodes are calculated by multiplying the percentage of AIDS cases who are expected to have at least one episode, by the number of episodes which they are expected to have, given they have at least one. These figures are based on clinical impressions and small studies to date. Longitudinal cohort studies are underway which will provide better data for further analysis (Pallangyo and Laing, p. 19). ** Pain is assumed to be an underlying general symptom present throughout the life of the AIDS patient, and is not counted in episodes. Cost of pain relievers are included in cost calculations in Table V.3. 3. Costs-of Treatment Treatment Protocols and Cost Assumptions 5.10 Pallangyo and Laing's assumptions about Tanzanian treatment protocols were based on WHO draft guidelines for treatment modified to incorporate local practice variations as described in discussions with well informed Tanzanian specialists. Each treatment protocol estimated types and quantities of drugs needed, as well as days of care required by type of facility (e.g. referral hospital, dispensary, home care, etc.). Treatment protocols were developed separately for an episode of each type of opportunistic infection, and for adults versus children. - 115- 5.11 Costs were considered in two broad categories: drugs and nursing costs. Drug costs were established by multiplying quantities specified in the treatment protocols by prices obtained from the Tanzania Medical Stores and, where necessary, from international reference price lists. Nursing costs were calculated by multiplying the number of days of care required by each type of facility times the daily cost of care at that facility, then summing across all types of facilities used for treatment of an episode of illness. The nursing-day cost included all recurrent costs except drugs. In addition, home care was estimated to cost TSh 20 ($0.10) per day. Further information about the methodology used to calculate nursing costs is given below. 5.12 Because almost all drugs in Tanzania are imported, their costs must be paid in foreign exchange. This means that the true economic cost of drugs (the opportunity cost to the health sector and to the economy) is actually higher than the nominal cost would suggest. Likewise, the economic cost of health workers' and household care-givers' labor is probably different than the budget figures used in the analysis. Difficulty in ascribing shadow prices prevents a more accurate assessment of economic costs of patient care. Costs ger Episode of Illness 5.13 Table V.2 summarizes costs per illness episode for adult and child AIDS cases, under the current referral system, with 100% drug availability. Total cost per episode for adults ranges from TSh 723 ($3.75) for prurigo (an allergic skin condition) to TSh 45,699 ($237) for tuberculosis. Pediatric conditions range from TSh 1,514 ($7.84) for HIV-associated skin disease to TSh 28,100 ($146) for tuberculosisAl' Cost ler AIDS Case 5.14 While the cost of treating an episode of tuberculosis (TB) may be high, only 25 percent of adult patients and 35 percent of pediatric patients will suffer from this particular opportunistic infection. The average cost of TB per AIDS patient is therefore much lower: TSh 11,425 ($59) for adults and TSh 9,835 ($51) for children. In fact, the average cost for each type of opportunistic illness is different from the cost of one episode of that illness. Table V.3 combines the information on expected episodes of illness and cost per episode to produce estimates of the average lifetime cost per AIDS case for both adults and children. The average costs are disaggregated by type of illness, as well. The average lifetime ster adult AIDS case is estimated to be TSh 55.917 (02909. while the cost per pediatric case is TSh 37.651 ($195). Again, this assumes the current rEfal sym and 100% dEu availability. i' In their background paper, Pallangyo and Laing provide this kind of detailed cost data only for the one scenario mentioned, the current referral system with 100% recommended drug availability. For the other scenarios, the authors cite total cost per AIDS case, although the necessary assumptions for calculating cost per episode of opportunistic infection are embedded in their spreadsheet model. - 116- TABLE V.2: Cost per Episode of Onportunistic Illness Adult and Child AIDS Cases (TSh) Nursing Cost Drug Cost Total Cost A. Adult ConditIons er EEpisode ert Eoisode Chronic diarrheal disease 4,038 1,015 5,053 Oral & oesophageal thrush 859 344 1.203 Tuberculosis 35,565 10,134 45,699 Septicaemia 6,036 6,414 12,450 Pneumonia 6,123 1,633 7,756 Neurological conditions 10,630 0 10,630 Severe headache ° 10,559 86 10,645 Eczema (allergic skin condition) 2,701 82 2,783 Bacterial skin infections 825 159 984 Prurigo (allergic skin condition) 711 12 723 Pain 0 N/A N/A Viral skin infections 1 325 0 1,325 Kaposi sarcoma 5,146 0 5,146 B. Pediatric Conditions Tuberculosis 27,689 411 28,100 Pneumonfa 4 781 3,908 8 689 Oral & oesophageal thrush 3,315 1,008 4,323 Persistent or recurrent fever 4,091 2,649 6,740 Chronic diarrheal disease 3,184 124 3,308 HIV-associated skin disease 1,469 44 1,514 Source: Pallangyo and Lafng, 1990. Note: Pain is assumed to be an underlying general symptom present throughout the life of the AIDS parties, and is not counted In episodes. Cost of pain reli2vers is included in cost calculations in Table V.3. TABLE V.3: Average Lifetime Cost per AIDS Case. Adults and Children fTSh) Total Cost Total Expected Cost/Opportunistic Share of A. Adult Conditions Per Esisode Enisodes Illness Cost/Case Chronic diarrheat disease 5,053 2.40 12,128 21.7X Oral & oesophageal thrush 1,203 10.00 12,028 21.5X Tuberculosis 45,699 0.25 11,425 20.4X Septicaemia 12,450 0.70 8,715 15.6X Pneumonia 7,756 0.50 3,878 6.9X Neurological conditions 10,630 0.20 2,126 3.8X Severe headache 10,645 0.15 1 597 2.9X Eczema (allergic skin cond.) 2,783 0.45 1,252 2.2X Bacterial skin infections 984 1.05 1,033 1.8X Prurigo (allergic skin cond.) 723 1.05 759 1.4X Pain N/A N/A 741 1.3X Viral skin infections 1,325 0.10 133 0.2X Kaposi sarcoma 5,146 0.02 103 0.2X Total Adult 16.87 55,917 100.0X B. Pediatric Conditions Tuberculosis 28,100 0.35 9,835 26.1K Pneumonia 8,689 1.00 8,689 23.1X Oral & oesophageal thrush 4,323 2.00 8,645 23.0X Persistent or recurrent fever 6,740 0.75 5,055 13.4K Chronic diarrheal disease 3,308 1.00 3,308 8.8X HIV-assoclated skin disease 1,514 1.40 2.119 5.6X Total Child 6.50 37,541 100.0X sgurce: Pallangyo and Laing, 1990. Njgt: Pain is assumed to be an underlying general symptom present throughout the life of the AIDS parties, and Is not counted in episodes. Cost of intermittent use of pain relievers throughout life of AIDS patient (from diagnosis to death) has been included in this tabte. - 117 - 5.15 In addition to the total cost per case BREAKDOWN OF AIDS TREATMENT COSTS figures, several other By apun.tic IC observations can be drawn from Tables V.2 and V.3. First, OTUEA C 3 A septicaemia) account for about 80 percent of AIDS treatment costs in adults, as shown usH C21 5 graphically in Figure V.1. In children, TB, rUECULOSIS 20 4S) thrush and pneumonia account for almost three-quarters of all CGsts. A policy Figure I Contributions of Opportunistic Illnesses to implication of these AIDS Treatment Costs findings is that efforts to improve treatment protocols and efficiency of drug supply should target these diseases first, since potential cost savings are greatest. 5.16 The cost analysis also reveals that about 75 percent of all treatment costs fall in the category of nursing and institutional care. while drugs account for only a fourth of total cost. This finding indicates the need to focus attention on controlling use of services rather than use of drugs. When per case cost estimates are multiplied by the expected number of AIDS cases, as calculated in Chapter m, the estimated national costs of about $25 million (40-50 percent of government recurrent health expenditures) also immediately suggests the need to explore alternative treatment options. 5.17 One way to do this is to move toward a model of decentralized nursing care, where more patients are treated at the peripheral levels rather than being referred to expensive tertiary care facilities. This is certainly not a new idea to health policy makers; however, in the light of the AIDS epidemic, the potential cost savings of a decentralized system become much more important. This option is described further below. - 118 - Cost Implications of Alternative Treatment Scenarios 5.18 As a treatment alternative to the current practice of referring known AIDS patients to hospitals, Pallangyo and Laing have estimated days of care needed under a decentralized nursing model emphasizing primary care close to the patient's home. Tables V.4 and V.5 show the distribution of days of care by type of facility under this model, which may be compared to Table III.4 and 111.5 in Chapter 3, describing the current referral system. Due to increased efficiency and use of outpatient care, the model predicts a reduction of 5-8 percent in total days of care required for adults and children. While this decentralized pattern of referral is not currently in place in Tanzania, Pallangyo and Laing suggest that it would not be extremely costly to implement, and could decrease AIDS-related recurrent health care expenditures in the long-run. Implementation costs would include retraining of health personnel and strengthening of the infrastructure and supplies available in primary care facilities. Specific recommendations are contained in their background paper. TABLE V.4 - Days of Care, by Episode and Treatment Setting, Adult AIDS Cases Decentralized Referral System* Illness Days of Care Per Episode** Total - - - No. Total Total Days Epi. Hosp Home Care X Hospit Prim Home Tot-al per Care Care per Hospit X Home at Care Care Case Days Days Case al Chroni 2 5 10 17.0 2.40 4.8 24.0 40.8 12X 59% c Diarrhoea __ Oral Thrush & 0 4 5 9.0 10.00 0.0 50.0 90.0 0% 56X Cand. Pneumonie 9 2 5 16.0 0.50 4.5 2.5 8.0 56% 31X Tuberculosis 55 20 180 255.0 0.25 13.8 45.0 63.8 22X 71X Severe Headache 9 5 30 44.0 0.15 1.4 4.5 6.6 20X 68S Viral Skin Inf. 0.7 4 30 34.7 0.10 0.1 3.0 3.5 2% 86% Bact. Skin Inf. 0.7 4 5 9.7 1.05 0.7 5.3 10.2 7K 52X Eczema 3 6 10 19.0 0.45 1.4 4.5 8.6 16X 53% Prurigo 0 5 10 15.0 1.05 0.0 10.5 15.8 oX 67X Kaoi Sarcoma 4 0 10 14.0 0.02 0.1 0.2 0.3 29% 71X Neurol. Disease 9_ 0 45 54.0 0.20 _ 1.8 9.0 I 10.8 17X 83X Septicaemia 5 0 0 5.0 0.70 3.5 0.0 3.5 100% 0X Total 16.87 31.9 158.5 262.0 1l2 61X - 119 - TABLE V.5 - Days of Care, by Episode and Treatment Setting, Pediatric AIDS Cases Decentralized Referral System* ltIness Days of Care Per Episode** Total No. Totat Total Days Epi. Hosp Home Care X Hospit Prim Home Tot-at per Care Care per Hospit X Home aL Care Care Case Days Days Case al Chronic i 5 4 5 12.5 1.00 3.5 5.0 12.5 28X 40X Diarrhoea Oral Thrush & 0.5 4 5 9.5 2.00 1.0 10.0 19.0 5S 53K Cand.III Pneumonia 6 4 0 10.0 1.00 6.0 0.0 10.0 60X 0X Skin Diseases 0.9 3 3 6.9 1.40 1.3 4.2 9.7 13X 43X Recurrent Fever 4.5 3 5 12.5 0.75 3.4 3.8 9.4 36X 40X Tuberculosis 30 11 330 371.0 0.35 10.5 115.5 129.9 8X 89K Total I__ a_1 6.5 25.6 138.5 190.0 13K 73K * Source: Pattangyo and Laing 1990Appendix 1B. ** While total days of hospital care versus primary or home care may not have changed for certain conditions, some shifting from more expensIve care to less costly care (e.g. reference hospftal to district hospital) may have taken place. In some cases, total days of care needed went down under the decentralized model. This is because care Is assumed to be more efficient when decentralized, so less is needed. Also, patients don;t have to stay in the hospital or other facility to receive treatment, but can come and go in one day. In the case of diarrhea in children, the treatment days required under the decentralized model are actually higher than under the current system; however, cost savings are still achieved due to the shift in types of facilities used. More information can be found in the Appendix to Pallanryo and Laing 1990. 5.19 Pallangyo and Laing find that a decentralized system of patient referral cgn reduce patient treatment costs by about 25-28 percent in adults. and 20-22 percen in pediatric cases. The resulting per patient costs under the decentralized nursing model are TSh 42,104 ($218) for adult AIDS cases and TSh 29,823 ($154) for pediatric AIDS patients, assuming 100 percent drug availability. With 60 percent drug availability, costs are even lower: TSh 36,338 ($188) for adults and TSh 26,582 ($138) for children.2 This costs are summarized in Table V.6. 5.20 These estimates include recurrent operating costs, and do not take into account the investment costs which would be needed to institute changes in the existing referral system. Likely investment costs include development, publication and dissemination of new treatment protocols; curriculum development and training for care-givers at all levels (including home care); additional equipment needed to upgrade services at peripheral levels; changes in the drug and medical supply distribution systems to the periphery; supervision costs; or possible needs for increased staff. ' Since the peripheral health care institutions are located close to patients' homes, this kind of decentralization would, on average, reduce the travel costs of family members also. The analysis did not explicitly model the costs of family members, however. - 120- TABLE V.6 - Impilcatlon of Alternative Treatment Scenarios on Cost per Case Cost per Case Cost Per Case in TSh in US S C,*) Referral Drug Availability Cost Adult Child 2ption* Optfon Category . Adult Child Current Referral 100X Drug Availabilfty Nursing 41211 29405 214 152 System Drugs 14706 8246 76 43 Total 55917 37651 290 195 60X Drug Availability Nursing 41211 29405 214 152 Drugs 8928 4990 46 26 .________________________ Total 50139 34395 260 178 100X Drug Availability Nursing 41211 29405 214 152 Plus WHO Drugs for 1X of Drugs 19664 8246 102 Patients Total 60875 37651 315 195 Decentralized 100X Drug AvaiLability Nursing 27410 21592 142 112 Referral System Drugs 14706 8246 76 43 Total 42116 29838 218 155 60X Drug Avallability Nursing 27410 21592 142 112 Drugs 8928 4990 46 26 Total 36338 26582 188 138 100K Drug Availability Mursing 27410 21592 142 112 Plus WHO Drugs for 1% of Drugs 19664 8246 102 43 .____________ ._______ Patfents Total 47074 27983 244 155 Notes: * Referral and drug availability options are discussed further in text. Nursing and drug cost per case were derived from the Summary Spreadsheet, p.34 of Pallangyo and Laing 1990. ** Exchange Rate used is $1.00 = TSh 193. 5.21 The substantial reduction in treatment costs, coupled with the fact that clinical outcomes remain virtually unchanged from the baseline scenarioL', makes a decentralized nursing model for care a desirable policy direction for Tanzania. Specific steps to make decentralized nursing care for AIDS case management a reality need to be identified. As a starting point, some supply and drug distribution needs (e.g. additional ORS, gloves and IV sets at the peripheral levels) are mentioned in the section on clinical treatment of key opportunistic diseases in Pallangyo and Laing's paper. 5.22 Analysis of the 60 percent drug availability scenario under the current referral syst. shows that costs are lowered by about 10 percent. Under the decentralized system, limited drug availability lowers cost by up to 14 percent. Total cost is not affected more because drugs account for only 25 percent of patient treatment costs on average. This option is modeled to show what may be a more realistic cost estimation of the current situation in Tanzania, since many observers believe the current drug supply system is not without shortages. Whether 60 percent is a reasonable estimation of current availability of drugs is Jy Pallangyo and Laing 1990, p. 2. - 121 - not certain, but the model permits policy makers to see the general effect of drug shortages on total treatment cost per patient. Limited drug availability leads to suboptimal treatment and some treatment failures, and is not a desirable long-term option for case management. 5.23 Finally, Table V.6 shows that when all WHO-recommended drugs are made available to even a very small portion of patients (1%). the average cost per AIDS case jumps 9-12 percent for adults, from TSh 55,917 ($290) to TSh 60,875 ($315) under the current referral system, and from TSh 42,116 ($218) to TSh 47,074 ($244) under the decentralized nursing scenario. There is no change in cost per case for children, presumably because all WHO- recommended drugs are already included in the Tanzanian treatment protocols for pediatric conditions. Treatment outcomes may be improved for the 1 percent of patients receiving the more powerful drugs; however, the cost implications are substantial. It is likely that Tanzanian decision makers will want to evaluate WHO-recommended drugs which are outside the country's standard protocols on a product-by-product basis, weighing the added benefits from improved treatment outcomes against the cost for the specific pharmaceutical product in question. Pallangyo and Laing's model can be used to perform this exercise. 5.24 Table V.7 shows how the total direct costs of AIDS treatment in Tanzania are calculated under each of the treatment options. A summary comparing national AIDS treatment costs in 1990 under all options is presented in Table V.8. The analysis shows that national costs are lowest under the decentralized nursing model, ranging between $3.9 million and $21.1 million, depending on drug availability and the case projection scenario used. - 122 - TABLE V.7 - Implications of Alternative Treatment Options on Total Direct Costs. 1990 A. Under Current Referral System Drug Availability Case Per Capita Total Total Options* Projection Adult/ Treatment Current Estimated Cost Scenarios** Children Cost Cases 100% Drug 15% Monogamy Adult $290 69,010 $19,993,949 Availability Child $195 27.700 $5.403.796 Total 96.710 $25,397, 745 45% Monogamy Adult $290 16,030 $4,644,298 Child $195 6,260 $1,221,219 __________________ ______________ Total 22,290 $5,865,517 60% Drug 15% Monogamy Adult $260 69,010 $17,927,940 Availability Child $178 27.700 $4,936.484 Total 334.710 S22,864,424 45% Monogamy Adult $260 16,030 $4,164,395 Child $178 6,260 $1,115,610 Total 22.290 $5,280,005 100% Drug 15% Monogamy Adult $315 69,010 $21,766,755 Availability Plus Child $195 27.700 S5.403.796 WHO Drugs for 1% of Patients _ Total 572.710 $27,170,552 45% Monogamy Adult $315 16,030 $5,056,095 Child $195 6.260 $1.221,219 _________ ___ __ Total 622290 $6,277,314 B. Under Decentralized Referral System Drug Availability Case Per Capita Total Total Option* Projection Adult/ Treatment Current Estimated Scenarios** Children Cost Cases Cost 100% Drug 15% Monogamy Adult $218 69,010 $S15,059,198 Availability Child $155 27.700 $4,282,449 Total 810.710 $19,341,646 45% Monogamy Adult $218 16,030 $3,498,028 Child $155 6,260 $967,802 Total 22.290 $4,465.831 60% Drug 15% Monogamy Adult $188 69,010 $12,993,188 Availability Child $138 27,700 $3,815.137 Total 1.762,710 $16,808,325 45% Monogamy Adult $188 16,030 $3,018,125 Child S138 6,260 $862,193 __________________ _______________ Total 22.290 _ 3.880.318 100% Drug 15% Monogamy Adult $244 69,010 $16,832,004 Availability Plus Child $155 27.700 $4,282,449 WHO Drugs for 1% of Patients Total 2.952.710 $21.114.453 45X Monogamy Adult $244 16,030 $3,909,825 Child $155 6.260 S967,802 Total 22,290 S4.877,627 tote: Cost estimates fram Pallangyo and Laing 1990 ' Referral system and drug availability options are discussed further in text. " Case estimates are from Bulatao 1990, and are discussed in Chapter II. - 123 - TABLE V.8 - SuerY of Estimted Matiam A Trea - CoPtS in 1990 A ternat ve Tieotmmnt Scenarios Case Projection Scenario Referral Drug Avaflability Option Option 15X Monogamy 45X Monogamy Current 100X $25,397,745 S5,865,517 60X $22,864,424 $5,280,005 100X Plus WHO _ 27,170,552 $6.277,314 Decentralized 100X $19,341,646 $4,465,831 60X $16,808,325 S3,880,318 100X Plus WHO $21,114,453 $4,877,627 Sources: Bulatao 1990 and Pallangyo andLaing 1990 Methodological Considerations 5.25 Once again it must be stressed that the treatment protocols and cost assumptions used to make these projections should be verified and modified through further field research with input from local clinicians in the districts. Specific areas where additional data would be helpful include: a) Current percentage of facility expenditures allocated to salaries. The analysis built total costs from estimated salary expenditures in 1990, predicting what other expenditures should be given the percentage breakdown of costs for different facilities in 1980. One difficulty in using the 1980 cost structure is that today's cost structure may be different, given the economic recession during the past decade. As health resources diminish, countries often spend less on supplies and other recurrent costs but rarely cut payrolls. This leads to an increasing proportion of expenditures on salaries. Pallangyo and Laing's projections may therefore have overestimated current nursing costs.& b) Breakdown between use of inpatient and outpatient services and their relative costs. The analysis combines the two to compute an 'average day of care," when in fact outpatient care is much less expensive than inpatient care. A shift from inpatient to outpatient care at hospitals can produce cost savings even if the current referral system remains in place. This is an additional treatment option which can be considered once data are available. c) Proportion of AIDS patients who actually seek care. The analysis assumed that all AIDS patients will seek care for some episodes of illness. This assumption needs to be verified empirically. d) Distribution of patients by provider and the relative costs of facilities managed by different providers. Over et. al. (1989) note that health care costs will vary depending on the source of supply, e.g. public, NGO or for-profit. While private practice of medicine is not allowed in Tanzania, traditional healers are very involved in treatment of AIDS cases. The cost analysis assumed that the cost of service provision at public facilities was identical to the cost at NGO-managed facilities. Costs at traditional healers were not considered explicitly. Further research is needed Jw See Pallangyo and Laing 1990, page 23. - 124- to test the assumption that NGO and public facilities have similar costs, and to model traditional care costs. e) Deegree of inefficiency in the drug supply system and its effect on cost. The cost estimates by Pallangyo and Laing assume 100 percent efficiency in the drug supply system, which is highly unlikely. Assuming a 30 percent margin for losses, the estimates range from TSh 52,836 to TSh 60,329 ($274 to $313) for adult cases and from TSh 35,909 to TSh 40,125 ($186 to $208) for pediatric cases under the current referral system, 60 and 100 percent drug availability)1' B. SURVIVORS 1' 1. Growth In Numbers of Orphans and Other Dependent Survivors 17' 5.26 As the leading cause of death among adults in their most productive years, the AIDS epidemic is destined to have a serious impact on the dependents left behind. Survivors of the AIDS epidemic whose welfare may be at risk include orphaned children who have lost one or both parents, widows, widowers, and elderly family members aged 65 and above. 5.27 Widows and children, especially, may lack the means to support themselves, making it likely that they will encounter problems maintaining or finding adequate food, shelter, and funds for school fees and uniforms.Wy Most survivors will have incurred great costs associated with the illness of the productive household member, including medical care costs and lost productivity of both the sick person and the family members who had to stay home to provide nursing care.-W With the death of a husband from AIDS, the surviving widow (often ignorant of family rights laws regarding inheritance, or unable to afford a lawyer) may ly Losses may occur during storage or transit due to damage, spoilage, expiration and theft. Management Sciences for Health, in their book Managing Drug Supply, recommends a margin of 30% be anticipated until strong management systems are in place. The Essential Drugs Program in Tanzania should be consulted to determine if 30% is a realistic assumption for losses, given recent improvements in drug supply in the country. IV This discussion is drawn from the background paper by Ainsworth and Rwegarulira, 1991. 17/ The discussion on survivors is drawn from Ainsworth and Rwegarulira's background paper on survivor assistance (1991) as well as the background paper on the same topic by Bertozzi et. al. (1990). IF Bertozzi op. cite., p. 21. w Ainsworth and Rwegarulira, p. 6-7 and Bertozzi, p. 6. In addition, Tapia-Conyer, et. al. (1990) studied the economic impact of AIDS on 23 households in Mexico and found that household income decreased by 18% in over half the households during the nine month follow-up period after hospital discharge of the AIDS patient. Over 50% of expenditures in all households were related to patients' treatment and well-being. The patient was the main source of income in 60% of the households and a contributing source of income in an additional 20%, indicating that the approaching AIDS death would likely pose a serious economic hardship for these families. - 125 - not be able to claim inheritance rights and may lose access to her husband's land.A Others may blame her for the death of her husband and she may also lose access to her chlldrena' 5.28 Widowers may have difficulty caring for children and running the household, having lost their spouses (especially in monogamous marriages).A 5.29 These direct survivors are not the only ones to feel the effects of AIDS deaths, moreover. As Ainsworth and Rwegarulira note, even households not experiencing an adult death wIll feel the impact of the epidemic indirectly, as they are asked "to contribute to the funerals of neighbors and relatives, assist in the financing of health care, help in field work or the care of the sick, and take in orphaned children."2' Numbers of Child Orphans 5.30 Only one field study has attempted to systematically count the number of orphans in Tanzania. An enumeration of orphans in the Kagera Region was conducted in May 1991 by the Social Welfare Office and UNICEF. The survey counted 35,291 children under age 15 who had lost one or both parents. It is not known how many of these children were orphaned by AIDS. 5.31 Ainsworth and Rwegarulira have analyzed data from the 1988 Census of Population to provide further information about orphans under age 15 in Tanzania and how the AIDS epidemic has affected the rate of orphaning.2' Looking at data from the twenty regions of mainland Tanzania, they found that about 210,000 children ages 0-14 (1.96%) were "maternal orphans" in 1988, i.e. had lost their mothers. (No information was collected in the Census about chDidren whose fathers had died, or who had lost both parents.) Mwanza had the largest number of maternal orphans, at 20,238, followed by Kagera with 18,125. However, the orphan = in Kagera Region (2.92%) - hard hit by the AIDS epidemic - was by far the highest of any region. The next highest maternal orphan rates are in Mwanza (2.3%), Iringa (2.31%) and Mtwara (2.27%). 5.32 Using various assumptions Ainsworth and Rwegarulira estimate the total number of orphans in the 20 regions, including paternal orphans and correcting for orphans who have lost both parents. Table V.9 summarizes these estimates, providing additional detail on the five districts of the Kagera region. Inclusion of paternal orphans (and correcting for two- parent orphans) raises the total orphan rate to about 5.3% (about 545,000) across all twenty regions, with about 11 % of orphans having lost both parents. Unfortunately, no data are 2' Bertozzi, p. 12 and 21. Z' Ainsworth and Rwegarulira, p. 17; Bertozzi, p. 22. 2' Bertozzi, p. 26. V Ainsworth and Rwegarulira op. cit., p. 7. 2'Rwegarulira and Mushi (1991), cited in Ainsworth and Rwegarulira (1991). An additional 12,795 orphaned children 15-21 were also counted, which is probably a severe underestimate since, for example, those who had married were excluded. SAge 15 was used as an upper limit, since many youngsters 15 and older marry and form their own households; they are for all intents and purposes adults. - 126- available to estimate the orphan rate by region during previous years, so we cannot see what variability exists in orphan rates over time. In Kagera orphan rates in 1988 were higher than in other regions, presumably due to AIDS: an estimated 7.5% of children (47,000) were orphaned from one or both parents, with 17% of these orphans having lost both parents. (The excess burden of orphans attributable to AIDS is discussed below.) TABLE V.9 - Estimated Proportion and Number of Orphans in Twenty Regions, 1988 Maternal Orphans Estimated Children with Estimated Children with Region (1988 Census) One or Both Parents Deceased Both Parents Deceased Number X Number X Number - Arusha 9,525 1.51 25,978 4.13 2,579 0.41 Ruvuma 6,744 1.99 18,393 5.42 1,839 0.54 Iringa 13,095 2.31 35,715 6.31 3,571 0.63 Nbeya 11,328 1.69 30,895 4.61 3,089 0.46 Morogaro 12,073 2.23 32,927 6.10 3,292 0.61 Kagera 18,125 2.92 46,604 7.51 7,771 1.25 Kagera District Biharamulo 2,339 2.28 6,379 6.22 638 0.62 Ngare 2,096 2.73 5,716 7.44 572 0.74 Karagwe 3,328 2.36 9,076 6.44 908 0.64 Muleba 3,937 3.13 9,663 7.69 2,148 1.71 Bukoba U/R 6,425 3.69 15,770 9.06 3,505 2.01 JoLc_e: Ainsworth and Rwegarutira, 1992. Ornhan Rates and Age of Child 5.33 Maternal orphan rates calculated from the 1988 Census were higher among older children, as might be expected since their mothers are also older and more likely to have died. For example, Figure V.2 shows that across six of the regions analyzed less than 1% of children under five were maternal orphans, while 1.6%-3.3% of children ages 5 through 9, and 2.7%-5.0% of children ages 10 to 14 reportedly had lost their mothers. - 127 - FIGUME V.2 Orphan rates by age group Six regions of Tanzania, l1988 4.01 b3.5- . 2.89---- 0 2._6fi,@S 0~~~~~~~~~~~. 0 .4- ARUJSHA RUVUMUA RNA tKA M~OROORO KAGERA Region Children 0-4 ChN&Ven 5-9 m iChdren 10-14| Source: AinsWorth and Rwegarulira, 1992 Onrhan Rates and Adult Mortality 5.34 Figure V.3 shows the relationship between adult mortality and maternal orphan rates. Regions with higher adult mortality rates (ages 1549) also have higher maternal orphan rates. Thus the region of Arusha, with a relatively low adult mortality rate of 3.5 deaths per 1,000, had a maternal orphan rate of 1.5%, while the region of Kagera, with a high adult mortality rate of 9.4 deaths per 1,000 had a maternal orphan rate of 2.9%, almost double that of Arusha. 5.35 When quantified, the relationship showed that for each increase in the adult mortality rate of one death per 1,000 adults, the maternal orphan rate increased by 0.14 percent (statistically significant at p < .05). The positive relationship between adult mortality and maternal orphan rate was also found to be significant when orphans were examined by age with the exception of the youngest orphans (under age five).W ' Ainsworth and Rwegarulira discuss possible reasons for the lack of correlation at younger ages, hypothesizing that children orphaned very early in life are not likely to survive, and are therefore not likely to be recorded as orphans. - 128 - FIGUE V.3 Adult mortality and orphan rates Six regions of Tanzanla, 1988 3 t-o -0 §2.5- 2.31 7.8 - 8 II -~~~~~~~~~~~~~~~~~~7 2 0 - CL 0~ ~ ~ ~~~~~~~4 ARUSHA RUVUNA R4A kKYA MOROGORO KAGER Region Orphan tates Adulf death rat Source: Ainsworth and Rwegarultra, 1991. Excess of Orhans Attributable to AIDS Epidemic 5.36 The overwhelming majority of the estimated 545,000 in Tanzania at the time of the 1988 census were not orphaned due to AIDS, but due to other common causes of adult mortality. While orphans are a social issue of general concern to policy makers, a key question for AIDS policy is to determine the excess burden of orphans age 14 and younger created by the AIDS epidemic.2z' This may be accomplished by comparing the expected number of orphans without AIDS to the number of orphans observed in areas hard hit by the AIDS epidemic. 5.37 Assuming that none of the regions except Kagera were likely to have been affected by significant numbers of AIDS cases and deaths by 1988, Ainsworth and Rwegarulira calculate the mean maternal orphan rate for the non-AIDS affected regions and compare it with the ;Z/A study (Mutuembei 1988, 1989) conducted in two districts of the Kagera region in 1988- 89, found about 7,000 children age 16 and younger who had lost at least one parent to AIDS; about 30% of these children had lost both parents. The study did not attempt to determine the orphan rate in the districts in the absence of AIDS, however, and may have overestimated the number of AIDS orphans through respondent bias. The 1991 orphan enumeration of Kagera Region has not determined what share of the orphans were due to AIDS. - 129 - maternal orphan rate in Kagera.' They note that Kagera's maternal orphan rate is about 50% higher than the mean maternal orphan rate for the other regions (2.92% versus 1.88%). 5.38 Examining district level data for Kagera (Figure V.4), Ainsworti and Rwegarullra compare the maternal orphan rates for districts with low seroprevalence with those of severely affected districts of Bukoba and Muleba. Districts with high seroprevalence have high adult mortality and high maternal orphan rates. Almost 60% of all maternal orphans in Kagera are located in the two districts hardest hit by the AIDS epidemic, Bukoba and Muleba, which are also the most populous districts of the region. By comparing Bukoba and Muleba to the other districts thought to be less affected by the AIDS epidemic, Ainsworth and Rwegarulira predict that 30% of all maternal orphans in Bukoba and Muleba were due to the AIDS epidemic, while 17% of maternal orphans region-wide could be attributed to AIDS. 5.39 Including all orphans in the Bukoba and Muleba districts (maternal, paternal and both parents) an estimated 22 percent of orphans in 1988 can be attributed to AIDS deaths, and 12 percent (5,648) of orphans throughout the region of Kagera probably lost their parent or parents due to AIDS.22' FIGURE V.4 - Orphan Rates and Adult Mortallt. Kagera Region. 198748 l~~~~~~~~~~~~~~~~~~~re iN.~ ~m NgNa Cwp UbU Distnict | Orphan rate _Adult death rate| Source: Amsworth and Rwegarulira, 1991 S Ainsworth and Rwegarulira notes that adult mortality (and tberefore orphaning) is also influenced by other factors, including socioeconomic status, which would lead to differences in rates among regions even without AIDS. They remark that HIV leads to dramaticw increases in adult mortality rates, however, which argues in favor of this type of analysis to show impact of AIDS. s' Ainsworth and Rwegarulira, p. 16-17. - 130- Numbers of Widows. Widowers and "Elderly Orphans" 5.40 An analysis of 1988 Census data found that between 1.8% and 3.6% of adults ages 20-49 were widowed in six regions of Tanzania from all causes, not just AIDS-related.-' No correlation between adult mortality rates and widowing rates was found.3l' Data analysis is hindered by probable under-recording of widows and widowers, since many may remarry, migrate, or die themselves from AIDS within a relatively short period of being widowed. 5.41 The number of elderly who have lost their adult children to AIDS, and thus lost a means of support is likewise difficult to estimate. The 1988 census did not collect information related to these survivors; however, one study in the Bukoba District of the Kagera Region found 1,084 persons over age 50 who had lost their adult children to AIDS.W' 2. Survivor Assistance krograms 5.42 The deaths caused by the AIDS epidemic will have widespread effects on survivors, increasing the number of orphans and elderly dependents and reducing the standard of living of families who have lost household members to AIDS, or who come to the assistance of AIDS survivors. Given the striking differences in geographic dispersion of HIV infection, whole communities in some regions have already been affected and are suffering greater economic hardships due to the epidemic.-' Programs to improve welfare are needed to cope with survivors' difficulties. Possible programs enumerated by Ainsworth and Rwegarulira include: 3 Community services (water supply, day care, schooling, medical care); o Subsidies for schooling, medical care, housing; o Transfers of food, clothing or cash; o Orphanages and boarding schools; o Technical training; o Income-generating projects; o Free or subsidized credit; o Free or subsidized farm inputs; o Legal aid. 5.43 Ainsworth and Rwegarulira note the similarity between programs designed to assist AIDS survivors and general poverty alleviation programs. In fact, they argue that assistance should be targeted to the neediest families, regardless of whether they have had an AIDS -' The census noted marital status at the time of data collection and did not ask about "ever widowed." 2V Ainsworth and Rwegarulira, p. 17. -W Mutembei 1989, cited in Ainsworth and Rwegarulira 1992. It is not clear from the original source whether the author meant that respondents had lost all their children, or one orm gf their children (i.e. the one(s) with whom the elderly respondent lived or who supported the respondent). The study did not cover the entire region, so it is not possible to express 'elderly orphans' as a percentage of the district population. -v Bertozzi, et. al. 1990, p. 6. - 131 - death or not, both to avoid the stigmatization of AIDS survivors and to ensure that the neediest community members receive assistance.;' Criteria for Selection of Alternatives 5.44 Criteria for choosing among alternatives for survivor assistance include a) the relative costs and effects or benefits of different programs; b) the ease with which the programs can be targeted an.d monitored to assure they are reaching truly needy individuals; and c) the long- term sustainability of the programs. 5.45 Very little information exists about the relative costs and effectiveness of alternatives for survivor assistance in Tanzania or other countries. It is difficult, therefore, to compare the cost-effectiveness of these strategies. Nonetheless, Ainsworth has described the rationale or expected benefits as well as the range of costs for four types of survivor assistance programs now in operation in Tanzania. These programs have been selected for analysis for illustrative purposes only and not because they are necessarily "better" than other programs. The four types of strategies include orphanages; direct relief; feeding posts/child care centers; and subsidization of school fees and uniforms. All four types of programs are either under discussion or already being implemented by the Government and/or NGOs in certain regions of the country. 5.46 Table V. 10 summarizes the analysis of the survivor assistance programs, each of which is described in further detail below. Analysis of Options 5.47 QDhanage. While orphanages can provide a high standard of living for AIDS orphans, demand is low for the beds now available even in the areas most affected by the epidemic. For example, the three orphanages located in Kagera Region were operating at only 57% occupancy in 1990. Low demand probably reflects the cultural preference for caring for children within the extended family and an aversion to the "de-socialization" perceived to occur in institutional settings like orphanages.v' Orphanages are also one of the most costly options for survivor assistance (about 125,000 - 179,000 TSh or $650 - $930 per child per year in Dar es Salaam for children 3-18; less in rural areas) and are not able to accommodate large numbers of children. M Ainsworth and Rwegarulira, p. 21. They note, for example, that families who take in AIDS survivors may be in need of assistance, even though they have not had an AIDS death in the family. ; Ainsworth and Rwegarulira, p. 23. - 132 - TABLE V. 10 - Analysis of Survivor Assistance Options Rationale Drawbacks and Estimated Costa Existing Programs Strategy and Benefits other Considerations (See Note) Sponsors, Beneficiaries Orphanages Large numbers of orphans Children do-socialized; 1265310 TSh (0649) to - 34 orphanages In Tanzania; many exceed coping better to stay with family 179,014 (8928) per child 1,083 orphans, age 0-18; capacity of extended - Orphans may become per year at Kurasini all but 1 run by NOOs. family permanent wards of State Natlonal Children's Home In NOO orphanages - Stigma of AIDS makes Orphanages expensive to DSM I (Govt run, no Infant subsidized by GOT and familles reluctant to build and maintaln care; Incl. rental value of supervised by MLYD adopt AIDS orphans Lack of demand bldg). - Orphanages may offer (operating at 57% of - Cost of GOT subsidy to high standards of living, capacity in Kagera 11990) NOO orphanages: 29,200 complete support for (11156/child/yr wards, including education & medical care Direct Raises consumption level - Unlikely to have impact - SWO Prog. In Kagera: Kagera: SWO of MLCSW- Transfers of survivors by providing on long-term welfare 383 TSh (8 21/chIld/mo. for 11,024 children); Tarnz Red food, clothing & basic unless sustained over food assistance Cross; Lutheran Church; commodities long period Baptist Church In Kagera: Baptist Church 150 - Improves short-term - Expensive and difficuit to 2,000 TSh families); PartageI welfare sustain (810/family/mo. for - DSM: WAMATA 1(72 - Helps poor families - Does not foster self. unspec. support orphans, 47 widows) through acute cries reliance - WAMATA In DSM: 2,500 - Lack of data on 8 t13)child/mo. for food experiences with this assistance. Also gave strategy In Tanzania same amount to HIV+ persons for any needs. Feeding Posts - Ensure that at least one - Programs rely heavily on - Eat cost for 40 child, day - 262 feeding posts In 6 and Child Care good meal/day for community resources, care & feeding center regions serving 13,362 Centers children <7, thus especially food and 3,338 - 7,394 TSh. or 817- children -6, most under reducing risk of organizational capacity 37 per child per yr age 2. Avg. child, per post malnutrton - Communities hard-hit by excluding donor Inputs = 53. Moat - Free up time of adults & AIDS may not have (additional 84 per yr), voluntary/commun. older children for enough resources to get training, community financed, with assigt. from productive activities & started and may need mobilization & other start- donors (UNICEF) schooling extemal help up costs - 1,692 day care centers in - Relies on self-reliance & all regions serving 11 7.413 iniftiation of community children. Age 2-6. Avg. - Long and short-term 70 per center. Open 9-1 2. benefits Supervised & financed by - More sustainable In long- GOT, MLYD- run - Substantial experience with this option In Tanzania Subsidies of - Long-term benefit ftrom - Ethics & falmess: target - WAMATA assistance: - WAMATA: 54 students In School Fees & Investment in human only AIDS orphans when 6,000 TSh (*26) /child/ primary school; 20 In Uniforms capital; raised eamings of other familis have needs year primary school; secondary school, Dar es beneficiaries too? 20,000 1I104)/chIld/yr Salaam - High perceived need - Leakage: How to assure secondary school - Partage/ELOT actve In among survivors only truly noedy benefit - Et. cost for out-of-pocket Kagera Region - Expensive and not expenses, per child/yr. - 200 children identified as sustainable Prim, school: 3,700 TSh needing assistance by - Unknown impact on (819) Secondary public day SWO In Kagera Region. enrollment and long-term 13,700 (871); second. welfare public boarding 17,000 - Doesn't promote self- (188); second, private add reliance 3,000 TSh 18161; doesn't count opportunity cost of foregone labor of child in household production OURE J In N WEGAUUA 1902. ACRONYMS USED IN TABLE (NOTED WITH '- SWO: bldd Wefare Office MLYD: MWnry of Lbo mnd Youth Development, formwly Ministry of Labor. Culture ond Social Welfsre Potage: An NOO, *Pwtog avec 1.. Enfont du Tie Mord.' OSM: DOr rEalsam GOT: Govmnant ot Ta*tanie NOTE: Estimad cas wr calculaed by Ainwrth a d Rwaonlra arid we docunwnttd In the Annexe to their report. Thes cot should be Interpreted with exm cAution given the leck of relabb dat on wMch to bas asaurnptkne. They sre me to give onry a reuh idea of rane of ceost. - 133 - 5.48 This strategy for assistance may become more appealing, however, as the epidemic progresses. Ainsworth and Rwegarulira explain: If child abandonment becomes commonplace because of the AIDS epidemic.. .orphanages may have to be considered in spite of their expense and inherent disadvantages [although] experimentation with other less expensive arrangements would be highly desirable before embarking on such a course3A' 5.49 Direct Transfers. Direct transfers of food, clothing, basic commodities, or funds with which to purchase these items would immediately raise the consumption levels of survivors, thus improving short-term health and welfare. While not widespread in Tanzania, a fair number of direct transfer programs are currently being administered in the Kagera Region and in Dar es Salaam by the Department of Social Welfare in the Ministry of Labour and Youth Development and by NGOs. 5.50 The disadvantage of direct transfers as a strategy is that it is a short-term solution which, given the expense and lack of emphasis on personal or community development, is not sustainable over the long run. Nonetheless, direct transfers can provide for acute needs temporarily, as more sustainable strategies are being planned and implemented. 5.51 Costs vary a great deal, depending on the level of support. Existing programs provide assistance valued at from 4,600 - 30,000 TSh ($24 - $156) per beneficiary per year. This does not include program administrative expenses. 5.52 Feeding Posts and Child Care Centers. Ainsworth and Rwegarulira highlight two principal benefits to feeding posts and child care centers for orphans and non-orphans alike: first, the centers can ensure that young children (usually under age seven) receive at least one nutritious meal per day, thus reducing the risk of severe malnutrition to some degree.37' Secondly, the provision of day care can free up adults and older children to engage in productive activities or schooling. 5.53 A key advantage of this strategy is that it relies on the community to initiate, direct and finance activities. In this way it promotes self-reliance and enhances long-term sustainabDity. While relatively more expensive than some forms of direct transfer assistance, program costs for feeding posts/child care centers are stil not exorbitant, ranging from about 3,400 - 7,400 ($17-38) per child per year excluding start-up costs and donor inputs (see Table V. 10 for further detail). 5.54 The community support involved in program implementation and financing may also be a disadvantage to this strategy, however. In communities hard-hit by the epidemic, internal resources may already be exhausted and external assistance may be necessary to start- up such programs. 5.55 Subsidies for School Fees and Uniforms. The final strategy analyzed by Ainsworth and Rwegarulira is subsidization of school fees and uniforms. Such a strategy would respond V Ainsworth and Rwegarulira, p. 34. ' Evidence that feeding posts can reduce malnutrition is described by Ainsworth and Rwegarulira who cite Tanzanian research (JNSP 1988). Evidence has also been accumulated in other countries, for example, Peru (USAID 1990). - 134 - to one of the greatest perceived need of survivor populations, as evidenced by focus group research conducted in the Kagera Region (Bertozzi, et. al. 1990). It could also have a long- run impact on survivors' welfare by increasing earnings potential, since people with education are more likely to earn higher salaries than people who have not received formal education. 5.56 Unfortunately, the connection between education and earnings potential may not be very great given recent declines in quality of instruction.A Subsidized school fees/uniforms may not increase enrollment among orphans if the opportunity cost to the family is perceived to be too high (i.e. if the family cannot afford to lose the contributions that the orphan provides to the household in terms of farm work, child care, food preparation, or other activities). 5.57 Ethical issues regarding targeting criteria are also of concern. Analysis of census data revealed that, although maternal orphans have somewhat lower enrollments than non-orphans, enrollments even for non-orphans were low. School fee subsidies provided only to orphans may not be targeted to the most needy children, and may mean that orphans who would have enrolled in school anyway receive unnecessary benefits while non-orphan children from extremely poor families are ineligible. Limiting subsidies to AIDS orphans only might improve the targeting of needy children, but runs the risk of stigmatization. 5.58 Finally, program costs could run very high given the large number of potential beneficiaries and moderately high unit costs (3,700 - 20,000 TSh or $19 - $104 per child per year, depending on level and type of school). Implications for Policy Decisions 5.59 The preceding analysis highlights several issues of concern for policy decision makers in designing strategies for assistance to survivors. Although only a small subset of options was examined, the implications for policy decisions are generalizable to other poverty alleviation or general welfare programs, as well. 5.60 First, policy makers must invest resources in identifying beneficiaries who most need assistance and are likely to benefit from the type of help being considered. The first part of Ainsworth and Rwegarulira's paper deals with estimating the number of AIDS survivors, including widows, widowers, elderly dependents, and children under age 14 who have lost one or both parents to AIDS. Further work needs to be done within individual districts and regions. Depending on the goals of the program, policy makers may have to estimate numbers of families in specific areas who a) have been directly affected by an AIDS death; b) have been indirectly affected by AIDS (having taken in an AIDS orphan, for example); and/or c) are in poverty and in danger of not having basic human needs met, whether due to the AIDS epidemic or not. 5.61 Secondly, policy makers need to determine an effective targeting stratey for reaching the intended beneficiaries of the program while excluding those who do not qualify for benefits. As discussed by Pfeffermann and Griffin (1989), targeting of welfare programs is vital since it is "a fundamental determinant of the effectiveness and cost of the interventions." ; Ainsworth and Rwegarulira, p. 29. - 135 - 5.62 Poorly designed targeting strategies can err both in being too restrictive and in being too generous. For example, if a program is targeted only to AIDS orphans or AIDS widows/widowers one danger is that potential beneficiaries will be stigmatized, possibly resulting in further decreases in welfare as they suffer discrimination. Also, as discussed earlier, some AIDS survivors may not be as needy as other extremely poor families who have not suffered an AIDS death. On the other hand, if a program does not clearly specify requirements for entitlement, or if the requirements are too difficult and expensive to verify, benefits may be dissipated to persons who do not qualify for assistance, and the program will be less effective than anticipated.2' 5.63 A third policy issue for further analysis is the relationship between program inputs and desired outcomes. or "program effectiveness". There is a surprising dearth of information on the costs, outputs and impacts of existing survivor assistance programs.0' This makes it difficult to evaluate how effectively programs are meeting their objectives. During the design phase, policy makers need to analyze program logic, to be sure that relationships among inputs, outputs, and outcomes are plausible and to identify any assumptions which have been made. For example, a school fee subsidization program involves the following assumptions: a) subsidies will lead to increased enrollment; b) enrollment will lead to attendance; c) attendance will lead to better jobs, more income, and better welfare. Ainsworth and Rwegarulira raise questions about assumptions a) and c), noting first that subsidies may not lead to increased enrollment if orphans are an important source of labor for the household, and secondly, that attendance may not lead to better welfare if the quality of education available is not up to a minimum standard. 5.64 Further analysis of existing programs could help policy makers to gain a better understanding of how survivor assistance programs can achieve improvements in welfare of needy individuals, families and communities. 5.65 Related to the problem of quantifying program effectiveness is the quantification of resource needs and costs. Further documentation of costs of existing programs is needed, as well as careful cost analysis during the design of future strategies. 5.66 finally, policy makers must remember to weigh short-term and long-term benefits in designing strategies for survivor assistance. Immediate assistance may be needed to help people through acute crises; however, assistance programs should aim to increase self-reliance and secure the long term sustainability of programs and welfare improvements. 3' Ainsworth and Rwegarulira, p. 20. 4' Ainsworth and Rwegarulira, p. 34. - 136- VI. TIHE NATIONAL AIDS CONTROL PROGRAM A. ESTABLISHMENT OF THE NACP Background 6.01 While the first suspected cases of AIDS in Tanzania were diagnosed in 1983, it was not until 1985, when several hundred cases had appeared and been confirmed with laboratory tests, that the presence of an AIDS epidemic was confirmed. In 1985, the Government created a national AIDS Task Force (eventually to become the National AIDS Technical Advisory Committee - TAC) and sought the assistance of the World Health Organization (WHO) to assist it in formulating a Medium Term Plan (MTP) for dealing with the epidemic. A MTP for the period 1988-1992 was presented to the donor community in 1987, and the National AIDS Control Program (NACP) was officially inaugurated in April 1988. 6.02 Budgeted at almost $11 million over five years, the plan had ten objectives, which are given below by major action area: Monitoring and Research - To assess the current status of the epidemic - To monitor progression of the epidemic - To improve research activities Prevention - To decrease sexual transmission - To decrease transmission by blood transfusion - To reduce transmission by injection and other skin piercing activities - To prevent HIV infection through contact with infected materials - To reduce mother to child transmission Coping - To ensure optimal quality of life of AIDS patients - To improve diagnostic capabilities NACP Organization 6.03 The NACP, which administratively is established within the Department of Preventive Services of the Ministry of Health, began implementing the first MTP in April 1988. Four technical units were established within the NACP to guide implementation of the MTP. These units include (1) Information, Education and Communications (IEC); (2) Laboratory; (3) Clinical Services; and (vi) Epidemiology/Research. The IEC and Laboratory units focus primarily on prevention strategies; i.e., education/behavior change and blood screening, respectively. The Clinical Services unit deals with both coping strategies - 137 - (improved management of Tanzania's referral system for AIDS patients, training of health workers, estimating drug requirements) and prevention strategies (treatment of sexually transmitted diseases (STD) to reduce AIDS transmission, provision of disposable needles and other supplies to prevent infection of care givers, etc.). The Clinical Services unit was also charged with conducting research on risk factors and improving diagnostic criteria. Finally, the Epidemiology unit is responsible for the monitoring and case reporting systems, research and evaluation of control activities. A fifth unit - for Counselling and Social Support - was established in 1990 after the number of AIDS orphans gained political attention. Overall management is provided by a small Administration and Planning Unit. B. PROGRAM ACTIVITIES TO DATE 6.04 In the three years since it was established, the NACP has made considerable headway in achieving its MTP objectives. Twelve sentinel stations have been established for surveillance of HIV prevalence, and the blood of women attending pre-natal clinics is now regularly monitored for HIV antibodies. This HIV and AIDS surveillance system has increasingly given the NACP the means both to assess the current status of the epidemic and to monitor its progress. The 168 blod transfusion centers have been equipped with simple rapid kits for testing for HIV-1 antibody. The MOH estimates that about 80 percent of donated blood are now being screened prior to transfusion. NACP has improved AIDS diagnostic capabilities by issuing instructions to clinical staff in applying a modified form of the Bangui criteria for the recognition of AIDS cases. About 35 percent of reported cases are said to follow the established criteria. Protective wear and disposable needles have been distributed, and training has been initiated for health workers to reduce possible transmission by injection and other skin piercing activities. Condoms have been distributed to a number of outlets throughout the country. The Program has provided counselling for families of AIDS patients to help them cope with the burden of care and to educate them about self-protection. And it has undertaken information, education and communication campaigns to inform the public about the disease and its modes of transmission, in hopes of decreasing sexual transmission and, indirectly, mother to child transmission. It has also sponsored research to evaluate the effectiveness of its IEC efforts. 6.05 While there is no way to measure precisely the impact of NACP's efforts to date in slowing the epidemic, these activities have undoubtedly been useful. And they have set the stage for a much more concerted effort which must now be mounted. But by trying to do a little bit of everything itself - rather than concentrating on a few interventions which could have had the greatest impact, the NACP's small staff has been spread too thin and the Program has failed to mobilize, as it must, all existing institutions in the fight against AIDS. In a certain sense, most of the NACP's activities have been directed at understanding "the tip of the iceberg," that is, the part of the epidemic which is most visible or identifiable (AIDS cases, bereft families, and those already infected). If prevention is to be the main thrust of the AIDS control program - and it must be if the course of the epidemic is to be effected - it is time, now, to focus attention on the vast majority of people who are not yet infected but are at risk of acquiring this deadly disease. Fortunately, as the analysis in Chapter IV indicates, interventions do exist for slowing the epidemic's spread. 6.06 The epidemiological analysis has indicated that 80 percent of HIV transmissions in Tanzania are the direct result of heterosexual transmission. From the perspective of prevention, 90 percent of infections can be attributable to this mode, given that - 138 - most perinatal transmission is probably also the result of the mother having been infected through sexual exposure. Reducing mother-to-child transmission will therefore be accomplished indirectly by measures applied to prevent sexual transmission. Decreasing sexual transmission of the virus should therefore be the malor intervention objective of the NACP'S MTP. and this should be renec-ted in how resoure are allocatd to AIMS control, The fact that HIV has no cure makes it all the more imperative to focus the majority of limited AIDS-designated resources on preventing the disease rather than on coping with its adverse affects. 6.07 There are three interventions available to the NACP for decreasing sexual transmission of the virus: (i) prevention and control of STDs; (ii) condom promotion and distribution; and (iii) the promotion of safe (and responsible) sexual practices through effective IEC. The remainder of this chapter reviews NACP's efforts to date in undertaking these important activities and makes recommendations for strengthening NACP's capacity to ensure that these are effectively delivered, including possible organizational changes. C. SID PREVENTION AND CONTROL 1. Efforts to Date 6.08 While the MOH has, for a number of years, included the treatment of STDs in the array of clinical services offered at health facilities, it has never put in place an effective STD control program, including appropriate health education or partner notification and follow-up. In 1985, the Ministry even discontinued the regular screening of prenatal attendees for syphilis. This is particularly unfortunate given the high risk of acquiring STD's faced by women of child bearing age and the risk to both them and their unborn children if these are not detected and treated early. The rapid increase in HIV prevalence among women attending prenatal clinics confirms this view. 6.09 The importance of preventing and controlling other STD's in the fight against AIDS has been recognized in several key Tanzanian documents, including the 1987 MTP for the NACP and the updated Plan of Action produced in 1989 - where specific STD activities were proposed (training, protocol design, prenatal screening). However, little follow-up action was taken. By the end of 1990, no NACP or WHO funds had been spent on STD control activities; no additional training undertaken; no additional educational materials produced. Thus, although STD treatment continues to be available upon demand through the public and NGO health service delivery systems, the Ministry (and perhaps WHO) has not given this intervention the active attention it deserves. 6.10 The European Economic Community (EEC) is the only donor to have made a concerted effort to initiate meaningful action in this area. An EEC advisor for STDs has been working within the NACP for two years, but he has not been able to have an impact to date. Only now that his tour is about to end has a counterpart been recruited to oversee clinical STD activities. Of the various proposals for EEC funding discussed, one has been initiated: an AMREF research project in the Mwanza region is evaluating STD management feasibility and cost and STD program efficacy in reducing both STD and HIV infection. That project has just completed its first baseline survey and is in the process of analyzing the results. - 139 - 2. Planned Activities 6.11 Other proposals have recently been accepted, however, signalling an improved commitment to the prevention and control of STDs. The EEC will fund the establishment of five "STD clinics" by upgrading existing clinics in Dar-es-Salaam, Mbeya, Moshi, Zanzibar Town, and Chake Chake, and will provide drugs and train personnel to enable these five clinics, six outpatient services and 120 of the 3,300 PHC units throughout the country to improve STD care treatment a two-year period. The Oversees Development Agency (ODA) has agreed to support syphilis screening and treatment of antenatal clinic attendees, whether as a pilot or national effort has not yet been decided. Lastly, USAID, in collaboration with the AMREF-directed education program for truck drivers and prostitutes, plans to upgrade four occupational clinics and to train personnel in STD services, with PHC facilities located along the major truck routes to be included the second year. While the above activities fall short of the needed comprehensive national program, they are welcome initiatives. 6.12 While recognizing that the eradication of any STD (and HIV is no exception) is difficult because it involves changing sexual attitudes and behaviors, it should also be recognized that because HIV is facilitated by the presence of other STDs. one can have an impact on HIV transmission. This is true, first, because public and private health sector experience already exists in responding (although not always well) to demands for care of conventional STD's, and, second, because many of the very people practicing high risk sexual behavior (the target population) are making themselves visible and accessible by seeking health care for conventional STD's. 3. Condom Promotion and Distribution 6.13 As mentioned above, close to 50 million condoms donated by USAID for AIDS control have been received in the country since 1988, of which over 37 million have been distributed free-of-charge throughout the country. Over one million five hundred of these were promoted and distributed through the USAID/AMREF truck driver and prostitute project. These numbers are supplemented by the condoms provided by other donors (SIDA, GTZ) in specific regions, as well as those provided under family planning (UMATI) auspices which are distributed through the health network. NACP's early attempts to distribute condoms widely encountered serious logistic problems, but these are gradually being overcome by streamlining storage and logistic procedures. 6.14 The MOH's policy appears to be one of promoting and distributing condoms for family planning purposes, and to make them available to those who wish to use them for non-family planning reasons. Promotion of condom use against disease transmission, which would normally be incorporated into an STD control program, is not inherent within this policy. UMATI's promotional efforts have been limited to the production of a booklet. NACP has not itself promoted condom use, although several small, donor-assisted projects mentioned in Chapter IV, are promoting their use among high-risk groups, reportedly with some success. 6.15 Thus, despite the proven efficacy of condoms for protection against HIV/STDs as well as for family planning, hardly any attention has been paid to the stimulation of condom demand in the general population. Existing user rates of condoms are extremely low; within the very low contraceptive prevalence rate of about 5 percent, less than - 140 - 2 percent are estimated to be using condoms. The KABP study revealed that condom use both in and outside of marriage is negligible. More active promotion of condoms is required, though a number of issues (see below) will need to be addressed. D. IEC INTERVENTIONS 1. Resour=es Available 6.16 From the outset of the AIDS control effort, effective IEC has been recognized as the major tool available to combat the epidemic. It was selected as one of the four principal strategic areas of intervention in the MTP, and a unit was established within the NACP specifically for the purpose of providing imaginative and effective IEC leadership. Although the NACP-IEC unit would normally have been expected to work closely with the Health Education Unit (HEU) of the Ministry of Health, the latter is currently extremely weak and could not be counted on to provide the necessary technical or logistic support. The NACP-IEC unit therefore had to find its own way in designing an effective IEC campaign, and was further handicapped because of the lack of adequate baseline information concerning the epidemiological situation, sexual behavior patterns and practices, and characteristics of high risk groups. 6.17 It was initially envisaged that about half of the funds allocated to the MTP would be devoted to IEC, but as annual budgets were prepared on the basis of identifiable activities, much less was actually budgeted. In 1990, for example, only 6.3 percent of the NACP budget was initially planned for IEC activities. However, strong donor interest for AIDS education increased the funds actually made available for IEC activities to 37 percent of the funds available in that year. Over the 1988-1990 period, the IEC unit received a total of $2.94 million for its activities. About $1.69 million has been allocated for IEC in 1991. There is thus no shortage of funds for IEC, but donor regulations sometimes constrain their utilization. 6.18 Technical guidance for IEC activities is provided by the IEC Subcommittee of the National AIDS Technical Advisory Commitee (TAC), the latter being responsible for establishing overall policy for all NACP activities. There is no IEC specialist on the TAC, the composition of which is overwhelmingly medical. On the IEC Subcommittee, there is only one media specialist and no representative of youth, or of women's interests. The NACP IEC Unit, itself, is headed by a full-time national who is assisted by an information officer and a WHO/health education specialist. The current head, who recently took over from the IEC specialist initially appointed, just returned from health education and health promotion training abroad. Two posts have not yet been filled: a research coordinator and health learning materials producer. The limited staff is overstretched. 2. Activities to Date 6.19 As mentioned in Chapter IV, national IEC interventions for AIDS control and prevention of HIV transmission have been largely confined to five activities: (1) cartoons which have appeared in two national newspapers; (2) "news flashes" which appeared in three national newspapers; (3) radio spots carried over Radio Tanzania during intensive broadcast periods, (4) posters and calendars with brief messages, distributed nationwide, and (5) - 141 - counselling and social support to families of AIDS victims. Since late 1989, when an effort was made to decentralize IEC activities, these have been supplemented by seminars and film shows offered locally. 3. Effectiveness of IEC Efforts 6.20 The results of the KABP study discussed in Chapter IV indicate that the JEC interventions to date have been very effective in educating the general population about the existence of AIDS, and its main characteristics and transmission routes. That this awareness level was achieved among 95 percent of the population within two years of program commencement is a remarkable JEC accomplishment, especially in the context of Tanzania's limited mass media channels. 6.21 However, that same study, and complementary research on sexual behavior in Tanzania, point to serious deficiencies in the IEC interventions to date. First, certain vulnerable groups - namely, rural women and adolescents - remain relatively ignorant of the disease. Second, certain vital messages have not been effectively communicated, for example, that one can be infected but look healthy. Third, the IEC interventions have not yet successfully moved far enough along the continuum of awareness-knowledge-motivation- action, which is the ultimate goal of IEC for social purposes, to change behavior in a substantial risk-reducing way. This must be the objective of future IEC for AIDS control. 4. Future Directions 6.22 Research has identified the main requirements for effective IEC as being: (a) a motivational program that bridges the gap between information and behavior; (b) one that is targeted to seek out the deprived and most vulnerable audience segments; (c) message design and channel selection appropriate to varying audience needs and circumstances; (d) focus on identified knowledge gaps; and (e) imaginative use of different media. Fortunately, the recently analyzed KABP and sexual behavior studies provide substantial information, not available earlier, for moving AIDS-IEC in the appropriate direction. Effective IEC requires that developed messages are continually tested and evaluated, so that they can be altered when effectiveness has waned. This must be a permanent, ongoing part of the IEC effort. 6.23 It is particularly encouraging to note that the Government of Tanzania has, from the earliest days of its development effort, recognized the importance of educational and social communication interventions to assist in the realization of national development goals; the national experience of participatory social communication campaigns for development purposes is well known and documented. In particular there is a notable track record of public education and mass mobilization for literacy, political organization and constructive nation-building activities, including health promotion. One of Tanzania's most effective IEC campaigns in the past was, in fact, a multi-sectoral, multi-media national campaign: Mtu ni , or Man is Health, conducted in 1973, which included a synergistic combination of radio programs, print materials, extension support, community dialogue and mobilization for action. Tanzania's past therefore offers rich models for orchestrated, positive behavior changes. It is this model which will need to be followed if Tanzania is to deal successfully with the crisis which AIDS represents. - 142 - E. ISSUES/RECOMMENDATIONS 1. Give the NACP the Status it Needs to be a Truly National Program 6.24 An essential ingredient of the above approach to the AIDS crisis is strong and visible support from the very highest levels of government and the Party. The strong political support which the AIDS control program received in the mid-1980s appears to have receded in recent years. Despite the instructions that AIDS was to be a permanent agenda item at all Party meetings and that Party members were to participate in AIDS preventive work, AIDS issues have been largely left to the Party's Social Services wing. No major political statement has been made recently to endorse the program or to encourage multisectoral cooperation in according it priority. 6.25 In the absence of a powerful supra-ministerial body for coordination, the NACP is seen by all other sectors to be a Ministry of Health program. It is difficult for a sectoral ministry to exert the necessary influence to ensure intersectoral coordination or to bring about an internalization of responsibility for AIDS prevention within the existing programs of other sectors. Instead, other sectors expect "contracting" or large project support if special programs are to be put in place, an approach which is not only resource heavy and unnecessary, but beyond available resources. The most critical challenge, therefore, for the further development of the AIDS prevention effort is to make all sectors competent to take up the issue as their own, so that the existing infrastructure can become "AIDS sensitive" with only marginal additional inputs. 6.26 Several actions are recommended to make the NACP truly and "national" program: (i) With the support of multilateral and bilateral donors, including in particular WHO/GPA and the World Bank, the President of the Republic should be encouraged to make a strong statement of concern to the population about the threat of AIDS, calling for the involvement of every sector of society in its war against the disease. (ii) Each department / ministry / sector / organization should be asked to review its own programs, identify suitable intervention points, suggest required policy actions, and frame a plan of action within its existing programs which requires only marginal additional funding. Accountability would be ensured through monitoring of AIDS prevention as well as submission of reports on its primary responsibilities. (iii) A regular system to review and coordinate the implementation of this multisectoral program should be established. 2. Concentrate NACP's Efforts on Recognized Priority Areas 6.27 The time has come for NACP to concentrate its time and resources on the priority areas noted above by giving priority to implementation of a national STD/HIV prevention and control program, including effective IEC for more responsible sexual behavior. Interventions to prevent other modes of transmission and the support for coping should receive lesser emphasis and probably be moved outside the responsibility of the NACP altogether to other departments in the Ministry, other ministries in the government and NGOs. Blood screening, for example, whose importance will grow as prevalence rates increase, is appropriately overseen by the Hospital Services Department as part of its supervision of hospital laboratories in general. This will allow the NACP under the Preventive Services Department to be consistent with the Ministry's original line of thinking and allow the Department to focus on what it does best. - 143 - 6.28 Such a focus would have another distinct advantage; it would ensure that the limited funds and resources available for preventing the spread of AIDS do not get diverted to coping activities, which will - unfortunately - exert enormous pressure on resources as the epidemic grows. This implies recognition that the treatment of AIDS patients, like all others, and the establishment of appropriate drug regimes and treatment protocols, is also appropriately overseen by the MOH's Hospital Services Department. The main responsibility for home care, orphan care, and counselling on how to cope with the disease will have to be shared with the Ministry of Labor and Youth Development, community organizations and NGO's, as is done for the social effects caused by other terminal illnesses. The Ministry of Health's responsibility for counselling should probably be limited to guidance to district health teams for advising families of AIDS patients on home care, linked with treatment being offered by PHC facilities. 6.29 If the Government agrees that prevention of sexual transmission of HIV should be the major thrust of NACP's activities, consideration should be given to renaming the program to include STDs. This would accentuate the major mode of transmission and underscore the complementarity of activities. More and more countries and international organizations are following this trend. 3. Undertake an Aggressive National SID Prevention and Control Program Essential Elements 6.30 The essential elements of a national STD prevention and control program include: (i) early recognition of STDs by all health care providers (who can be trained to diagnose on the basis of syndromes); (ii) use of the appropriate therapy for treating individual STDs (through training); (iii) ensuring the continuous availability of drugs required for STD treatment; and (iv) diminishing transmission frequency by modifying sexual behavior and practicing safe sex. 6.31 It is essential that all health care providers (public, private and NGO) be included. Obvious targets are MCH/FP clinics and outpatient departments, PHC clinics and NGO facilities where STDs are treated. Location of STD services is extremely important. Because implementation of a national program must be phased over time, care must be taken to give first priority to those health care providers who encounter, or are likely to encounter, the greatest number of STD cases in a given time period. Facilities that serve the urban young adult populations should receive priority in providing STD services over political or "potential for good reporting" considerations. The agreed upon EEC phased approach will need to be followed in order to ensure implementation of quality training and supervision. The first 60 PHC units chosen should be located in urban areas that guarantee potential access to young adults in general (target core group) or to prostitutes, truck drivers and the military specifically (high risk core groups). This will require some initial research to identify where people suffering from STDs go for treatment, and the range of services currently offered. - 144- 6.32 As a minimum, an STD/HIV control program should include the following: - Diagnosis and treatment based on syndromes for urethritis, cervicitis and genital ulcer diseases in PHC units; - Routine serological screening for syphilis among MCH/FP clinic attendees and treatment, expanded to include gonorrhea and chlamydia, assuming these STDs are found to be prevalent; - IEC/counseling on STD/HIV prevention practices, including use of barrier methods and STD infected patient-referral of partners in all of the above facilities, plus in pharmacies, traditional healing services and NGOs. 6.33 A "package" for implementing STD services will need to be developed by NACP and interested parties for (a) the planned five STD/Skin Clinics providing more sophisticated levels of care with access to laboratory backup; 'o) the first 60 PHC units to be upgraded; and (c) NGO and other private providers. Each package should contain guidelines for training personnel, diagnosing and treating patients, patient IEC (with materials), counseling and follow-up of partners, record keeping, drug requirements, and management concerns including supervision and monitoring. 6.34 The cooperation of the Essential Drugs Program (EDP) must be obtained for including STD drugs into the existing EDP system of kits to ensure integration and sustainability of services. 6.35 Determining what system of partner follow-up to be practiced by health personnel and the cultural issues this involves can not be avoided. If one recognizes the importance of making the most out of this one opportunity to interrupt disease transmission and instill future behavioral change with a client, partner follow-up may require more than a reliance on the client to notify his/her partner. Operations research on the range of partner follow-up efforts to determine the cost-effectiveness of several variations is critical. 6.36 Condom use is an important factor in decreasing STD/HIV transmission, but neither the Government nor donors can afford to provide, free of charge, all the condoms necessary to cut transmission in the population. Subsidized condom distribution programs should therefore be targeted at the core group of people with frequent partner change. If current social marketing efforts and EEC campaigns are successful, however, the demand for affordable condoms will grow among the general public. The Government should consider removing the current 20 percent duty tariff and 25 percent tax on imported condoms, and negotiate with low-cost producers to make them available in-country, so that condoms are available at the lowest possible cost within the private sector. 6.37 The program needs to decide which indicators will be used to measure effectiveness of STD program efforts. The obvious number of personnel trained, clinics upgraded and patients treated is sufficient for process monitoring. One may also consider such indicators as number of patients seen per facility-catchment area served, number of partners referred and number of STD cases treated. Such information would make it possible to estimate HIV cases averted in several years time, which could become important for securing additional funding for the program. - 145 - 6.38 Assuming that the EEC-funded program (including five STD clinics, six OPDs, and 120 PHC units) is the take-off point for a national STD/HIV control program, which would gradually be expanded to cover all health care facilities, such a program would cost approximately $1.25 million annually (in 1990 dollars) about five years after initiation. Such costs assume that, by that time, 960 PHC units would have received the intensive training and supervision in STD prevention and control, that all health centers and dispensaries would be receiving the drugs needed to treat, say, the five main STDs, in their Essential Drug Program (EDP) kits as well as an algorithm on STD syndrome treatment and guidelines for IEC and partner referral counselling. Training materials would include IEC pamphlets, RPR tests for syphilis, treatment algorithms, condom demonstration kits, partner referral cards, supervisory checklists, reporting forms, etc. About $150,000 has been assumed for operational research in determining the true involvement of private sector and effectiveness of various partner notification systems. 4. Strengthen IEC for AIDS Prevention A Communication Plan 6.39 The most critical gap in the current IEC effort for AIDS prevention is the lack of a comprehensive, overall "Communication Plan" - as distinct from an itemization of activities - which would clearly state the NACP's communication objectives, goals, thematic coherence, prioritization of target groups, and messages based on research findings, as well as a planned selection of communication channels/ actors and their effective mix to create synergistic action. 6.40 Because of this gap, AIDS/IEC activities have suffered from the following problems: a) over concentration on print media materials and radio slogans - which, further lacking overall design and basis in well researched audience analysis, has led to a plateau in IEC investment returns; b) fragmented, piecemeal, project development and insufficient activity on scale. The project approach has been ad hoc and resource intensive. Although there have been several pioneering exercises, they have not been linked to clear cut plans for upscaling and replicating successful experiments in a cost-efficient manner. This has also encouraged the approach that AIDS/IEC needs to be "sponsored", rather than taken up as a cause to be pursued. c) lack of a systematic activation of the basic health and education infrastructure, and hardly any involvement of key community groups beyond the top and middle levels of Party echelons. 6.41 A wealth of data providing important communication insights for formulating an appropriate new approach is now available through the KABP and partner relations studies, as also the number of pilot projects conducted in the country. It is recommended that top - 146- priority should be accorded to (a) the analysis of this research data within a strictly time-bound two-month period; (b) simultaneous collation of major project findings. The IEC unit should interact closely with the Behavioral Sciences Department at Muhimbili to ensure timely and pertinent information being tabulated. This information should form the basis for developing a comprehensive communication plan, delineating objectives, goals, strategies, priority target groups and messages, media mix and design, channels/actors/roles etc. The Population Communications Services of Johns Hopkins University, already working with HEU, has successfully assisted other African countries, including Zimbabwe, in this regard and could be asked to assist with this work. The IEC Unit within NACP 6.42 Ideally, IEC leadership for AIDS education should come from within a strengthened HEU chartered to link AIDS education with every major IEC activity unidertaken by MOH; most particularly in conjunction with the reactivization of the PHC movement. However, HEU's strengthening, although planned, is a number of years away from being a reality. In the meantime, AIDS/IEC has to be revamped, accelerated and expanded. It is therefore essential to not only continue the AIDS IEC Unit, but to further strengthen it to ensure a strong, central core capable of providing overall direction to IEC and to stimulate multisectoral involvement. 6.43 Among the necessary ingredients of a strerigthened NACP-IEC is the resolution of outstanding staffing issues. First, high priority should be accorded to filling the two vacant posts of research coordinator and HLM producer so that a full complement of a professional communication team is available in-house. Second, a two-year extension should be requested for the WHO health education specialist, whose community extension experience is relevant for the next phase and whose presence will provide important continuity at a time when a newly trained officer is at the helm. Third, contractual arrangements should be considered with the former Head of the Unit to assist the coordination of research for IEC until such time as a full-time suitable research coordinator can be positioned. Fourth, short term consultancies should be utilized as required, most particularly to bring in skills for communication planning and audio-visual software deveiopment, which are identified deficiencies of the program. However, such independent consultancies should only be utilized after exhausting opportunities to use expertise jointly with other allied programs, such as HEU, MCH/FP, and FPCP, in order to develop common communication strategies, materials, and convergence of programme actions. The terms of reference of all consultants recruited by NACP/IEC, HEU or MCH/FP should integrate the needs of the others' programs. AIDS Technical Committees 6.44 There is no media/IEC specialist on the TAC and the number of members associated with social development aspects are limited in comparison to those with medical backgrounds. The IEC Subcommittee has only one media representative and does not have anyone representing Information and Broadcasting, nor the critical target groups of women, youth, and organized labor. It is recommended that the composition of both Committees be reviewed and their membership revised/enlarged to ensure appropriate representation from the Ministry of Information and Broadcasting, Radio Tanzania, the Ministry of Community Development, Women Affairs and Children, and the Ministry for Labour and Youth - 147 - Development. Two meetings a year of the IEC Subcommittee should be specifically earmarked to assist in communication planning and strategy formulation/review. Decentralization of IEC 6.45 The review of the MTP in mid 1989 highlighted the need to decentralize AIDS-IEC, and it was decided to appoint one of the existing health officers or health assistants in each district as District AIDS Coordinator (DAC), to be charge of all AIDS control activitiec in their respective jurisdictions. These individuals were theoretically to be released from their regular clinical duties so that they could concentrate on AIDS-related activities, primarily the stimulation and coordination of IEC efforts at the local level. Communication skills training which was initially envisaged was, however, never carried out. 6.46 In practice, the time and attention which DACs have devoted to IEC has been patchy; both their orientation as clinical personnel and the demands of program management and patient care/counselling have pushed preventive IEC into the background, leaving a serious void at this level. There is critical need for a full-time health educator at the district level, someone who could develop a strategy for ensuring that crucial AIDS EEC messages developed nationally are effectively communicated to the local population. Just as the NACP should probably not devote its scarce resources to activities better handled by others, the DACs should probably not attempt to coordinate all AIDS activities at the district level, but allow the other institutions and systems meant to handle these activities (patient counselling, record keeping, etc.) to do so. It is recommended that DACs be transformed into full-time AIDS educators through appropriate communication skills training, with the intention that they would form the nucleus of health educators at the district level once the HEU was successfully decentralized. More Effective Electronic Software Development 6.47 Radio has provided the most effective means of dissemination in raising awareness of the AIDS problem - over 60 percent attribute it as a source of information. However, more versatile use of this powerful tool is now required. Special programs, specifically tailored to different audience segments, using non-prescriptive, more popular formats - such as radio soaps, jingles, a regular question and answer session, field based reporting - need to be developed. The skills exist within Radio Tanzania, Institute of Adult Education, and UMATI for the development of such programs, but need technical support; i.e. the key messages to be incorporated, the profiles of the audience des:red to be reached, etc. Radio Tanzania has just completed an Audience Analysis Survey in collaboration with the BBC. This analysis should be utilized to target programs more effectively to specific audiences, as also to select the most appropriate and effective broadcasting times for different categories of listeners. Radio Tanzania is willing to provide the time and production facilities. Its cooperation must be maximized by full technical support from the EEC Unit as a top priority in its work; also the inclusion of Radio Tanzania on the IEC Subcommittee. Development of a Multi-Media Campaign on AIDS 6.48 The Institute of Adult Education has the mandate to mount campaigns on subjects identified as national priority. Its multi-media campaigns using radio study groups, print communication materials and extension community channels over the seventies and early eighties have been landmarks in development communication. While some of the earlier - 148 - pioneers have left the Institute, there remains sufficient experience to revive the concept, given sufficient advance planning and support. Advocacy efforts need to be made to identify the AIDS issue as a key national priority requiring the mounting of a multi media multi-sectoral campaign by the NACP and Institute of Adult Education in collaboration with Radio Tanzania, using the schools, adult education groups and village committees' networks. This avenue should be fully explored to develop a powerful nationwide debate on the moral and social issues surrounding AIDS. The Department of Behavioral Sciences at Muhimbili should also be actively involved in the development of the campaign and communication materials. The campaign should further utilize AIDS films as a part of the multi-media strategy. Film Development 6.49 The Audio Visual Institute and the Tanzania Film Corporation have the capability to create effective film software. Films can be a good initiator of public discussion. With the availability of film projectors in every region being ensured this year, there will be need for more visual software. Funds should be allocated for the development of more visual materials. More Emphasis on Youth 6.50 Youth (10-24) constitute 30 percent of the total population of Tanzania. Small studies have observed high levels of sexual activity among the youth, including those in schools. The studies also reveal that reproductive ignorance is fairly prevalent and the potential health and social consequences alarming. Analysis of the age and sex distribution of early AIDS victims in Tanzania, given what is known about the incubation period, suggests that almost half the patients were infected before the age of 20. Dramatic increases in teenage HIV sero-prevalence rates have been observed between 1987-90. The need for sex education for the young is therefore critical. Resistance among school authorities to this has only recently been overcome. 6.51 There are currently three projects underway: the first, to include family life education (FLE) in the curriculum, a second to include AIDS education in the curriculum, and a third to undertake an AIDS campaign in the schools, but these are proving to be parallel exercises. All three are experiencing considerable delays and modalities for implementation have not been worked out. Very little has been done to date to reach out-of-school youth. 6.52 Youth must be given priority as the single most important target group to be reached. The effort should be systematic and comprehensive to cover all higher-primary-and- above school goers (including technmcal) with information suitable to their age and level of understanding; existing and potential school teachers; and out-of-school youths, reached through special programs with youth economic networks, the National Service Scheme, and NGOs. An imaginative, intensive multi-media campaign should be designed and directed to the teenage group. 6.53 In this connection it is recommended that the following steps be taken: (i) integration of the FLE and AIDS education curriculum prior to the completion of the FLE pilot testing and revision and early development of proposals for its nationwide institutionalization in the school system. - 149- (ii) steps to ensure timely printing of materials developed for the AIDS school campaign and finalization of the modalities for its implementation, in order to ensure it is accorded highest priority by the educational authorities and its implementation within the current year. (iii) A campaign for human values and responsible sexuality should be initiated through the involvement of the press, radio and popular folk groups/participatory theatre groups, so that the school discussions are supported by a backdrop of socio-cultural ferment. The development and recording of an upbeat, motivational theme song by a nationally popular singer for frequent use on the radio - as has been most effectively promoted in several Latin American, Asian and African countries for the family planning program by the JHU/PCS services - should be attempted. (iv) It is most essential that school education efforts are supplemented by channels of interpersonal contact and that further, special programs reach out-of-school youth. It is recommended that Tanzania explore the possibility of organizing mobile communication and counselling teams, through NGO efforts, for a program of school visits and group contact, backed by static counselling centers in at least major cities - on the lines of Zimbabwe's successful Youth Guidance program. Condom Promotion 6.54 The issue of condom promotion is fraught with dangerous emotions - there is considerable, open opposition from religious leaders and elders who see the condom as a message for promiscuity and corruption, particularly of the young. This image of the condom in Tanzania is influenced by its availability and use almost exclusively, so far, at bars and guest houses. The separation of the AIDS and the family planning program efforts have further reinforced the condom's negative connotations. The matter therefore needs to be handled most delicately. There is a need for a carefully considered IEC strategy for its promotion to adults in general in a way that can combine its benefits for family planning as well as AIDS/STD prevention. The promotion of condom use by very high risk people, as under the AIDSTACH project, should be seen and treated as a different exercise, so as to distance it from that formulated for the general population. The cost-effectiveness analysis presented in Chapter IV indicates that one can target ':ery high risk groups (the core population) with discrete but intensive. one-on-one promotion techniques with high payoff. 6.55 Condom promotion to the young, as advocated by some donors and intellectuals, may be justified in the context of the high levels of sexual activity noted among adolescents. However, to even moot such a strategy at this juncture, when a breakthrough has finally been made in promoting FLE and AIDS education in schools, would be to unleash a backlash that might create a considerable setback to such efforts in general. Counselling services for the young at sites away from the schools can be the only strategy to reach the young who remain sexually active, notwithstanding moral education in schools. 5. Strengthen Intra-Sectoral Coordination within MOH 6.56 The Ministry of Health has for some years been struggling with the need to integrate its various programs more effectively in order to attain its primary health care goals. The existence of numerous vertical programs, each with its own budget and supervisory links - 150- with field staff, has complicated headquarter's efforts to train staff, collect data, design IEC, and implement health strategies in an integrated and cohesive fashion, and has hindered rather than supported the ability of districts to plan and implement their own health development programs. 6.57 Within the NACP, the lack of adequate intra-sectoral coordination has adversely affected Program goals in several ways; for example, (i) AIDS training of health workers has been ad hoc, and has not yet been incorporated into either the basic training or continuing education curricula; (ii) the NACP encountered serious problems when it first attempted to distribute condoms to remote sites all over the country, despite the fact that a well-functioning system for distributing drugs and medical supplies already existed under the Essential Drug Program; (iii) as mentioned above, pregnant women are not being screened for nor counselled on STDs under the MCH/FP program, thereby missing the best possible opportunity for one-on-one AIDS counselling to almost half of the adult population; (iv) AIDS-IEC has not been integrated with other health education messages; and (v) more attention has been given to high-cost blood screening than to reducing the need for transfusions by better malaria treatment and control. 6.58 Within MOH, intrasectoral coordination on AIDS is meant to be achieved through a National AIDS Management Committee, chaired by the NACP Coordinator. To improve intrasectoral coordination, the MOH is considering re-establishing its Primary Health Care Committee, under the chairmanship of the Chief Medical Officer. A high priority of this committee should be the review of the scope for coordinated planning and implementation of components of the MCH/FP and NACP, in order to integrate training and IEC activities. Progress should be reviewed quarterly. 6.59 The current weaknesses in program coordination are well recognized within the Ministry of Health, which has committed itself to a reorganization which would integrate all vertical programs at the central level among the general public health functions of epidemiology, training, laboratory services, health education, etc. One major aspect of this reorganization will be the strengthening of the HEU, with DANIDA support, including the establishment of a health education infrastructure at the district level under a Health Education Project (HEP). While the NACP-IEC cannot expect much support from the HEU until completion of this project, it will be essential that the immediate strengthening of NACP-IEC is consistent with its eventual integration into the new HEU. 6.60 Among the steps which should be considered to ensure the eventual integration of NACP-IEC with the strengthened HEU are the following: - establishment of a suitable mechanism to ensure that the Head of HEU and the DANIDA advisor to the HEP (when in position) are regularly briefed on developments in AIDS-IEC and their compliance/guidance secured for measures with longer term implications. Joint planning of annual work plans should be undertaken to avoid duplication or compartmentalization of efforts. - nomination of the DANIDA advisor to the TAC's ECr Subcommittee. - the eventual evolution of the TAC into the high-level Health Promotion Committee which is envisaged under the HEP. - 151 - - recognition that field structures should eventually be integrated. 6. Strengthen Donor Coordination 6.61 In addition to funding provided through WHO/GPA, AIDS activities have received ad hoc support of about $4.8 million from multilateral and bilateral donors. This has made integrated planning and implementation difficult and is causing a heavy administrative load, particularly with the necessity to monitor some donors' expenditures separately. Closer donor coordination, particularly among those most involved - e.g. DANIDA, USAID, UNFPA, UNICEF, UNDP and WHO - will be essential. Ideally, agreement will be reached among all donors on the next MTP. 7. Decentralize AIDS Planning 6.62 It is increasingly recognized that effective implementation of a PHC strategy depends to a large extent on the involvement of the local population in planning and implementation. In Tanzania, this means shifting the focus of health planning from the center (MOH headquarters) to the districts, who are finally responsible - through the district health team - for program implementation. Tanzania has a long history of decentralized planning initiatives. Most recently, in 1983, responsibility for development planning was formally shifted to the district administration, with the creation of the Ministry of Local Government and the decision to allocate a large proportion of the development budget directly to the districts. Progress has been slow, however, in building the districts' capacity for development planning, because district authorities lack both the experience and the tools for making the hard choices which planning within a constrained environment demands. 6.63 Unlike other health programs (for example, diarrheal disease control), where a standard response (oral rehydration therapy) can be defined for countrywide implementation, STD/HIV control programs cannot rely exclusively on a single intervention to control the epidemic. As demonstrated in Chapter IV, the cost-effectiveness of various interventions depends dramatically on the prevalence of HIV (and other STDs) in the population in question. HIV prevalence among adults varies significantly among Tanzanian's 100+ districts, from less that one percent to 17 percent. Even within districts, HIV prevalence is known to vary significantly. Similar variations occur for other STDs as well. While there is generally an association between STD and HIV levels, there are some areas with relatively high STD levels and (as yet) low HIV levels. A priority must be to prevent and control STDs in such regions in order to slow transmission of HIV. 6.64 Once interventions have been selected for STD/HIV control, as well as for other health program, their implementation must be carefully planned to take into account local constraints and opportunities. Among the factors which must be considered in designing implementation activities are the following: - access; i.e., communication facilities; - existing public and private health infrasEructure; - educational levels of the population, and education facilities; - income levels of the population; - socio-cultural norms and beliefs of the population; and - health-seeking behavior of the population. - 152 - 6.65 These factors influence both promotion of the epidemic and controlling it. "Access," for example, can be both negative and positive: good roads facilitate the movement of staff and materials, but may also encourage the presence of high risk groups, or facilitate movement to and from high risk areas. Access varies significantly among districts, as do the other factors listed. 6.66 Combatting the AIDS epidemic will require behavior change, which can only be achieved if the individuals and communities are themselves involved in planning for that change, preferably from the earliest stage. It will require a coordinated effort from all sectors of the community. And, because AIDS will inevitably be a long-term problem, it will also require a sustained response, not exclusively dependent on external resources. All of this argues for focussing planning for AIDS at the district level. 6.67 The Ministry of Local Government has a decentralized planning structure and system in place at the district level which incorporates village level inputs into an intersectoral planning process. This planning process is being strengthened in collaboration with various donors (IDA, DANIDA, UNICEF, GTZ). The IDA-supported Health and Nutrition Project, for example, includes a pilot program in ten districts, whereby district health plans are being developed and funded, including PHC initiatives undertaken by the communities themselves. Assistance in being provided to district authorities in commnunity mobilization, management, planning, financial control and other key areas. The Government (MOH and Ministry of Local Govermment) has expressed interest in building on this experience by undertaking an AIDS planning exercise with selected districts, utilizing the information developed in the background studies for this report on the relative cost-effectiveness of different interventions to slow the spread of HIV. 6.68 A manual has therefore been developed for testing with selected district representatives in a workshop environment. If successful, this exercise would be the basis for developing draft guidelines for implementing STD/HIV control programs at the local level, for possible use in Tanzania and elsewhere. - 153 - VII. RECOMMENDATIONS 7.01 A number of recommendations have been made throughout this report. They are summarized here for ease of reference. The reader is directed to the main chapters of the report for a full discussion of the relevant issues. 1. 'The NACP should give highest priority and devote the major portion of its human and financial resources to those interventions which are most likely to have the greatest impact on slowing the epidemic; that is, to a comprehensive, national STD/HIV prevention and control program. NACP's organizational structure should facilitate this thrust, and available interventions should be promoted with confidence and commitment. (paras. 4.24-4.35, 6.27-6.38) 2. The proven effectiveness of condoms to prevent STD/HIV transmission should be acknowledged, and their use should be actively promoted for disease prevention as well as for family planning, using techniques and messages specifically designed for different target groups. (paras. 4.36-4.60) 3. IEC for AIDS prevention and control should be given new life by making better use of existing technology to develop a multi-media and multi-sectoral campaign, and by incorporating more effectively lessons learned from Tanzanian and international experience. The aim should be to raise national consciousness of the seriousness of the epidemic and to create a collective will to alter its future course. Within an overall communication plan, IEC messages should be specifically designed for and directed at different target groups identified, through research, as being particularly at risk or uninformed. While the biggest payoff in the short run will be messages directed to prostitutes and their clients, high priority should also be given to a campaign directed at the youth - both in and out of school. Such a campaign should include a revised curriculum at all levels of the education system, integrating family life education and AIDS education; a campaign for human values and responsible sexuality involving the press, radio and popular entertainers; and interpersonal contacts. (paras. 6.39-6.55) 4. The increased demand for health care brought about by the AIDS epidemic should not be allowed to divert resources from preventive health measures, both those to prevent AIDS and those directed at the many other preventable conditions which present the greatest threat of illness and death to the vast majority of the population. All possible measures should be taken to keep the cost of AIDS treatment low, by limiting drug use, by strict adherence to treatment protocols, and by caring for patients at peripheral facilities and at home. In addition, the budgeting process must build in safeguards to protect moneys allocated for "preventive" health programs. (para. 3.44, paras. 5.18- 5.21) 5. Programs to assist the survivors of the AIDS epidemic must be carefully screened to ensure that they are affordable, that they reach those most in need, and that they are equitable. Evidence to date suggests that well-designed poverty alleviation programs which depend upon community participation and commitment have a greater probability of reaching the most vulnerable members of the community and of being sustained than programs directed at AIDS survivors, per se. The NACP should WI - 154 - have responsibility for providing such assistance, which should rather be integrated with other programs for those in need of community support. (paras. 5.59-5.66, para. 6.28) 6. The Ministry of Health should examine its many programs to ensure that all parts of the Ministry are aware of the role they should play in the fight against AIDS. Much more could be accomplished under the MCH/FP program, for example, by providing counselling to women of child-bearing age on STD prevention and control, and by strengthening the treatment of malaria in women and children so that fewer blood transfusions are required. The Ministry should move forward with its plans to integrate its vertical programs at headquarters, so that the requirements of the NACP for better health education, data collection, training of service providers, etc. can be fully integrated with and support a cohesive primary health care strategy. (paras. 6.56-6.60) 7. Assistance should be provided to districts to enable them to plan better how to utilize available financial and human resources effectively in the fight against AIDS. (paras. 6.62-6.68) B. The Government and donors should recognize the AIDS epidemic in Tanzania for the crisis it is, but one which can be combatted with forceful and committed action on their part. The highest levels of Government should support the fight against AIDS in highly visible and constructive ways in order to mobilize all segments of the population in a concerted effort to slow the epidemic's progress, recognizing that the National AIDS Control Program, although administered by a small unit in the Ministry of Health, is a national effort which touches all sectors of the economy and all segments of society. (paras. 6.24-6.26) 9. The Government should pursue its economic reform program vigorously and determinedly. The negative macro-economic consequences of the epidemic will be minimized to the extent that price distortions are removed. And only with substantial real growth can the economy generate the public revenues required to maintain the current public health effort, including that necessary to prevent the spread of AIDS, while meeting even a small proportion of the costs of caring for AIDS patients. (paras. 3.43-3.44) 10. Donors, for their part, should be prepared to increase their assistance to Tanzania to help counter the adverse impact which reduced public and private saving is likely to have on economic growth. The Bank should estimate the magnitude of these increased requirements as part of its economic work and stand ready to help mobilize the necessary funds provided the economic reform stays on track. Particular attention should be directed at human capital formation, not only to replace productive members of society who die prematurely, but also to plan now for the loss of skills which are normally acquired only through years of work experience. (paras. 3.21, 3.45-3.55) - 155- BIBLIOGRAPHY Abel, Nick, Tony Barnett, Simon Bell, Piers Blaikie and Sholto Cross. "The Impact of AIDS on Food Production Systems in East and Central Africa Over the Next Ten Years: a Programmatic Paper." in Fleming, A., et. al. (eds.) The Global Impact of AIDS. New York: Alan R. Liss, Inc. 1988. p. 145-154. Ainsworth, Martha, and A. A. Rwegarulira. "Coping with the AIDS Epidemic in Tanzania: Survivor Assistance." Background Paper prepared for the Tanzania AIDS Assessment and Planning Study. Washington, D.C.: World Bank. December 1991. Anderson, R. M., R. M. May and A. R. McLean. "Possible demographic consequences of AIDS in developing countries." Nature. 332:228-234. March 17, 1988. Becker, Charles M. "The Demo-Economic Impact of the AIDS Pandemic in Sub-Saharan Africa," World Development, Vol. 18, No. 12., pp 1599-1619. Great Britain: Pergamon Press. 1990. Belsey, Mark A. and Erica Royston. "A global overview of the health of women and children." in Wallace et. al. (eds.) Maternal and Child Health Practices. Third Edition. Oakland, CA: Third Party Publishing Company. 1988. Bertrand, Jane T., Bakutuvwidi Makani, Susan E. Hassig, Kinavwidi Lewu Niwembo, Balowa Djunghu, Mbadu Muanda, and Chirwisa Chirhamolekwa. "AIDS-related knowledge, sexual behavior and condom use among men and women in Kinshasa, Zaire." American Journal of Public Health. January 1991. 81 (1): 53-109. Bertozzi, Stefano, Maxine Ankrah and Magdalene Ngaiza. "Tanzania: Assistance to Survivors of the AIDS Epidemic, A Review of the Policy Options." Background Paper prepared for the Tanzania AIDS Assessment and Planning Study. Washington, D.C.: World Bank. June 1990. Bitran, Ricardo. "A Household Health Care Demand Study in the Bokoro and Kisantu Regions of Zaire: Volume II, Patterns of Utilization." Report prepared for the Resources for Child Health Project (REACH). Arlington, VA: REACH. October, 1989. Bjorkman, Anders. "Alternative Medical Interventions to Slow the Spread of HIV Infection." Background Paper prepared for the Tanzania AIDS Assessment and Planning Study. Stockholm, Sweden: Department of Infectious Diseases, Karolinska Institute, Roslagstull Hospital. Revised draft, January 25, 1991. Bulatao, Rodolfo A. "The Demographic Impact of AIDS in Tanzania." Background Paper prepared for the Tanzania AIDS Assessment and Planning Study. Washington, D.C.: World Bank. June 1990. Bulatao, Rodolfo A. "Estimates and projections of mortality by cause: a global overview, 1970-2015." Draft, December 1989. Chapter 2.1 in Jamison, Dean T. and W. Henry Mosley, eds. Evolving Health Sector Priorities in Developing Countries. Washington, D.C.: World Bank, forthcoming. - 156- Cabral, A.J.R. "Keeping donor support for the next phase of NAPs in the Third World (and Sub-Saharan Africa in particular): the additional demands created by the increasing number of AIDS patients and medical/social care to them." Paper prepared for GPA/WHO-HQ. Feb-March 1990. Carballo, M. and Carael, M. "The Impact of AIDS on Social Organization." in Fleming, et. al. (eds.) The Global Impact of AIDS. New York: Alan R. Liss, Inc. 1988. pp. 81-93. Center for International Health Information. "Demographic Indicators," in their country series of USAID Population and Health Profiles. Arlington, VA: ISTI. 1990. Chin, James and F. Sonnenberg. "The Epidemiology and Projected Mortality of AIDS in the United Republic of Tanzania." Background Paper prepared for the Tanzania AIDS Assessment and Planning Study. Washington, D.C.: World Bank. No date (revised draft received Jan. 30, 1991). Chin, James. "The Epidemiology and Projected Mortality of AIDS." Chapter 13.1 in Feachem, R.G. and D.T. Jamison, eds. Disease and Mortality in Sub-Saharan Africa. Volume II. Washington, D.C.: World Bank. 1989. (forthcoming in 1990 from Oxford Univ. Press) Chin, James. "HIV and International Travel," in Fleming, et.al. (eds.) The Global Impact of AIDS. New York: Alan R. Liss, Inc. 1988. p. 61-65. Cooper, Ellen C. "Treatment Issues in AIDS." Chapter 9 in Corless, I. B. and Pittman- Lindeman M., eds. AIDS: Principles. Practices. & Politics. Washington, D.C.: Hemisphere Publishing Co. 1988. Cuddington, John T. "Modelling the Macroeconomic Effects of AIDS with an Application to Tanzania. " Background paper prepared for the Tanzania AIDS Assessment and Planning Study. Washington, D.C.: Georgetown University, August 19, 1991. Foote, Dennis. "A Cost/Effectiveness Model for Comparing Interventions in Information, Education and Communications About AIDS in Tanzania." Background Paper prepared for the Tanzania AIDS Assessment and Planning Study. Menlo Park, California: Applied Communication Technology. October 1990. Greenberg, Alan E., Phuc Nguyen-Dinh, Jonathan M. Mann, Ndoko Kabote, Robert L. Colebunders, Henry Francis, Thomas C. Quinn, Paola Baudoux, Bongo Lyamba, Farzin Davachi, Jacquelin M. Roberts, Ngandu Kabeya, James W. Curran, Carlos C. Campbell. "The association between malaria, blood transfusions, and HIV seropositivity in a pediatric population in Kinshasa, Zaire." Journal of the American Medical Association. 259(4):545- 549. 1988. Heymann, Sally Jody. "The Future of AIDS Prevention and Control Programs: Preparing for AIDS Vaccines and Treatment." Report prepared for WHO GPA. Center for Population Studies, Harvard University. Boston: December 28, 1990. - 157 - Heald, Gary R. "Estimating the Cost/Effectiveness of AIDS Information, Education and Communication Interventions in Tanzania." Background Paper prepared for the Tanzania AIDS Assessment and Planning Study. Tallahassee, Florida: Communication Research Center. March 5, 1991. Hunt, Charles W. "Migrant Labor and Sexually Transmitted Diseases: AIDS in Africa." Journal of Health and Social Behavior. 1989:30 (December), p. 353-373. Jager, Helmut, Bosenge N'Galy, Joseph Perriens, Kifuani Nseka, Farzin Davachi, Claire- Mulanga Kabeya, Gertrud Rauhaus, Gabriele Peyerl, Robert W. Ryder, and Thomas Rehle. "Prevention of transfusion-associated HIV transmission in Kinshasa, Zaire: HIV screening is not enough." AIDS. 4(6):471-574. 1990. Jamison, Dean T. and W. Henry Mosley. "Selecting disease priorities in developing countries." Third Draft, July, 1990. Chapter 1 in Jamison and Mosley, eds. Disease Control Priorities in Developing Countries. Washington, D.C.: World Bank, forthcoming. Killewo, Japhet, Klinton Nyamuryekunge, Anita Sandstrom, Ulla Bredberg-Raden, Stig Wall, Fred Mhalu, and Gunnel Biberfeld. "Prevalence of HIV-1 infection in the Kagera region of Tanzania: a population-based study." AIDS. 4(11):1081-1085. Nov. 1990. Kirkwood, Betty R. "Diarrhoea." Chapter 9 in Feachem, R.G. and D.T. Jamison, eds. Disease and Mortality in Sub-Saharan Africa. Volume II. Washington, D.C.: World Bank. 1989. (forthcoming in 1990 from Oxford Univ. Press) Kirkwood, Betty R. "Acute respiratory infections." Chapter 10 in Feachem, R.G. and D.T. Jamison, eds. Disease and Mortality in Sub-Saharan Africa. Volume II. Washington, D.C.: World Bank. 1989. (forthcoming in 1990 from Oxford Univ. Press). Lepage, Philippe and Philippe Van de Perre. "Nosocomial transmission of HIV in Africa: what tribute is paid to contaminated blood transfusions and medical injections." Infection Control and Hospital Epidemiology. 9(5):200-203. 1988. Leshabari, M.T. "Partner Relations and Risk of HIV Infection in Tanzania (Preliminary Report)." Paper presented at the workshop on the development of the Second Medium Term Plan for AIDS Control. Kibaha, Tanzania: May 1991. Leshabari, M. and G.K. Lwithula. "AIDS Education for School Children: a Challenge to Teachers and Parents." Paper presented at a seminar on AIDS, Women and Children. Dar- es-Salaam, Tanzania: March 1990. Lopez, Alan D. "Causes of Death in the Industrialized and the Developing Countries: Estimates for 1985." Health Sector Priorities Review Paper HSPR-20. Washington, D.C.: World Bank. November 1990. MacMahon, Brian and Thomas F. Pugh. Epidemiology: Principles and Methods. Boston: Little, Brown and Co. 1970. - 158 - Micklin, Michael and David S. Sly. "International Population Movements and AIDS: Patterns, Consequences and Policy Implications." in Fleming, et. al. (eds.) The Global Impact of AIDS. New York: Alan R. Liss, Inc. 1988. p. 67-77. Moses, Stephen, K.N. Ngugi, N.J.D. Magelkerke, M. Bosire, P. Waiyaki, F.A. Plummer. "Cost-effectiveness of an STD/AIDS Control Programme for high frequency STD transmitters in Nairobi, Kenya." Abstract F.D. 837. Sixth International Conference on AIDS, San Francisco, CA. 22 June 1990. Muhondwa, E.P.Y. "Review of KABP Findings, (Preliminary Report). Paper presented at a workshop for development of the second medium-term plan for AIDS control. Kibaha, Tanzania: May 1991. Neequaye, A.R., L. Osei, J.A.A. Mingle, G. Ankra-Badu, C. Bentsi, A. Asamoah-Adu and J.E. Neequaye. "Dynamics of Human Immune Deficiency Virus (H1V) Epidemic-the Ghanaian Experience." in Fleming, et. al. (eds.) The Global Impact of AIDS. New York: Alan R. Liss, Inc. 1988. p. 9-15. Mujinja, P., G. Koda, G. Lwihula, I. Semali, M. Ainsworth and M. Over. "Coping with the Economic Impact of Fatal Adult Illness: An Analytic Framework," Dar es Salaam: University of Dar es Salaam, Tanzania, and Washington, D.C.: World Bank, July 1991. Nguma, J.K., M.T. Leshabari, and G.S. Mpangile. "Perceived Risk and Adoption of Protective Behaviour Against HIV Infection Among Truck Drivers, Casual Labourers and Post Secondary School Pupils in Dar es Salaam." Paper presented at a worlkshop for development of the second medium-term plan for AIDS control. Kibaha, Tanzania: May 1991. Nyamuryekung'e, K.M. "National AIDS control programme in Tanzania: achievements, constraints and future perspectives." Speech presented to the participating parties meeting, Tanzania Room, Kilimanjaro Hotel, 25th October 1990. Over, Mead. "Costs and Effects of Blood Screening and Social Marketing of Condoms as Strategies for AIDS Prevention in Tanzania." Background Paper prepared for the Tanzania AIDS Assessment and Planning Study. Washington, D.C.: World Bank. April 10, 1991. Over, Mead. "Costs and Effects of STD Treatment as a Strategy for AIDS Prevention in Tanzania." Background Paper prepared for the Tanzania AIDS Assessment and Planning Study. Washington, D.C.: World Bank. February 12, 1991. Over, M., M. Ainsworth, P. Mujinja, G. Koda, G. Lwihula, and I. Semali. "The Economic Impact of Fatal Adult Illness in Sub-Saharan Africa: A Research Proposal," Washington, D.C.: World Bank and Tanzania: University of Dar es Salaam, November 1989. Over, Mead and Peter Piot. "HIV Infection and Other Sexually Transmitted Diseases." Draft, April 1990. Chapter 10 in Jamison, Dean T. and W. Henry Mosley, eds. Evolving Health Sector Priorities in Developing Countries. Washington, D.C.: World Bank, forthcoming. -159- Over, Mead, Stefano Bertozzi and James Chin. "Guidelines for rapid estimation of the direct and indirect costs of HIV infection in a developing country." Health PolicY. 11, pp. 169- 186. 1989. Over, M., S. Bertozzi, J. Chin, B. N'Galy, and K. Nyamuryekung'e. "The Direct and Indirect Cost of HIV Infec'on in Developing Countries: The Cases of Zaire and Tanzania." In Fleming, A., at. al. (eds.) The Global Impact of ADS, pp. 123-135. Alan R. Liss, Inc. 1988. Pallangyo, K. and R. Laing. "Background Study on Alternative Approaches to Managing the Opportunistic Illnesses of HIV-Infected Persons: Costs and Burden on the Tanzanian Health Care System." Background Paper prepared for the Tanzania AIDS Assessmenz and Planning Study. Tanzania: University of Tanzania Faculty of Medicine, and Boston, Massachusetts: Management Sciences for Health. September 1990. Russell, Sharon Stanton, Karen Jacobsen and William Deane Stanley. "Internal Migration and Development in Sub-Saharan Africa, Volumes I (Overview) and II (Country Analyses)." World Bank Discussion Papers 101 and 102, Africa Technical Department Series. Washington, D.C.: World Bank. 1990. Rwegarulira, A.A. "Strategies for Community-Based Care: A New Approach for Service Delivery and Care of Orphans." Paper presented at the Regional Officers' Workshop on the Impact of AIDS, Tangra, Tanzania. October 1991, mimeo. Scitovsky, Anne A. and Mead Over. "AIDS: costs of care in the developed and the developing world." AIDS. 2 (suppl 1) 1988, pp. S71-S81. Smith, William A. "AIDSCOM technical assistance for Tanzania: trip report to review condom promotion opportunities in Tanzania and -'iake recommendations for future activity." Washington, D.C.: Academy for Educational Development. Aug. 14-17, 1988. Tapia-Conyer, R., Martin, A., Revuelta, A., and Rodriguez, S. "The Economic Impact of AIDS at the Household Level in Mexico." Abstract F.D. 839. Sixth International Conference on AIDS, San Francisco, CA. 22 June 1990. Thompson, Mark S. Benefit-Cost Analysis for Program Evaluation. Beverly Hills, CA: Sage Publications. 1980 UNICEF. State of the World's Children 1990. Oxford: Oxford University Press. 1990. United Republic of Tanzania, Ministry of Health. "National AIDS Control Programme AIDS Surveillance." Report No. 4, March 1991. Dar-es-Salaam, Tanzania: NACP Epidemiology Unit. March 1991. Vian, Taryn. "Trip Report: Economic and Financial Aspects of the Condom Social Marketing Project of the AIDS Control Programme in Cameroon." Research Triangle Park, N,C.: AIDSTECH Project, Family Health Intervational. April 1989. - 160 - Volberding, Paul A. "AIDS Overview," Chapter 8 in Corless, I. B. and M. Pittman- Lindeman, eds. AIDS: Principies. Practices and Politics. Washington, D.C.: Hemisphere Publishing Co. 1988. World Bank. Tanzania Population. Health and Nutrition Sector Review. Report No. 7495- TA. Washington, D.C.: World Bank. October 1989. World Bank. _T2,agnia. Teachers and the Financine of Education. Washington, D.C.: World Bank. May 10, 1991. World Bank. Tanzania - Wmen and Development. Report No. 9108-TA. Washington, D.C.: World Bank. June 1991. World Bank. Tanzania - Economic Report: Towards Sustainable Development in the 1990s. Washington, D.C.: World Bank. June 1991. World Bank. World Development Report 1990. New York: Oxford University Press. June 1990. Worrall, Robert P. "Report and Recommendations: Tanzania Health and Nutrition Project Preparation Mission, December 1-6, 1988: Information, Education, and Communication Component." Report prepared for the World Bank. January 1989