33004 W O R L D B A N K W O R K I N G P A P E R N O . 5 7 Health System Innovations in Central America Lessons and Impact of New Approaches Edited by Gerard M. La Forgia THE WORLD BANK W O R L D B A N K W O R K I N G P A P E R N O . 5 7 Health System Innovations in Central America Lessons and Impact of New Approaches Edited by Gerard M. La Forgia THE WORLD BANK Washington, D.C. Copyright © 2005 The International Bank for Reconstruction and Development / The World Bank 1818 H Street, N.W. Washington, D.C. 20433, U.S.A. All rights reserved Manufactured in the United States of America First Printing: June 2005 printed on recycled paper 1 2 3 4 5 07 06 05 World Bank Working Papers are published to communicate the results of the Bank's work to the development community with the least possible delay. The manuscript of this paper therefore has not been prepared in accordance with the procedures appropriate to formally-edited texts. Some sources cited in this paper may be informal documents that are not readily available. The findings, interpretations, and conclusions expressed herein are those of the author(s) and do not necessarily reflect the views of the International Bank for Reconstruction and Devel- opment/The World Bank and its affiliated organizations, or those of the Executive Directors of The World Bank or the governments they represent. The World Bank does not guarantee the accuracy of the data included in this work. The boundaries, colors, denominations, and other information shown on any map in this work do not imply and judgment on the part of The World Bank of the legal status of any territory or the endorsement or acceptance of such boundaries. The material in this publication is copyrighted. Copying and/or transmitting portions or all of this work without permission may be a violation of applicable law. The International Bank for Reconstruction and Development/The World Bank encourages dissemination of its work and will normally grant permission promptly to reproduce portions of the work. For permission to photocopy or reprint any part of this work, please send a request with complete information to the Copyright Clearance Center, Inc., 222 Rosewood Drive, Danvers, MA 01923, USA, Tel: 978-750-8400, Fax: 978-750-4470, www.copyright.com. All other queries on rights and licenses, including subsidiary rights, should be addressed to the Office of the Publisher,The World Bank,1818 H Street NW,Washington,DC 20433,USA, Fax: 202-522-2422, email: pubrights@worldbank.org. ISBN-10: 0-8213-6278-X ISBN-13: 978-0-8213-6278-5 eISBN: 0-8213-6279-8 ISSN: 1726-5878 DOI: 10.1596/978-0-8213-6278-5 Gerard M. La Forgia is Lead Health Specialist in the Latin America and the Caribbean Region of the World Bank. Library of Congress Cataloging-in-Publication Data has been requested. Contents Foreword v Acknowledgments vii Acronyms and Abbreviations ix Introduction 1 Monique Vledder and Gerard M. La Forgia 1. Is the Perfect the Enemy of the Good? A Case Study on Large-Scale Contracting for Basic Health Services in Rural Guatemala 9 Gerard M. La Forgia, Patricia Mintz, and Carmen Cerezo 2. Contracting for Basic Health Care in Rural Guatemala--Comparison of Performance of the Performance of Three Delivery Models 49 Isabel Danel and Gerard M. La Forgia 3. The San Miguelito Hospital Reform in Panama--Evaluation and Lessons 89 Ricardo Bitrán, Cecilia Má, and Paulina Gómez 4. Contracting Primary Health Care Services--The Case of Costa Rica 109 James Cercone, Rodrigo Briceño, and Varun Gauri 5. The Nicaraguan Social Security Health Insurance Scheme-- A Promising Work in Progress 135 John L. Fiedler and Gerard M. La Forgia 6. A New Dimension for Health Reform--The Integrated Community Child Health Program in Honduras 173 Marcia Griffiths and Judith S. McGuire 7. Improving Efficiency and Impact in Health Care Services-- Lessons from Central America 197 Maureen Lewis Contributors 209 iii Foreword T his book, Health System Innovations in Central America, comes at an opportune time. Many countries in Latin America are taking stock of a decade of reform actions carried out in the health sector. The contents of this book will inspire and provide valuable insights to this evaluation. The book presents a series of case studies of health systems innovations by the Central American republics in the 1990s. The cases have a common theme of efforts to improve specific aspects of health system performance through the introduction of innovative and alternative financial, organizational, or delivery models. Over the past decades, the Central American countries have made significant progress in improving the health status of their populations. Despite these improvements, several countries face major challenges in reaching the Millennium Development Goals for health. Recognizing the need to make their health systems more efficient, equitable, and cost- effective, each country launched innovations to address specific problems or deficiencies in a particular program or function of the health system. The case studies in this book report on the results of these experiences, encompassing a range of issues from the expan- sion of primary care to the use of public-private partnerships and the establishment of a social security-financed delivery system. The lessons from these case studies are highly relevant for other countries in Latin America and other regions of the world. Ensuring high performance of health care deliv- ery systems is a challenge facing all governments. Dealing with the incentive problems underlying public health care delivery to improve productivity, quality, and performance is a common theme of health sector reforms in many countries. However, the impact of these reforms is often hard to establish. The case studies in this book provide an important opportunity to reflect on lessons learned that are relevant to all countries seeking to intro- duce changes in their health systems. I congratulate the authors for their groundbreaking work and hope that this report will stimulate discussion and innovation aimed at strengthening health system performance. Such efforts are critical to achieving the Millennium Development Goals for health. Evangeline Javier Director Human Development Department Latin America and Caribbean Regional Office v Acknowledgments T he papers presented in this volume were commissioned by the World Bank and supported in part by grants from the Dutch Trust Fund, Development Marketplace, and the Gates Foundation. I am grateful for this support from each. This volume was two years in the making, and many individuals helped it along the way toward publica- tion. First and foremost are the contributors who have a shared concern for improving health system performance in Central America. At the early stage of formulation of the research effort, the cogent advice and tireless encouragement from Charles Griffin were invaluable. Subsequently, the volume would not have been possible without the active support of Ariel Fiszbein and Evangeline Javier. I also acknowledge the patience and confidence of Ana-Maria Arriagada, Jane Armitage, Helena Ribe, and Laura Rawlings as the volume took form. Special thanks are due to the peer reviewers for the papers: April Harding, Benjamin Loevinsohn, Alexander Preker, Hedi Deman, Roberto Gutierrez, Milla McLachlan, John Fiedler, and Marcelo Bortman. The papers benefited greatly from the comments and sug- gestions. I would also like to thank Patti Borthwick from Grammarians, Inc., for her rigor- ous editing. Finally, MoniqueVledder did an outstanding job coordinating final review and production of the volume. vii Acronyms and Abbreviations ACECSA Asociación de Servicios Comunitarios de Salud (Guatemala) ADMSS administradora de servicios de salud (health services administrator [Guatemala]) AIN-C Atención Integral a la Niñez en la Comunidad (Integrated Community Child Health Program [Honduras]) ALOS average length of stay ARIs acute respiratory infections B balboa (Panamanian currency) BCG Calmette-Guérin bacillus C$ córdobas(s) (Nicaraguan currency) CBGP community-based growth promotion (CBGP) CCSS Caja Costarricense de Seguro Social CESAMO Centro de Salud con Médico (health center staffed by a doctor [Honduras]) CESAR Centro de Salud Rural (rural health center staffed by a nurse or nurse auxiliary [Honduras]) CLAP Centro Latino-Americano de Perinatologia (Latin American Perinatol- ogy Center) COMSAIN Programa de Comunicación en Salud Infantil (Honduras) CT cuota técnica (technical quota) DASs directores de áreas de salud (district leaders) DPT diphtheria-pertussis-tetanus EAP economically active population EMPs empresas médicas previsionales (previsional medical firms) ENSMI Encuesta Nacional de Salud Materno-Infantil (Guatemala) GDP gross domestic product GOG government of Guatemala HACyA Habilitación, Adjudicación, Certificación y Acreditación (Licensure, Adjudication, Certification, and Accreditation [Guatemala]) HISMA Hospital Integrado San Miguel de Arcángel (Hospital San Miguelito) HPs health posts IMCI integrated management of childhood illnesses INSS Instituto Nicaragüense de Seguridad Social (Nicaraguan Social Security Institute) IVM Invalidez, Vejez y Muerte (Incapacity, Old Age and Death Insurance [Nicaragua]) LCE Ley de Contrataciones del Estado (Public Contracting Law [Guatemala]) MDGs Millennium Development Goals MINSA Ministerio de Salud (Ministry of Health [Nicaragua]) MOF Ministry of Finance MOH Ministry of Health MSP Modelo de Salud Previsional (Previsional Health Model [Nicaragua]) ix X Acronyms and Abbreviations MSPAS Ministerio de Salud Pública y Assistencia Social (Ministry of Health and Social Assistance [Guatemala]) NGOs nongovernmental organizations NIR nosocomial infection rate OB/GYN obstetrical and gynecological OLS ordinary least squares ORS oral rehydration solution PBS paquete básico de servicios (package of basic services) PHC primary health care PMSS Projeto de Mejoramento de los Servicios de Salud (Health Services Strengthening Program [Guatemala]) PNC prenatal care PRAF Programa de Asignación Familiar (Honduras) PSS proveedora de servicios de salud (health provider NGO [Guatemala]) RFPs requests for proposals RI régimen integral (integrated regime) RP Riesgos Profesionales (Professional Risks Insurance [Nicaragua]) SEM Seguro Enfermedad-Maternidad (Sickness-Maternity Insurance [Nicaragua]) SIAS Sistema Integral de Atención en Salud (Integrated Health Care System [Guatemala]) SNUS Sistema Nacional Único de Salud (Universal National Health System [Nicaragua]) SSI Social Security Institute TIPS Trials of Improved Practices UNICEF United Nations Children's Fund USAID U.S. Agency for International Development WHO World Health Organization Introduction Monique Vledder and Gerard La Forgia T his volume presents case studies of recent health system innovations launched by the Central American republics in the 1990s. The studies report on the perfor- mance of the innovations and the policy environment in which they were devel- oped, as well as nuts-and-bolts features and processes incorporated into their design and implementation. Criteria used to select the experiences include: (a) the innovation departed from the modus operandi of the traditional public health delivery system; (b) the innovation was developed and implemented by a government agency that was a major player in the health sector; (c) availability of evaluative or analytic work on performance; (d) accessibility of information on the evolution and process of reform implementation; and (e) evidence that the reform is "here to stay." Each of the cases reports on a reform that has been under implementation for at least eight years (as of 2004) and has survived at least two changes of government. Similar to developing countries elsewhere, during the 1990s, Central American coun- tries faced pressures to improve the performance of their health systems. In most coun- tries, there was a consensus, often negative, that the systems were failing to live up to their potential. Rather than take on system-wide change, each country opted to step into reform through launching innovations to address specific problems or deficiencies in a particular program, function, or intervention of the system. The studies report on how these experi- ences fared--a hospital in Panama, a nutrition program in Honduras, primary care exten- sion in Guatemala, a subset of hospitals and primary care units in Costa Rica, and a social security-managed health care program in Nicaragua. The cases can be characterized as efforts to improve specific aspects of system perfor- mance through alternative financial, organizational, or delivery models. Five involved exper- imentation with health service purchasing through public-private partnerships. One case entailed a public-community partnership. Harnessing the private sector was an alternative 1 2 World Bank Working Paper to command-and-control, public sector provision typical of public and social security health systems in Central America. Although the changes intrinsic in these reforms reflected broader political and economic changes taking place in the region--such as divestiture of state enterprises, opening of markets to competition, and relaxing of gov- ernment controls--the rationale for and context of engagement with the private sector var- ied considerably. As such, an array of strategies emerged over time. Briefly, each country recognized the need to make their systems more equitable, effi- cient, accessible, and cost-effective. However, the route taken depended on the priority placed on each of these goals as well as external pressures. For example, in Guatemala, the goal to rapidly expand access to primary care was mandated by the 1996 Peace Accords, which ended decades of civil strife, and constituted the main driver of nongovernmental organization contracting strategy adopted there. In Costa Rica, cost escalation, user dis- satisfaction, and decreasing production led to partnering with private cooperatives and firms and the application of performance contracting. In Panama, the reform model, con- sisting of the creation of a purchasing agency and contracting out of hospital services, was a compromise measure to avoid wasteful overlap between public and social security- managed delivery systems. In Honduras, the search for cost-effective interventions to com- bat persistent child malnutrition buttressed the development of a pioneering and publicly managed nutrition program. Nicaragua presents the only case of a reform emerging from a heated ideological debate. The establishment of a social security-financed delivery system was linked to the transformation of the centrally planned economy to a market economy in the early 1990s. Nevertheless, the resulting model was a political compromise in which services were pur- chased from both the public and private sectors. The remainder of this introduction provides a thumbnail sketch of the countries and their health systems to facilitate understanding the context of the reforms that shaped the efforts reported in this volume and shed light on their potential for replication in other settings. Overview of Social and Health Conditions in Central America The isthmus of Central America forms a bridge between North and South America. It includes the countries of Guatemala, El Salvador, Honduras, Nicaragua, Costa Rica, and Panama and spans 521,500 square kilometers. The region has a population of approxi- mately 36.4 million. Population growth in much of the region remains high because of continuing high birth rates and falling death rates. Central America contains some of the poorest countries in Latin America as well as a couple of the better off. Table 1 presents summary social and economic indicators. In Nicaragua, Honduras, Guatemala, and El Salvador, half of the population lives in poverty. However, the economic situation is very different in countries such as Costa Rica and Panama, which boast of the high per capita income, low unemployment, and more favor- able social development indicators. With the exception of Costa Rica, most of the coun- tries in the region have large income inequalities. Panama and Guatemala, for example, have among the most highly skewed distributions of income in the world. Traditionally, Central American countries have been reliant on agricultural exports such as coffee and bananas to generate a large portion of their gross domestic product Health System Innovations in Central America 3 Table 1. Summary of Social and Economic Indicators in Central America, 2003 Adult External Population literacy Value debt GNI per below rate added in present capita Total national (% age agriculture value (current population poverty Gini 15 and (% of (% of Country US$) (millions) line (%) index above) GDP) GNI) Costa Rica 4,280 4.0 22.0 46.5 96 8.3 33 El Salvador 2,200 6.5 48.3 53.2 80 9.4 46 Guatemala 1,910 12.3 56.2 48.3 80 22.3 21 Honduras 970 7.0 53.0 55.0 70 13.5 50 Nicaragua 730 5.5 47.9 55.1 77 17.8 77 Panama 4,250 3.0 37.3 56.4 92 5.6 84 LAC 3,260 534.2 N/A 59.0 89 7.0 N/A GNI gross national income. LAC Latin America and the Caribbean Region, World Bank. N/A not available. Source: World Development Indicators database (2004). (GDP). During the past decade, most Central American economies have become more diversified and much of their growth now comes from nontraditional exports and so-called maquila industries. More recently, remittances from large expatriate populations residing in the United States have contributed to income levels, particularly in the cases of Nicaragua and El Salvador. Nevertheless, agriculture continues to play an important role in most of the region. Over the past decades, the Central American republics have made significant progress in improving the health status of their populations. Since 1970, life expectancy at birth has increased by 10 years or more in each country while child mortality rates have fallen dra- matically to, on average, 40 per 1,000 live births in most of the countries. Maternal mor- tality rates have been cut by roughly one-third. Tables 2 and 3 present summaries of measures of health status and system performance This progress has been in part due to improvements in the general standard of living. As average per capita income increased, the education levels and the proportion of the population with regular access to water and sanitation services have steadily improved. The countries have also been successful in improving access to primary care, raising vaccina- tion coverage, and increasing the proportion of births attended by skilled health person- nel. However, with the exception of Costa Rica, Central America has made less progress in reducing malnutrition (table 2). Despite these advances, health status in several Central American countries remains well below Latin American averages. The pace of the progress has been slow relative to other countries with similar, or even lower, income levels. This is especially true of the four poorest countries of the region--El Salvador, Guatemala, Honduras, and Nicaragua-- each of which faces major challenges in reaching the Millennium Development Goals for health. 4 World Bank Working Paper Table 2. Central America: Health Status Indicators (2002) Under-five Infant Maternal Stunting Life mortality mortality mortality of children expectancy per 1,000 live per 1,000 per 100,000 under Country at birth births live births live births five (%) Costa Rica 77.6 11 10 25 6a El Salvador 70.1 37 28 150 23a Guatemala 65.5 56 45 240 49 Honduras 66.1 44 35 110 32.9 Nicaragua 68.7 38 32 230 21 Panama 74.9 25 19 160 18a LAC 71.0 34 28 193 12 a. Health, Nutrition, and Population statistics, World Bank (1995). Source: The World Health Report 2004 malnutrition data from national surveys; Latin America data from World Development Indicators (2004). The improvements in health outcomes have also been very uneven, varying enor- mously by income, urban versus rural residence, and ethnicity. For example, in Nicaragua, between 1998 and 2001 the reduction in infant mortality rates in urban areas was nearly twice that in rural areas and chronic malnutrition fell four times faster in urban than in rural areas. Similar inequalities in health outcomes exist between indigenous and non- indigenous groups. In Guatemala, for example, infant mortality rates decreased four times faster among non-indigenous groups than the indigenous rate in the period from 1987 to 2001. In short, health problems are increasingly concentrated geographically in rural areas, among the poor, and, in the case of Guatemala, among indigenous populations. The cases from Guatemala and Honduras in this volume report on strategies and programs to address the health needs of these populations. Table 3. Selected System Performance Indicators (2002) Births One-year-olds attended Population immunized Total by skilled with access against fertility health to safe Country measles (%) rate personnel (%) water (%) Costa Rica 82 2.3 98.2 95 El Salvador 82 2.9 58.0 77 Guatemala 91 4.5 40.6 92 Honduras 95 3.8 62.0 88 Nicaragua 99 3.8 64.6 77 Panama 95 2.7 90.0 90 LAC 91 2.0 86.0 86 Source: The World Health Report 2004; Latin American data from World Development Indicators (2004). Health System Innovations in Central America 5 Health System Issues Similar to most Latin American countries, the main actors in the health sector in Central America are ministries of health (MOHs), social security institutes (SSIs), and the private sector. The health systems are highly fragmented: an MOH attends to low-income popu- lations; an SSI provides health coverage to mostly salaried, middle-income populations; and the private sector, dominated by fee-for-service solo practitioners, serves all popula- tion groups. The SSIs have commonly enjoyed levels of financing that exceed those of the MOHs. In Panama, for example, the per capita expenditure of the MOH is US$146, ver- sus US$207 of the SSI. Table 4 shows that total and government per capita spending as well as social insur- ance coverage levels in Costa Rica and Panama dwarf levels observed in the rest of Central America. The health care systems in El Salvador, Guatemala, Nicaragua, and Honduras currently provide inadequate financial protection for the population against the financial consequences of ill health. This is evidenced by the high levels of out-of-pocket financing of health care (see Figure 1). Significantly, out-of-pocket expenditures as a percentage of total health expenditures have been increasing over the years. Throughout the region, the institutional environment is not conducive to sustained health system performance. In all countries, the MOHs and SSIs provide most care through facilities that they own and operate. Input-focused, hierarchical bureaucracies and rigid public sector management practices that contribute to diffuse accountability and lack of incentives for performance characterize these systems. Allocation of resources is based on historical budgets; the systems are subject to civil service systems that result in inflexible human resource management; and purchasing of inputs is invariably centralized. Facility, program, and district managers have little decision-making authority on the allocation and use of resources. In most countries in the region, the fragmented nature of the system con- tributes to duplication of service supply that in turn undermines the effectiveness and effi- ciency of the health system. The case studies from Costa Rica, Panama, and Nicaragua report on initiatives to integrate service delivery systems of the government, social insur- ance, and private sectors. Table 4. Health Financing in Central America (2001) Total Per capita Per capita Population expenditure total government covered by on health expenditure expenditure SSI health Country as % of GDP on health (US$) on health system (%) Costa Rica 7.2 562 385 74 El Salvador 8.0 376 175 15 Guatemala 4.8 199 96 14 Nicaragua 7.8 158 77 12 Honduras 6.1 153 81 12 Panama 7.0 458 316 60 Source: The World Health Report 2004; social security data from household survey-based estimates. 6 World Bank Working Paper Figure 1. Share of Government, Social Security, Out-of-Pocket, and Private Insurance Expenditures as Percent of Total Health Expenditure (2001) 70 60 50 40 % 30 20 10 0 El Salvador Guatemala Honduras Nicaragua Panama Costa Rica Government Social security Out-of-pocket Private insurance Source: World Health Report 2004. In sum, although there has been significant progress in improving health outcomes in Central America over the last decades, important challenges remain to achieve the Millen- nium Development Goals. Poor health outcomes are increasingly concentrated geographi- cally among poor and indigenous populations. Epidemiological changes provide additional challenges for the future. The fragmentation of the health systems, lack of financial pro- tection of the poor, and weak incentives inherent in public sector management capacity are all issues that need to be addressed to raise system performance. During the 1990s, Cen- tral American governments began to reassess their role in health service delivery. This vol- ume reports on the performance of experiences that sought to alter financial, organizational, and service delivery arrangements. Volume Preview This book consists of a series of case studies describing the structure and impact of the more promising innovations with the goal of promoting cross-regional fertilization. We have not intended this volume to be a guidebook on the design and implementation of spe- cific reforms, nor do we recommend any particular reform. Indeed, each of these initiatives emerged from historical developments of the health system, contemporary conditions, negotiations among stakeholders, and shifting political and policy contexts. Rather, we seek to present the results of these initiatives in comparative perspective based on available evidence while providing sufficient information on key design elements and implementa- tion processes. Most of the studies deal with innovations related to introducing health purchasing arrangements that link government, social insurance, and private sectors. The arrangements that emerged are varied because each country fashions purchasing processes and procedures while rearranging or establishing organizational and institutional structures to implement purchasing arrangements. To a greater or lesser extent, each case addresses the questions on purchasing and institutional functions posed in Table 5. On the narrower topic of contract- ing, a major element of the purchasing arrangements presented in this volume, La Forgia, Health System Innovations in Central America 7 Mintz, and Cerezo present an opera- Table 5. Major Purchasing Functions tional framework in the first chapter. A brief preview of the cases follows. Policy and design Purchasing The first two chapters examine questions functions the performance and process of a How much Define resource envelope to buy? Collect and manage money large-scale contracting-out effort in Guatemala to extend basic health For whom Define and target beneficiaries to buy? coverage to rural, and mostly indige- What to buy? Define benefit package nous, populations. La Forgia, Mintz, What to exclude? and Cerezo provide an in-depth view Whom to buy from? Select and contract providers of the contracting process in Gua- In what form? temala and how it changed over time. How to pay? Define price and payment The case study provides insights into mechanism the possibilities and limits of contract- Source: Adapted from Preker and others (2001); ing, particularly in terms of large-scale Figueras, Robinson, and Jakubowski (2005). contracting in situations of weak insti- tutional capacity. The second chapter by Danel and La Forgia analyzes the comparative performance of three service delivery models used to extend basic services in Guatemala: direct contract, mixed contracting, and traditional public models. Based on household and provider surveys, the evaluation suggests mixed results that depend on contextual factors affecting the supply of and demand for services. In the third chapter, Bitrán, Má, and Gómez present the results of an evaluation of alternative financial and organizational arrangements applied to the Hospital Integrado San Miguel de Arcángel (Hospital San Miguelito, also known as HISMA) in Panama City. This reform is noteworthy because it pooled heretofore separate financial flows from the MOH and SSI, established a jointly funded agency that purchases services on the behalf of the MOH and SSI, created an independent governance structure for the hospital, and con- tracted out services to an array of private providers through competitive bidding. Com- pared to two control hospitals, HISMA demonstrated superior performance. The fourth chapter compares the performance of private providers contracted by the Costa Rican Social Security Institute (CCSS) with providers directly managed by the CCSS. Using a panel data set of ambulatory care outputs, Cercone, Briceño, and Gauri found that contracted providers, including cooperatives and private firms, were more efficient than CCSS direct providers with no apparent reduction in essential services or outcomes. The fifth chapter analyzes the evolution and performance of a health financing and delivery scheme introduced by the Nicaraguan Social Security Institute. Fiedler and La For- gia found that the model receives high marks for patient satisfaction, capacity for resolu- tion, cost containment, and increasing available resources in the broader health system. However, several shortcomings were identified, including low population coverage, hid- den subsidies, and weak contract management. The sixth chapter by Griffiths and McGuire assesses the development, organization, and impact of a community-based public health and nutrition program, Atención Integral a la Niñez en la Comunidad (AIN-C) in Honduras. Unlike the other programs described in this volume, AIN-C constitutes a public-community partnership. According to the authors, active community participation is the key ingredient in the program's success. They present results from two evaluations suggesting that AIN-C is cost effective. 8 World Bank Working Paper In the final chapter, Lewis provides a fitting conclusion to this volume. She examines the policy implications and lessons learned from the cases, stressing the importance of how the innovations altered incentives to improve performance and productivity. Many of experiences were successfully launched and later sustained under uncertain policy and institutional conditions, suggesting that performance can be attained in difficult policy environments. Finally, the cases are also deemed innovative because of the willingness of public authorities to support systematic data collection and evaluation. Without full understanding of results as well as reasons for success and failure, health systems can only grope in the dark in terms of raising performance. We hope that the case studies consti- tuting this volume will provide some insight into innovative ways to improve health sys- tem performance. References Figueras, J., R. Robinson, and E. Jakubowski, eds. 2005. Purchasing to Improve Health Sys- tem Performance. Berkshire: Open University Press. Preker, Alexander S., Cristian Baeza, Melitta Jakab, and Jack Langenbrunner. 2001. "Resource Allocation and Purchasing Arrangements that Benefit the Poor and Excluded Groups." World Bank, Washington, D.C. Processed. CHAPTER 1 Is the Perfect the Enemy of the Good? A Case Study on Large-Scale Contracting for Basic Health Services in Rural Guatemala1 Gerard La Forgia, Patricia Mintz, and Carmen Cerezo Abstract This case study examines a large-scale contracting effort in Guatemala over a six-year period. In 1997, the government embarked on an ambitious effort to extend basic health coverage to impoverished rural and indigenous popu- lations through the contracting of private, nongovernmental organizations (NGOs). By 2002, the government had signed 160 contracts with 88 NGOs to reach approximately 3 million beneficiaries. Two distinct phases are described for the period under study: a developmental, improvisational phase during the early years of implementation, and, later, a correction and consolidation phase. Drawing on the Guatemalan experience, this case study aims to provide an in-depth analysis of the "nuts and bolts" of contracting practices, how they have changed over time, and how they have affected the delivery of services by contracted NGO providers. It also draws lessons for policy makers who are considering contracting out the provision of health services. The case study complements a companion chapter in this volume on the impact of contracting on meeting performance targets. 1. This case study is based on fieldwork conducted in 2000 and 2001 in Guatemala. A companion chapter in this volume evaluates the performance of the contracting experience. This case study draws on field research commissioned to Gestión y Tecnologia en Salud y Desarrollo (GETSA), S.A. It also draws on the work of Nieves and La Forgia (2000) and Mintz, La Forgia, and Savedoff (2001). 9 10 World Bank Working Paper Introduction Contracting out health services by public agencies is not new. Municipalities in the United States began contracting out an array of health services and programs in the 1950s (Shulman 1982). Beginning in the 1980s, contracting services became increasingly common in other developed countries, especially the United Kingdom, Finland, New Zealand, Sweden, Australia, and Canada (Vining and Globerman 1999). In general, the experience in devel- oping countries is more recent, with the launching of most initiatives in the 1990s.2 Build- ing upon lessons learned from contract experiences in the developed world, these countries are also pioneering innovations of their own. Although still dominated by maintenance, catering, security, and laundry services contracts in hospital settings, developing countries are increasingly seeking to provide a range of health services to poor and underserved pop- ulations through direct contracting with private organizations. Recent experiences include contracting of: (a) nongovernmental organizations (NGOs) in Afghanistan, Cambodia, Bangladesh, Colombia, India, Mali, and Panama to provide primary clinical, preventive, and reproductive health services; (b) managed care organizations to provide coverage for a comprehensive service plan in Colombia, Estonia, and Nicaragua; and (c) medical coop- eratives and other provider organizations to supply clinical services in Brazil, Costa Rica, and Panama.3 Evaluations of government contracting experiences in developing countries are scarce. Those that do exist focus on the impact of contracting on achieving stated goals related to coverage, service delivery, and health.4 Although contracting has not been systematically examined or a major focus of available evaluations, some studies have attempted to under- stand how government's contracting infrastructure (or lack thereof) and processes may affect overall performance.5 Similar to the experience in developed countries, the contract process itself affects service delivery and therefore performance. In a review of the literature drawn from developing countries, Taylor (2003) identifies a number of factors affecting the perfor- mance of contracting out: the supply of public providers and their capacity to manage and supervise service delivery; contract management and negotiating capacity of the government; payment strategy, including the adequacy and timeliness of payments; confidence in the legal and regulatory framework; transaction costs related to preparing for and implementing con- tracting; and government capability in technical monitoring and evaluation.6 2. Governments in a limited number of developing countries, such as South Africa and Brazil, have had contractual agreements with private hospitals for decades (see Mills [1998] and World Bank [1994]). 3. For a review of contracting experiences in developing countries, see Mills (1998), Mills and Broomberg (1998), and Bennett (1997). For a database on contracting experiences, see http://www1. worldbank.org/hnp/hsd/documents/ContractDatabase.pdf; for Nicaragua, see Fiedler and La Forgia (in this volume); for Panama, see Bitrán, Má, and Gómez (in this volume); for Costa Rica, see Cercone and Rosenmöller (2000) and Abramson (1999); on contracting of reproductive health services in Colombia and India, see Rosen (2000) and Abramson (1999). 4. For Cambodia, see Loevinsohn (2002); for Haiti, see Eichler, Auxila, and Pollock (2001); for South Africa, see Mills and others (2004). 5. One exception is Cambodia, which identified several issues related to contracting out, including weak skills of staff hired by contracted nongovernmental organizations (NGOs), low contract manage- ment capacity of government, and lack of communication between government district authorities and contracted NGOs (Loevinsohn 2002; Souters and Griffitths 2003). Mills and others (2004) also examined how contract design can influence primary care performance in South Africa. 6. See Domberger (1998) for a review of issues affecting contracting out in developed countries. Health System Innovations in Central America 11 In 1997, the government of Guatemala (GOG) embarked on an ambitious effort to extend basic health coverage to impoverished rural and indigenous populations through contracting with private NGOs. By 2002, the GOG had crafted 160 contracts with 88 NGOs to reach approximately 3 million beneficiaries. Before this effort, the GOG had provided health services only through inhouse providers and had not entered into service contracts (of any kind) with private providers. The Ministry of Health and Social Protection (Ministerio de Salud Pública y Assistencia Social [MSPAS]) contracted with NGOs to deliver a basic package of services, giving priority to prevention as well as maternal and child care in rural areas where publicly managed services were irregular or unavailable. The magnitude of this effort together with the accelerated scaling-up process were un- precedented among developing countries. The MSPAS developed and expanded the pro- gram by trial and error during a three-year period. In addition to forging the "downstream" infrastructure related to NGO selection, contractual design, payment mechanisms, and infor- mation and monitoring systems, the Guatemalans had to develop the "upstream" institu- tional infrastructure to support service purchasing, including the policy framework, regulatory environment, consensus-building strategies, financial structures and processes, and auditing systems. This case study focuses on the direct contracting model applied by the government.7 The model entailed the contracting of NGOs to manage and provide a set of services to specified catchment areas. These NGOs received a per capita payment and were responsi- ble for the procurement of all inputs except vaccinations. Drawing on the Guatemalan experience, this case study aims to provide an in-depth analysis of the "nuts and bolts" of contracting practices, how they have changed over time, and how they have affected the delivery of services by contracted NGO providers.8 It also draws lessons for policy makers who are considering contracting out the provision of health services. The story of Guatemala's contracting is one of tradeoffs between moving quickly on the one hand and "checking off all the boxes" of an ideal contracting strategy on the other. It is also an example of the capacity to learn and respond to new information and to develop new capabilities. The case provides insight on the do's and don'ts of contracting practices, scaling-up processes, and creating the institutional capacity to purchase services. The main message of this case study is that successful contracting is not a question of getting it all right. Indeed, if Guatemala had diligently checked off all the boxes and applied a contracting strategy based on best practice, implementation of the large-scale effort might have been elusive. Given the lack of experience with service contracting, the time required to develop the capacity to implement a more vigorous contracting strategy might have seri- ously compromised achievement of population coverage extension goals set by the GOG. Many of the elements associated with effective contracting were absent during the early years of the program--for example, a transparent and competitive selection process, clear specification of service responsibilities of contracted providers, use and measurement of performance, and development of contracting management and monitoring functions. Failure to address these items probably contributed to less than anticipated performance, 7. Government also applied a mixed contracting model. About two-thirds of all contracts entailed the direct contracting model. 8. Chapter 2 in this volume examines the impact of different contracting models on service delivery performance. 12 World Bank Working Paper as evidenced by the failure of most providers to deliver the full range of services desired by the GOG and lower than expected population coverage with priority interventions. How- ever, the GOG did enable the contracting process by garnering political and institutional support, establishing a supportive legal and regulatory framework, and developing neces- sary finance management and payment systems. More important, the Guatemalans learned from their mistakes. Three years after the first round of contracts was signed, the GOG began a series of in-flight corrections that addressed most of the early shortcomings. Main- taining political and policy support for contracting while adjusting instruments and reshap- ing processes, based on information derived from implementation, led to significant improvements in contracting and substantially enabled continuity. The next section begins with a description of the policy context that led to the decision to contract. It is followed by a section that summarizes the main design features of the con- tracting model as applied in Guatemala. A brief review of a nine-step operational frame- work on contracting, which outlines major aspects of the contracting process, follows, and the remainder of the study applies the nine-step framework to the Guatemalan experience. Each step is analyzed to compare the "ideal" to what actually occurred in Guatemala. The case study concludes with a contracting scorecard and summary of lessons learned. The results are based on document analysis and in-depth interviews with nearly 30 key actors and stakeholders. To obtain a balanced assessment of the reality of designing and implementing contracting out, the following participants were interviewed: (a) current and former program officials; (b) MSPAS officials at the central, area, and district levels; (c) a sample of NGOs providers under contract with the MSPAS; (d) NGO providers whose contracts were cancelled; (e) NGO providers that participated in the selection process but were not awarded contracts; and (f) representatives of donor agencies supporting the MSPAS contracting. Why Contract? The Political and Policy Context in Guatemala in 1996 In 1996, Guatemala was emerging from a lengthy and bitter civil war. In that year, a newly elected government instituted talks and Peace Accords were signed, officially ending the civil strife. These accords specifically contained provisions regarding health improvement in the country, especially in those areas where care was most needed and least available-- the rural areas. Specifically, the accords stated: The Government undertakes to increase the resources it allocates to health. By the year 2000, the Government proposes to step up public spending on health as a proportion of gross domestic product by at least 50 percent over its 1995 level. This target will be revised upwards in the light of future developments in State finances. The system would give priority to efforts to fight malnutrition and to promote environ- mental sanitation, preventive health care and primary health care, especially maternal and child care. The Government undertakes to allocate at least 50 percent of public health expenditure to preventive care and undertakes to cut the 1995 infant and maternal mortality rate in half by the year 2000. In addition, the Government undertakes to maintain the certification of eradication of poliomyelitis, and to eradicate measles by the year 2000. --article VII, paragraphs c and d, Peace Accords, December 29, 1996 Health System Innovations in Central America 13 The accords and the need to increase resources directed at rural, largely indigenous, communities demanded a rapid, well-targeted response. The political cycle, with elections scheduled for 1999, also contributed to the government decision to accelerate basic health coverage extension in rural areas, where more than half of the population resided. Faced with these time constraints, the Minister of Health became convinced that increasing the level of services provided internally by the ministry for the targeted popula- tion would take too long. To mount such an effort would require screening, hiring, train- ing, and placing a large number of staff in areas where historically the MSPAS had been mostly unsuccessful in recruiting and maintaining personnel. Purchasing and providing supplies to support coverage extension posed additional obstacles. Where MSPAS providers did exist in remote areas, supply-side constraints related to provision of logistical, equip- ment, and material support were notoriously deficient (Annis 1981; Pebley and Goldman 1992; Goldman and Pebley 1994; Cosminsky 1987). Contracting was also envisioned to enable the MSPAS to expand coverage rapidly by facilitating the acquisition of inputs through bypassing the traditionally slow and cumbersome government personnel hiring and material procurement systems. Alternatively, the new MSPAS authorities were aware that an undetermined number of NGOs, supported by external agencies, were providing health services in rural areas. How- ever, little systematic information existed about the precise number of NGOs, their loca- tion, or the type and volume of services provided. Another enabling factor was a recently approved loan from the Inter-American Development Bank, which included financing for a demonstration project on service contracting. Epidemiological imperatives also drove the placement of health mandates in the accords and the ministry's decision to contract out. Table 1.1 compares health indicators for the poor- est and richest income quintiles (in 1995) and for indigenous and nonindigenous populations (in 1998­99). In the mid-1990s, Guatemala had a population of about 10.5 million, of which Table 1.1. Summary Health Status and Service Indicators, 1995 and 1998­99 Poorest Richest Non- National quintile quintile Indigenous indigenous average Indicator (1995) (1995) (1998­99) (1998­99) (1995) Infant mortality 56.9 35.0 56.0 44.0 51.0 Children stunted (%) 64.6 12.1 N/A N/A 49.7 Total fertility rate 8.0 2.4 6.2 4.6 5.1 Diarrhea prevalence 22.8 16.0 13.7 13.1 20.9 DPT3 coverage 54.3 70.8 49.5 65.3 59.4 Births assisted by 9.3 91.5 16.7a 54.8a 34.8 trained personnel (%) Prenatal visit by 34.6 90.0 45.6a 68.0a 52.5 trained personnel (%) DPT diphtheria-pertussis-tetanus. N/A not available. a. Referred to physician or nurse attendance. Source: For 1995, Pan American Health Organization (1999) and Gwatkin and others (2000); for 1998­99, Instituto Nacional de Estadística (1999). 14 World Bank Working Paper about 40 percent were indigenous. Indigenous populations had traditionally resided mainly in the northern and northwestern highlands, but they had also migrated to the southern coasts to sell their labor to large coffee and sugar growers. Documented health conditions showed high national rates of infant mortality (57 per 1,000 live births), high maternal mortality (190 per 100,000 live births), and a national life expectancy rate of 65 years. However, all these rates were at least 20 to 60 percent worse in the rural highland areas, where the program was to be targeted. Diarrhea and acute respiratory infections (ARIs) were the main killers of children, particularly in rural areas. Wide gaps in health service use were also evident. The MSPAS esti- mated that 40 percent of the population had little or no access to basic care, and most were concentrated in rural areas, particularly among indigenous groups. All of these conditions, coupled with the indigenous communities' lingering antipa- thy toward and distrust of the government, contributed to the MSPAS decision to contract out basic care provision. The decision was made in late 1996 to support a program to extend coverage for a defined package of basic health services to targeted rural populations through contracting with existing NGOs. Program Overview The coverage extension effort was part of a broader health reform policy that sought to mod- ernize and decentralize the MSPAS, improve the efficiency and quality of health services, strengthen management and governance functions of the MSPAS, and increase financial resources to the sector. This set of initiatives was known as the Integrated Health Care Sys- tem (Sistema Integral de Atención en Salud [SIAS]).9 The reform process also sought to reorient the role of the MSPAS, introduce alternative management and organizational arrangements, and make the service delivery system more responsive to clients and payers. The design involved contracting with NGOs to deliver a basic package of services, giv- ing priority to prevention, maternal and child care, and basic curative care. As stipulated in a contract-like instrument (a convenio), private organizations performed two distinct functions: both provision and administration of services. A health provider NGO, known as a PSS (proveedora de servicios de salud), directly provided the basic package in predeter- mined geographic areas. This was known as the direct contracting model. A health services administrator NGO, or ADMSS (administradora de servicios de salud), operated exclusively as administrators, financial managers, and payers for MSPAS-operated provider teams consisting of both MSPAS- and NGO-hired personnel.10 This became known as the mixed contracting model. About two-thirds of the convenios have been awarded to organizations that operate as PSSs. A number of NGOs act as both PSSs and ADMSSs. SIAS also encom- passed the extension of coverage through the MSPAS delivery system. 9. This case study does not deal with aspects of the program unrelated to contracting. 10. ADMSSs served as financial and administrative agents for MSPAS-managed teams. The original design called for the application of the direct contracting model only. However, protests from MSPAS area personnel resulted in the negotiation of a mixed contracting model. The ADMSSs paid personnel (mainly physicians) and administrative staff, purchased supplies, and managed the books for the MSPAS, but did not manage, supervise, or provide technical oversight for the provider teams. The latter was the responsibility of MSPAS area offices. Health System Innovations in Central America 15 The PSSs and ADMSSs received capitated payments in four yearly installments. The amounts received varied, depending on whether the contract was for an ADMSS or PSS, the number of inhabitants in the assigned catchment area, and considerations for distance. The ADMSSs received a lower capitation rate than the PSSs for two reasons. First, an ADMSS contracted a reduced number of personnel compared to the PSSs (see footnote 10). Second, because most PSSs were assigned to remote areas, they received higher rates to account for elevated transportation costs. The PSSs were assigned a catchment area containing at least 10,000 inhabitants, usu- ally distributed among dispersed and distant rural hamlets. The payment covered the direct cost of the basic package plus administrative expenses and expenditures related to institutional strengthening. The MSPAS supplied vaccines, but the NGOs were respon- sible for purchasing all other inputs. In 1998, the average per capita payment was approx- imately US$6.25. As of 2002, it had not been adjusted for inflation. The total per capita cost was around US$8.00, including the cost of vaccines that the ministry supplied directly to the PSSs.11 Supported by a loan from the Inter-American Development Bank, the plan originally proposed by the MSPAS involved an initial pilot period with three departments: Escuintla, Alta Verapaz, and Chiquimula. The plan also called for scaling up in a second phase, start- ing in 2000 and drawing on the results of an evaluation of the pilot phase. By the end of 1997, seven NGOs had 14 agreements in these three departments. Pressed by the need to show rapid improvement in health outcomes and system cover- age, which in turn responded to provisions in the accords and to the ruling party's political mandate to show results before the next round of elections (in 1999), the ministry made the decision, only six months into the pilot program, to expand contracting out nationwide. This effectively ended the pilot experiment and forced Table 1.2. Summary Statistics on SIAS Implementation large-scale contracting. Table Agreements, NGOs, and Estimated Coverage, 1.2 shows the exponential 1997­2002 growth of contracts and peo- Estimated ple served from 1997 through No. of signed No. of NGOs population 2002. The greatest increase Year agreements under contract covered took place between 1997 and 1997 14 7 460,000 1998. Since then, growth has 1998 110 74 2,180.000 been much slower because the 1999 119 78 2,516,000 MSPAS has sought to con- 2000 136 89 2,940,000 solidate gains and focus on 2001 154 89 3,000,000 installing the support and 2002 161 88 3,200,000 monitoring systems to foster stronger performance. Source: MSPAS and SIAS annual reports, various years. 11. The MSPAS also purchased and supplied drugs for the treatment of tuberculosis and malaria for NGOs demonstrating need. However, some NGOs purchased these drugs directly from suppliers when the MSPAS was unable to supply them in a timely manner. 16 World Bank Working Paper Contracting Framework This section briefly outlines a stylized framework of best practices for the "how" of con- tracting. It draws on experiences in public sector contracting of health services in the United States and Europe and, to a lesser extent, on experiences in developing countries.12 These experiences range from subcontracting laundry services in a hospital to purchasing compre- hensive health insurance for low-income groups. In countries as diverse as Sweden (which has experimented with contracting private administration of hospitals) to the United States (where states are innovating in expanding insurance coverage for the poor through new Medicaid programs), contracting is being adopted as an important instrument for the pub- lic sector to achieve social goals. The framework identifies the distinct steps necessary to select, execute, and moni-tor a successful contract. As described in Box 1.1, it consists of nine steps. The steps are similar to "critical" or "successful" contracting features pre- sented by Mills (1998), Mills and Broomberg (1998.), and Taylor (2003). Although the framework is by no means exhaustive, its nine steps provide an analytic basis for assessing the design and implementation of contracting practices for health services. Box 1.1. Contracting Framework 1. Assess the feasibility of contracting, including the costs, political consequences, availability of suppliers, regulatory framework, and readiness to contract. 2. Gain political and institutional support for contracting, including building public support from communities, organizations, and unions. 3. Define service specifications, including services to be purchased and the target populations to be served. 4. Select performance measures. 5. Define payment methods and link payment to performance. 6. Select providers and maximize competition in the bidding process. 7. Negotiate and write the actual contract. 8. Monitor and evaluate the contract, and ensure the capacity for contract management. 9. Encourage competition over the long term to avoid monopolistic abuses in contracting. Source: Mintz, LaForgia, and Savedoff (2001). Analysis of Guatemala's Contracting Process The remainder of this chapter is organized according to the contracting framework pre- sented in Box 1.1. For each step, the salient tasks are identified and described in terms of best practice. The discussion centers on practical advice as to the do's and don'ts pertinent to success at each step. These are summarized in a box at the beginning of each section. Then each step of the Guatemalan direct contracting experience is described--the extent to which specific elements of the stylized framework were applied, modified, or ignored, as well as the consequences of actual practices for service delivery. In several steps where 12. A more indepth description of the framework can be found in Mintz, La Forgia, and Savedoff (2001). Health System Innovations in Central America 17 there was a significant change between the early (1997­99) and later (2000­02) phases of the project, these phases are described separately. The analysis also includes what was miss- ing in each step. Step 1: Assess feasibility Conceptual Framework Deciding which services to contract out must take into account the fac- Box 1.2. Assessing Feasibility tors that can support or inhibit con- tracting (see Box 1.2). The first factor Evaluate relative costs of administration. to be considered is cost. What are the Assess the political readiness for contracting. total costs of providing the service Evaluate the market. through a contract compared to Assess the existing regulatory framework. direct, inhouse provision? Another is Evaluate the existing contracting infrastructure. the market of suppliers of the ser- vice(s) to be contracted. Among the factors that should be considered is the extent of competition in the supply of services. In comparing the total costs of contracting with the costs of retaining services in the public sector, it is also important to compare them to the true cost of providing an equivalent quantity and quality of service through direct, inhouse provision. A second major issue that affects feasibility of contracting is the political context. Specifically, how do political leaders view contracting at the national and local levels, and at the Ministry of Health? At the national level, there needs to be support for market forces and the involvement of the private sector, including nonprofit organizations, in the con- tracted provision of goods and services previously provided exclusively by the public sec- tor (or not provided at all). These policy considerations also include evaluating the readiness of the population at large for contracting. In the public agency itself, there needs to be a commitment to move toward a role that emphasizes policy, planning, and oversight as opposed to direct provision of all services. Contracting need not replace existing services that are directly administered. Instead, con- tracting can be focused on previously unserved or underserved populations or on services that have not been previously provided. This is commonly the case with basic health and reproductive health services (Rosen 2000). At the local level, there needs to be recognition of the impact of contracting on spe- cific facilities and local governmental institutions. Whether the local population views this as a threat to their services or a welcome opportunity to get more and better services will critically affect the degree of acceptance. Local government institutions may also be required to take on new functions that they are not prepared for. For example, they may be asked to manage new funds or monitor contracts. This can sometimes facilitate and some- times obstruct implementation. Furthermore, local relationships among private providers, NGOs, public facilities, and the local government will also play a role. A third issue is the availability and quality of suppliers. Who are the available con- tractors, how numerous are they, and do they compete with one another? In general, the 18 World Bank Working Paper greater the number of suppliers available, the greater the leverage of the contracting orga- nization for cost, quality, and outputs. In cases where the market of service providers is weak, governments can initiate small-scale experiments in contracting, incrementally expanding the number of suppliers. Potential providers may need technical assistance and training to raise standards and create capacity to become acceptable contractors. A fourth component in evaluating contracting feasibility is the presence and sophisti- cation of a supportive regulatory framework. Nearly all governments have laws defining the procedures of public contracting, but these vary considerably. Unfortunately, most were drafted with procurement of goods and infrastructure in mind, and they may not be applicable to health service contracting. A final element that will determine the feasibility of contracting is the organization's ability to contract and oversee contracts once they have been awarded. Contract manage- ment is critical to the success of contracting and often requires the establishment of a purchasing unit within the contracting agency. Nonexistent or weak capacity would rec- ommend an incremental approach to contracting through small-scale arrangements. Mov- ing slowly would allow time for the development of institutional capacity. The Guatemalan Experience The decision to proceed with contracting was in part a political decision made by MSPAS authorities and in part a contractual obligation specified in a loan financed by the Inter- American Development Bank.13 The matrix of conditionalities stipulated that by the third year of project implementation, projected for 1999, the GOG was to sign at least eight con- tracts with NGO providers in three pilot departments: Alta Verapaz, Chiquimula, and Escuintla. During the preparation of the loan-financed project, known as the PMSS (Health Services Strengthening Program [Programa de Mejoramento de los Servicios de Salud]), the MSPAS explored options for contracting with NGOs in these areas. Consul- tants prepared several reports that provided a basis for decision making by the MSPAS, including legal feasibility analysis, definitions of health care delivery model and service package, and estimated service delivery costs. The last was based on a small sample of NGOs operating in the three target areas. No market analyses or other feasibility analyses were conducted in other regions of the country. As mentioned, the idea was to implement and evaluate a pilot program and then, based on progress and results, scale up to other areas of the country in a subsequent phase. Organizational capacity to contract was nonexistent, but this was not surprising because the GOG was inexperienced in service contracting. Because the PMSS called for a small pilot program of contracting out, no feasibility analyses were conducted to assess expertise, institutional capacity, or data-processing needs for contract management inside the MSPAS. Expertise was to be built up during pilot implementation. 13. The Health Services Strengthening Program (PMSS) consisted of two loans: an Adjustment Loan for US$25 million and a Technical Cooperation Loan for US$13.5 million. The Adjustment Loan con- tained conditions oriented toward increasing public health financing and allocative efficiency, extending basic coverage and improving the management of public hospitals. The Technical Cooperation Loan pro- vided financing for technical assistance activities to support achievement of the adjustment conditions. Both loans were approved by an outgoing government in December 1995. Health System Innovations in Central America 19 The MSPAS performed no formal analysis of the comparative costs of internal ver- sus external contracting. The decision to contract was largely based on the fact that inter- nal resources at MSPAS were inadequate and, when available, difficult to allocate and convert into service delivery, especially in poorly served rural areas where the MSPAS presence was minimal at best. Also, the NGOs were seen as more likely than the govern- ment to be accepted by indigenous people. Their cost structures, however, were not known before initiation of the contracting process. It is not clear whether increased information about internal versus external ("make versus buy") costs would have changed the deci- sion to contract in Guatemala. It might, however, have provided a benchmark against which to judge the cost of the package of basic services that was later developed and implemented. Contracting for basic health services was heavily supported at the highest levels of the Ministry of Health (and the GOG) and given further impetus by the requirements for rapid improvement of health outcomes created by the Peace Accords and the political cycle in which elections were closely followed by the target date (2000) of the health provisions of the accords. However, the contracting program also faced a number of challenges and vulnerabilities. First, although political support for the new program was strong, it was greeted at the district and local levels with mistrust and suspicion. Many NGOs distrusted the GOG's intentions and capacity to implement, especially because there was no precedent of gov- ernment contracting of services. Many NGOs touted their separation from government and their ability to provide higher-quality services than the MSPAS. Some were hesitant about public perception of directly or indirectly supporting any underlying political cam- paign of the government to use SIAS as a springboard to solidify bases of political support. Others were more concerned with well-known government practices of not paying or delaying payments to suppliers. Still others feared that they would be drawn into corrupt procurement practices that were well known to be plaguing the government. Second, although many NGOs existed,14 there was little systematic information, by way of market analysis, as to the specific capabilities and capacity of each one. The lack of a broader market analysis (beyond the sample of NGOs operating in the pilot departments) and the use of prevailing assumptions about NGOs' capacities and capabilities created an inaccurate assessment of the ability of these contractors to deliver the prescribed services. This led to delivery system problems downstream. Taking the time to do a quick scan of the true capabilities of the NGOs in the country would have created a far more realistic pic- ture of capacity and identified areas for development. A structure and a process for the cer- tification or accreditation of NGOs initially did not exist. Financially, the MSPAS had insufficient resources to support the program, but funds were buttressed by the PMSS and additional funds were earmarked for the program by the Ministry of Finance (MOF).15 However, getting the funding from the MOF to the MSPAS in order to pay NGO providers was a potential impediment to contracting. No payment mechanism existed, nor was there even a line item in the government budget to allocate 14. An earlier MSPAS survey indicated that nearly 200 NGOs were working in the health field in Guatemala. 15. The PMSS did not directly finance the NGO contracts. 20 World Bank Working Paper public funds to contracted providers. There was no mention of contracting public social services in the legal and regulatory framework governing public spending.16 Finally, the infrastructure and regulatory framework for public contracting in the coun- try was extremely weak, according to a legal assessment performed by the MSPAS. There was no mechanism for formally contracting (or paying) nonprofit health service providers. Existing contract law focused on commercial businesses and transactions primarily of works and goods, a problem noted in the framework. This represented another barrier to imple- mentation and took much-needed time away from contract planning and implementation. In general, the lack of a clear regulatory framework to support contracting with NGOs both required innovations and created obstacles to implementation. The MSPAS team reached a consensus that moving forward with contracting out would require legal reforms. Step 2: Gain political and institutional support for contracting Conceptual Framework Public concerns and political considerations are just as important as economic and legal ones. Politics can provide an important source of support for contracting as well as a source of unyielding resistance. At the broadest (and frequently least organized) level are consumers and communities. At a second level are groups that have a specific stake in the process. This in- cludes nonprofit organizations (including NGOs) and private Box 1.3. Gain Political and Institutional Support firms (including private hospi- tals) that may be awarded con- Analyze what will be necessary to gain public sup- port of contracting. tracts. It also includes public Evaluate the extent of public sector dominance of sector unions, professional associ- the service. ations, and public employees. The Consult with NGOs and other relevant organizations. section summarizes the key factors Understand the public sector unions and work col- and actors that influence the deci- laboratively with union leadership. sion to contract (see Box 1.3). Level of public support for the contracting strategy. Community support or opposition to a contracting plan can make or break the effort. Before the first contract is signed, there needs to be a focused public relations campaign that educates the community and then involves them in the decision-making process (for example, development of evaluation cri- teria, selection of finalists). This process involves discussions about the problems or limi- tations of the status quo, and the benefits that can be realized through contracting. Role of nonprofit organizations and for-profit firms in the contracting process. Private sec- tor trade groups, along with associations of private firms and nonprofit agencies, are obvious but often untapped sources of support for the contracting of services. Their interests can balance those of public employees, who generally seek to retain inhouse 16. In preparation for the PMSS, financial flow analyses were performed to determine options to enable the MSPAS to pay private contractors. Health System Innovations in Central America 21 provision. To enlist support from these associations, government agencies should consider how contracting will affect those businesses, their competition, and job creation. Potential providers are generally most interested in the transparency of procurement and bidding processes and the definition or negotiation of contractual terms. However, they may obstruct implementation if their perspectives are excluded from the design of both the con- tract and the contracting process. Presence or absence of a unionized workforce. In general, the possibilities and support for contracting will be weaker where service delivery is concentrated in public agencies. The role played by unions is critical to the contracting process. Dealing confrontationally with unions or ignoring them often leads to strikes and rallying of public opinion against the contracting initiative. By contrast, planning for union participation will improve the chances of success. A policy of no layoffs also can defuse labor opposition. This can be com- bined with assisting workers displaced by contracting with finding jobs in other localities and facilities or elsewhere in government. Prevalence and quality of good information. It is considered best practice to concretely demonstrate the potential benefits from contracting to the community, NGOs, and unions, and keep them informed of evaluations of the contracting process. This is easier when good information on current costs and levels of service can be made available. The Guatemalan Experience Before the institution of SIAS, health care for most Guatemalans with access to services was provided by public providers. There was no tradition of the types of public-private part- nership that contracting represents. Consultations with public employees, NGOs, and other stakeholders were considered after the decision to contract and scale up. This made implementation more difficult. District and local branches of the MSPAS saw the contracting program as a threat-- to their jobs and their control--a potential first step in "privatizing" the national health care delivery system. A major problem was that members of the district leadership (direc- tores de áreas de salud [DASs]) were not involved in either the development of the program, including the identification of services to be covered, or the identification of the NGOs with whom initial contracts were developed. Strong opposition from MSPAS area and district personnel cast doubts on the ability of the MSPAS to launch the contracting-out proposal. Local personnel questioned why the MSPAS was abandoning cadres of well-trained but underused rural health personnel (rural health technicians). Supported by area authorities, this well-organized group of health work- ers claimed that if they were provided with sufficient inputs and support, they too could extend services effectively and efficiently. Recognizing that coverage extension could be severely compromised without area and district support, the MSPAS and area personnel negotiated a mixed contracting model whereby "administrative NGOs" were contracted to hire additional personnel (to complement MSPAS health workers) and purchase supplies for MSPAS-managed provider teams. This compromise contributed to area and district per- sonnel's acceptance of NGO direct contracting. 22 World Bank Working Paper Increased early involvement and communication could have substantially improved program effectiveness by eliminating an "us versus them" mindset among MSPAS offi- cials and NGOs at the district level. Under the program design, DASs were to have over- sight of the NGOs. However, because of the resentment between the DASs and the NGOs (as well as the resentment toward the MSPAS for "imposing" the new system), there were initial efforts to disrupt service provision by these fledgling organizations. For example, a there were DASs who often held back allocations of supplies and, in some cases, financ- ing to the NGOs. Partly because of time pressure and partly as a management decision, there was little initial consultation with NGOs.17 In addition, there was relatively little communication overall with the NGO community. In the early phases of the program, there was limited understanding on the part of NGOs about what the expectations of SIAS, in part because the program itself was a work in progress. These misunderstandings later led to service delivery problems in the rural commu- nities served. More time spent educating NGOs at the beginning about the components and expectations of the contract could have ameliorated these difficulties. One umbrella NGO organization, Asociación de Servicios Comunitarios de Salud (ACECSA), which pro- vided training services to NGOs, strongly resisted the new program and militated against it publicly. ACECSA came to represent a group of health NGOs that wanted a voice in the design and implementation of SIAS. ACECSA, together with some NGOs and donor financiers, correctly argued that the NGO sector had not been taken into consideration when the contracting model was conceived and launched. Other groups, however, responded more cautiously. For example, church groups, unions of health workers, and others, although initially skeptical, saw the model as an opportunity to participate in the system. The MSPAS was proactive with the unions, reach- ing out and communicating the potential benefits of contracting. These activities helped defuse initial union concerns. Although suspicious of the government's ability to pay, a number of large NGOs, wholly dependent on the vagaries of international funding, also saw an opportunity in the government's initiative. They were willing to risk participating in the government scheme as a means to sustain their operations in Guatemala. In sum, most of the elements of the framework relevant to gaining support were addressed in the Guatemalan case. In some cases, they were addressed poorly, which slowed implementation and increased distrust, most notably with the DASs and the NGOs. In other cases, efforts were made to reach out to health workers and unions, for instance. Ultimately, over time, the sheer size of the program created momentum for support. Step 3: Define what will be purchased Conceptual Framework Once a decision has been made to contract, and political and labor issues have been consid- ered, the contract needs to be defined. At this stage, the contracting organization determines 17. Seven NGOs were selected by the MSPAS for pilot projects in 1997. Health System Innovations in Central America 23 what it will be purchasing (for exam- ple, what goods and services), how Box 1.4. Define Services much (volume) to purchase, and for whom (the target population)-- Specify scope and type of services to be provided. see Box 1.4. These are the building Define volume of services to be offered. blocks of both the bid solicitation Define the population to be served. and the subsequent contract, and Stimulate provider innovation. they are addressed through a set of specifications.18 Specifying the required services can be accomplished in several ways. Most contracts specify outputs--the specific services that will be purchased. Outputs should be specified according to the following criteria: (a) completeness (description of all tasks), (b) clarity (unambiguous definition of all terms), (c) measurability (to facilitate contract monitoring and compliance), and (d) focus (congruent with the mission and objectives of the con- tracting agency). Poorly specified services are difficult to monitor or evaluate, but clearly specified services foster accountability of the provider (to the purchaser) for the contracted service. The services identified in the contract must be measurable and independently verifi- able. In this way, the service measures can be turned into clear, objective, and credible per- formance indicators and can be linked to the contractor's compensation. It is preferable for the purchaser to focus more attention on indicators of desired outputs than on process indicators. By doing so, incentives are linked to outputs, or at least to services provided, leaving the contractor with the discretion needed to best achieve the purchaser's ultimate aims. Output definition involves specifying the scope, expected volume, and number of beneficiaries of the contract.19 Although it is preferable to detail outputs rather than focus on resources and processes,20 there are occasions on which the latter can be included in specifications. When outputs are ambiguous or not predictable, it is common for specifications to focus instead on the application of predetermined resources. Nevertheless, first-time contractors of social services are often tempted to define the processes of service delivery together with the configuration of resources that correspond to these processes. However, specifications that focus on the "how" of service delivery may impede contractors from innovating to improve quality or reduce costs. Contracts defined to purchase a specific cluster of services for a given population should give the providers flexibility to innovate by providing ser- vices in new ways, with different kinds of staff, or by substituting elements of the input mix. If the purchaser specifies the inputs in great detail, then the contractor functions much like an inhouse direct provider and the gains from contracting are lost. By contrast, specifying 18. A specification is "what the purchaser seeks to buy and, consequently, what a bidder must be responsible to in order to be considered for award of a contract" (Council of State Governments and National Association of Purchasing agents [1983], cited in Marlin [1984, p. 40]). In short, they define the scope of the services that the contractor is to provide. 19. Establishing target populations will vary depending on the specific characteristics of that popula- tion. Geographically isolated and discrete populations are perhaps the easiest to identify and quantify. In rural and isolated regions, the possibility of leakage toward other groups is minimal. 20. Quality specification is discussed in step 4 in this chapter. 24 World Bank Working Paper outputs sends a strong signal to contractors to craft low-cost and high-quality ways to pro- vide a service. The Guatemalan Experience In developing the package of basic health services (paquete básico de servicios [PBS]), MSPAS leaders worked from epidemiological data for the country and the goals established in the Peace Accords to define what would be covered. The objective was to have a maxi- mum impact on infant and maternal mortality and morbidity at a cost that was affordable and sustainable. The final package included the major elements of prenatal care; postnatal monitoring of mother and infant; well-baby care, including growth and weight monitor- ing; general treatment for common morbidities; and health promotion and sanitation (see Table 1.3). This was the minimum package of services that an NGO could offer, and it was deemed non-negotiable by the MSPAS. Nevertheless, supplementary services could be added at the discretion of an NGO if alternative financing existed. The MSPAS also assigned each NGO a catchment area (jurisdicción) consisting of an estimated 10,000 people, based on available census tract data. If the populations were known to be highly dispersed, a smaller population base was assigned. These catchment areas also represented areas in which the MSPAS had little or no presence. One of the first deliverables of a contracted NGO was to produce a population census of the catchment area. Table 1.3. Composition of the PBS Package Maternal care Prenatal care, birthing, and postpartum care, supplementation, preg- nancy spacing Infant and child care Immunizations, growth monitoring for children younger than 2, and control of acute respiratory infections, diarrhea, and nutritional deficiencies Demand for acute care Cholera, dengue, malaria, tuberculosis, rabies, HIV/AIDS and other sex- and disease control ually transmitted diseases, and urgent care Environmental actions Vector control, zoonosis control, promotion of safe food handling, hygiene and sanitation, and water quality monitoring This package was developed without local input from MSPAS line personnel or the NGOs, who often criticized SIAS directors for forcing a "one size fits all" health care model. The MSPAS was convinced that it knew the priorities of these rural populations and did not see the need to consult with local authorities. It focused on pathologies that impacted mortality. This meant, however, that other common and frequently seen pathologies-- such as skin infections, gastrointestinal problems (in adults), and musculoskeletal dis- orders--would not be included in the PBS. In late 1997, within a few months of contracting the first group of seven NGOs, the MSPAS developed a set of normas de atención (medical guidelines) that broadly defined the services of the PBS but also specified how they would be provided based on national guide- lines. The norms reinforced the "one size fits all" approach, specifying such inputs as the number and type of personnel to be contracted and the responsibilities of each. All NGOs had to adopt the MSPAS-mandated personnel structure, training program, materials, and Health System Innovations in Central America 25 equipment. The norms also defined outputs in terms of the volume and type of services to be provided by each team, which were based on estimates of expected number of annual cases of pregnancies and diseases (such as diarrhea, ARI, and so on) in the catchment areas. Overall, the norms allowed the providers little flexibility in determining how best to deliver the PBS. According to the MSPAS, many NGOs had little experience in pro- viding the services constituting the PBS and welcomed the guidelines.21 However, some NGOs considered the norms a "straitjacket model" that was not appropriate for all sit- uations. For example, the MSPAS mandated the type and numbers of personnel; for example, one health promoter for every 20 homes; one physician per 10,000 popula- tion; and the number of household visits per month by team members. NGOs had to follow the MSPAS program for training personnel and use only MSPAS-prepared train- ing materials and techniques. Not doing so could result in cancellation of the contract. Nevertheless, over time and in light of the heterogeneity of health and geographic con- ditions, the MSPAS began to allow NGOs to part from the PBS and the norms if suffi- ciently justified. The MSPAS's failure to involve or inform the NGOs in its design of the PBS, impre- cise service specification, and weak monitoring (discussed later in this chapter) led to lack of clarity on the part of the NGOs as to the volume and frequency of services that they were expected to provide. As a result, in the early years of the program (1997 and 1998), most NGO providers did not consistently provide the interventions specified in the PBS. Each provided those services in which it was strongest and most experienced. So, for example, vaccinations would be emphasized but growth and weight control might not. Greater upfront involvement and communication, along with specific volume and service indica- tors, although appearing to slow down the process, could in fact have ultimately speeded and smoothed effective implementation. Step 4: Measure and ensure performance Conceptual Framework Once the services and the population have been identified, performance measures are needed to provide the benchmarks for monitoring contract compliance and quality of service provision as Box 1.5. Measure and Ensure Performance well as fulfillment of the broader health goals sought by the purchas- Ensure that the performance measures are ing agency. Specification of perfor- directly associated with the purchasing agency's health objectives. mance indicatorsandcorresponding Select measures that are independently verifiable. reporting requirements are part of Focus on output and outcome measures. the contract, and they form the basis of an effective contract-monitoring Be sure to specify the quality standards and cor- responding measures. system (see Box 1.5).22 21. A training guide was developed to train providers in how these norms would be applied. 22. Monitoring is discussed in step 8 in this chapter. 26 World Bank Working Paper First, the purchaser needs to ask whether it can develop performance measures linked to the public health objectives of which the contracting strategy is one component. Second, is the measure independently verifiable? Although some measures will be dependent on record keeping by contractors, independent sources of information are needed to verify and validate the results. Such sources might include household surveys, health facility sur- veys, or medical record audits. The purchaser faces a wide range of potential measures. These can include: input measures, such as staff hours worked, number of vaccines supplied, oral rehy- dration solution (ORS) kits provided; intermediate or output measures, such as the number of primary care visits, family planning classes provided, home visits by community health workers, and techni- cal quality (that is, application of clinical protocols); and outcome measures, such as a lower incidence of diarrhea among children, reduc- tions in maternal mortality, and user satisfaction with health services. Input measures are perhaps the weakest measures of performance because, alone, they give little information on the result of the contractor's actions. Output measures are important and can correspond closely to the scope and volumes specified in the contract. However, output measures (such as productivity) do not gauge efficacy or quality. Outcome measures are most directly linked to overarching policy objectives but may be more difficult to attribute to the particular contractor's efforts because of long lag times and multiple determinants. An important outcome measure--often overlooked in the specifications--is quality. The contract should specify quality standards expected for each service or activity. Clinical or technical quality measures for most types of health services are increasingly available, and they can be adapted to contractual specifications. Also, con- sumer satisfaction measures are always important in ensuring that the quality of the ser- vices provided is satisfactory to the ultimate recipient. The Guatemalan Experience Although service was weakly defined, performance measures were absent, especially in the early years of the contracting process. When contracting began in early 1997, there were no performance measures included as part of the convenios between the MSPAS and the NGOs. The norms of care focused on process but not on outputs or ultimate results. No baseline health or utilization data were collected in the catchment areas.23 According to SIAS norms, contracted NGOs had to comply with three measures: (1) regularly send demographic, service production, financial, and some epidemiological information to the MSPAS; (2) provide all the interventions contained in the PBS; and (3) cover the entire population in the assigned catchment area. In practice, the MSPAS only effectively monitored and held NGOs accountable for supplying the information specified in the first requirement during the first three years of SIAS implementation. As long as the 23. Given that all contracted NGOs had to conduct a household census to establish the population in its catchment area, this was an opportunity to collect other types of data to facilitate setting performance measures and subsequent evaluation. Health System Innovations in Central America 27 NGOs provided the required Table 1.4. National Goals of the SIAS Program information on a more or less regular basis, they were in com- Measure Specific Five-year Goal pliance with MSPAS contractual Vaccination coverage in >90% of covered population requirements.24 children under 5 The SIAS program had es- Prenatal care At least 75% of covered population tablished an overarching set of Coverage for antitetanus At least 75% of covered goals that it wanted to accom- vaccination in women in population plish at the outset. Key items are their childbearing years shown in Table 1.4. These met- Growth monitoring of At least 75% of covered rics were not, however, initially children under 2 population linked to contractual measures Availability of ORS kits in 100% of families covered of performance of the NGOs. every home Therefore, the accountability of Presence of medications At least 80% of essential medications (at health posts NGOs to move toward meeting and NGO-operated services program--and national--goals (botiquines básicos) of service provision and mortal- ity reduction was slight. In regional meetings, the MSPAS provided often inconsistent signals, directing the NGOs to emphasize specific interventions (such as immunizations) of the PBS. In sum, for most NGOs, the "what" and "how" of performance assessment was unclear at best. Three years after startup, however, with experience and program expansion, perfor- mance indicators were gradually brought into the contracting process. Supporting this process was a performance-monitoring system, Habilitación, Adjudicación, Certificación y Acreditación (HACyA (licensure, adjudication, certification, and accreditation), that annually assesses NGO performance based on a series of metrics related to community organization, training, service coverage, and financial management (see Table 1.5).25 These standards also became the threshold for new organizations seeking to enter the program. The HACyA indicators focused on a subset of PBS interventions--vaccination, growth monitoring, micronutrient supplements, and prenatal care. This made sense because they were the key interventions of the PBS and, given the weak capacity of the MSPAS,26 they were also the activities that the MSPAS would be capable of monitoring. The HACyA sys- tem also assessed financial management performance.27 Thus, over a two- to four-year period, performance measurement in the SIAS program evolved from a somewhat haphazard process to one that gradually moved toward a more focused and performance-based orientation. Since 2000, the performance indicators have been written into the contracts and used to determine contract renewal. However, only the indicators are included. The performance targets for each NGO are not specified, in part 24. Performance monitoring is discussed in step 8 in this chapter. 25. The ADMSSs were rated on financial management criteria only. 26. At this writing (September 2004), only the certification and adjudication aspects of the program have been implemented. 27. As described in step 8, a point system was developed for both technical (community organization, training, and coverage) and financial management aspects. The combined technical and management scores determined the type of contract renewal (full renewal, conditional renewal, or cancellation). 28 World Bank Working Paper Table 1.5. Indicators for Determining NGO Contract Renewal Assessment category Indicator Community organization Number of volunteer community workers selected Number of community midwives selected Existence of community census Existence of community map Existence of community pharmacies Training Number of workers trained in at least 6 basic themes: mapping, cen- sus taking taking, ARI, diarrhea, vaccination, and growth promotion monitoring Coverage Prenatal care (%) Iron and folate supplementation for pregnant women (%) Pregnant women with a 2nd dose of tetanus toxoid (%) Infants with DPT3(diphtheria, tetanus, and pertussis), polio3, BCG1 (Calmette-Guérin bacilli, for tuberculosis), and measles vaccination (%) Children under 2 with growth monitoring (%) Children under 2 with iron supplementation (%) Administrative-financial Complete ledgers of all expenditures Inventory system up to date All personnel hired with formal contracts Monthly financial and bank statements Quarterly financial execution reports Evidence that all public financing used for PBS activities (% budget execution) because of the absence of baseline data. Rather, global targets are used to rate the NGOs on coverage (see Table 1.4). As discussed in step 8, the contract assessment system essentially became equivalent to the monitoring system. Performance indicators signaled NGOs as to the MSPAS's pri- orities related to a subset of PBS interventions. However, despite the turn toward perfor- mance measures, including an increasing focus on outputs and away from process indicators, the role of the HACyA process as it relates to ensuring quality was not discern- able. Another shortcoming is that performance continues to be self-reported. Household- based assessments of provider performance have yet to be implemented. Step 5: Define the payment system Conceptual Framework The choice of payment mechanisms and the linkage of payment to performance are two of the greatest drivers in the ultimate success of the contract (see Box 1.6). Payment systems can consist of standard payment mechanisms as well as incentive payments. Although no one payment mechanism will be appropriate in all cases, there is evidence that whatever Health System Innovations in Central America 29 method of payment is selected, linking some portion of it to key performance indicators Box 1.6. Define a Payment System improves overall performance. In health service contracting, the range Choose a payment mechanism (such as fee for service, case rate, or capita- of payment mechanisms available to the pur- tion) that creates incentives for effi- chasing organization is wide. These can in- ciency and quality. clude global budgets, fee for service, capitation, Define systems of performance-based and combinations of these mechanisms. Each reimbursement. must be weighed in terms of the incentives or disincentives they imply for important criteria such as efficiency, quality, volume, and potential for fraud. Payments may be increased for good performance or reduced for poor performance. Including incentives in contract specifications signals to the provider which services, activ- ities, and quality standards are of special interest to the purchaser. There are two types of incentives that have been commonly applied to health service contracting: performance withholds and bonuses. Performance withholds are portions of the contractor's compensation that are held back by the public agency and released only upon evidence of success in attaining agreed- upon performance goals. There is substantial evidence that relatively modest withholds (as little as 5 percent) can encourage significant improvements in contractor performance. Bonuses are an additional payment, predetermined in the contract, that are paid to the contractor upon the successful attainment of specific performance indicators. Similar to withholds, it appears that relatively modest sums can significantly improve contractor performance. These methods are not necessarily exclusive of one another. A withhold or a bonus can be tied to overall performance (quantity and quality of all specified services) or applied to a particular subset of services or indicators. Developing performance measures and selecting a payment method or methods are necessary but not sufficient conditions for putting in place a complete performance-based reimbursement system. The power of such a system is that contractors realize financial gains for performing in ways that optimize the results of the program under contract and further the health policy goals of the contracting government organization. For this sys- tem to be effective, a subset of performance indicators must be selected, and weighted, through joint discussions with the contracting organization and the contractor(s). Second, the specific terms of the incentive have to be negotiated between the two par- ties. The incentive should be significant enough to have an impact on the contractor with- out transferring an unmanageable amount of risk and thereby inhibiting the ability to provide services. Generally, 5 to 10 percent of costs have been shown to be effective as with- holds or bonuses, particularly in early phases. Over time, the share of payment that is at risk can increase as each party gains experience in the administration of the contract provisions. The Guatemalan Experience The MSPAS developed a per capita financing system to pay NGO contractors. For a given catchment area (the level at which contracts were provided), these data were derived from estimating each of the budget items associated with providing services, such as salaries of 30 World Bank Working Paper health care staff, equipment, medications, training, and overhead. Based on MSPAS spend- ing ledgers, costs were developed for each service element, aggregated, and divided by the population expected to be served under a contract. This amount totaled US$8 per capita per year, including in-kind distribution of vaccines. The MSPAS permitted a 10 percent adjustment for higher transportation costs for NGOs working in remote areas. Echoing the level of distrust that existed between the MSPAS and the NGOs during the early years of implementation, DASs of the MSPAS saw this as a very high payment and commented that the NGOs would become wealthy from the program. Contrarily, NGOs and international agencies providing technical assistance to SIAS suggested that the per capita rate was too low for the services provided. Cost data from the NGOs were never col- lected by the MSPAS to assess the appropriateness of the capitated rate or the NGO claims of "low-balling." Further, as of 2002, the per capita rate applied in 1997 had not adjusted for inflation. The MSPAS's refusal to adjust the rates resulted in five NGOs to dropping out of SIAS in 2000 and 2001. According to the representatives from these NGOs, the MSPAS would not consider their economic proposal even when they presented actual costs. These were mainly local NGOs without access to international and other sources of financing to fill the funding gap attributed to below-cost MSPAS payments.28 The development of a capitated approach to a defined population for a specified set of services made a good deal of sense as a choice of payment, but there were no offsetting incentives to encourage appropriate provision and use of services. Together with the low payments, this may have contributed to NGO provision of only a subset of the PBS and a lower than expected population coverage. To make the payment system work, MSPAS financial managers had to create an entire financial flow system to enable the channeling of resources (and corresponding manage- ment and auditing systems) from the MOF to the MSPAS and then to the NGOs. This was no minor feat, given the highly centralized context of government financial management. The MSPAS was fortunate to participate as a pilot ministry in a general reform process to decentralize government financial management. The overall financial reform effort involved29 creation of a financial management unit within the MSPAS that was ultimately responsible for paying providers, establishment of the regulatory framework and corre- sponding procedures to permit MSPAS management of the payment system, creation of budgetary line items and a financial reporting system for contracted services, and the set- ting up of an open contract system of pooled procurement and decentralized purchasing and delivery of drugs and medical supplies. This system allowed NGOs to purchase directly from approved suppliers at favorable prices negotiated by the MSPAS. Payments were made quarterly to NGOs. However, the consistency of payments was a problem because the MSPAS was often late. Because the ADMSSs and PSSs were depen- dent on these payments to purchase supplies and pay staff, this often resulted in empty pharmacies and limited supplies. To alleviate this problem, a more complex arrangement 28. Applying price competition features to the bidding process could address this situation, but was rejected by both the MSPAS and NGOs. Price competition is discussed in step 6 in this chapter. 29. It is important to note that most of these items were disbursement conditions included in the Inter-American Development Bank's adjustment loan, providing a strong incentive to the MOF and MSPAS to work together. Health System Innovations in Central America 31 was established under which the necessary funds were directly transferred to the NGOs, using the banking system, with the approval of the MoF as the intermediary. Although this was an improvement over previous arrangements of channeling payments through area offices, payment delays persisted. Step 6: Select providers and maximize competition Conceptual Framework Selecting providers through a competitive selection process Box 1.7. Select Providers and Maximize Competition is considered best practice in public contracting (see Box 1.7). Evaluate the concentration of providers. The power to select and encour- If the market is limited, evaluate the feasibility of a age competition is an important contestable contract. public policy tool. Also, trans- Evaluate readiness of providers to compete. Choose a parent competitive bidding is competition-based selection or bidding method. generally the best way to avoid Avoid pitfalls that can result in a legal challenge after award of the contract. favoritism and fraud. In some cases, competition is not possi- ble because of an undeveloped market or because of the nature of the service. Consequently, contracts are awarded pursuant to negotiations. Selecting providers and maximizing com- petition will be a function of at least three factors, which are discussed below. First, is there a sufficient concentration of providers (and concentration of population to be served) to support competitive bidding? In general, the greater the number of poten- tial providers, the greater the likelihood of sustainable competition. Second, even if a market has limited providers and consists of a geographically isolated and unserved or underserved population, the market can still be made contestable. Con- testability occurs when providers need to bid for concessions over particular areas, where there are few barriers to entry, or both. Third, what is the readiness of providers to compete for contracts? If the contractors are well prepared and have concentrated market shares, this may create severe challenges for the contracting agency. However, if potential contractors are inexperienced in con- tracting, especially with public agencies, technical assistance may be needed to prepare these providers to develop the management, cost accounting, and information systems necessary to conform with performance-based contracts. The structure of the contract may also need to be relaxed in the first few years to ensure that the barriers to competition are not so high as to exclude potentially successful bidders. Expectations of administrative and clinical performance can be gradually tightened. Thus, the role of the contracting agency is also as a facilitator in developing the market readiness of potential contractors. Both parties can benefit in the long run. There are multiple approaches to the actual bidding. Providers generally compete for health service contracts based on requests for proposals (RFPs) in which potential providers are requested to submit separate technical and financial proposals. The selection of providers and procurement must comply with national public contracting regulations, but 32 World Bank Working Paper most public contracting laws in developing countries are oriented toward the purchase of goods and works rather than services. Although specifications and price offers are generally non-negotiable in bidding for standard goods, RFPs for services generally involve pre- and post-award negotiations on both technical and price aspects of the winning proposals. This difference in approach is necessary because of the complexity of defining the services to be provided, the importance of ensuring quality, and the fact that the contract will create a relationship rather than sig- nify a discrete transaction. In preparing and conducting the bidding process, public agencies should exercise cau- tion to avoid future litigation or challenges from (losing) bidders. Potential pitfalls include using non-neutral consultants in the preparation of bid solicitations, applying service spec- ifications supplied by a particular provider, specifying services and resources that favor a particular bidder, having informal contacts with bidders, and failing to maintain accurate and complete records of compliance with the procurement process. If there are multiple bidders who are relatively sophisticated with regard to contract compliance, then the public purchaser may be able to unilaterally issue a well-specified bid with terms, measures of performance, and payment mechanisms in the expectation that the providers will be able to respond to those terms. If, however, contracting is a relatively new concept, contracts are of limited dollar value, a monopoly provider dominates the market, or there are few bidders, then competitive bidding may be difficult to implement. In these cases, services can be procured through negotiation--that is, noncompetitive pro- curement of services. The Guatemalan Experience The provider selection process for SIAS underwent a substantial transformation from its early days in 1997 to 2002. Among the issues that the MSPAS and SIAS managers had to address were who would select providers and what decision criteria would be used in mak- ing the selection. The initial plan called for a pilot project in three departments that would run for three years. Using data that had been gathered by the MSPAS, seven NGOs in the three depart- ments were approached and invited to become the providers of services. Consultants guided the MSPAS in preparing an initial RFP and evaluation guidelines. Beginning in 1997, pilot programs were put in place in Escuintla, Alta Verapaz, and Chiquimula. The seven selected NGOs also received technical assistance in preparing their proposal. No for- mal assessment of their capacity was performed. All proposals received from the groups were accepted. Because not every NGO in the departments was included (especially smaller NGOs), there were complaints from some concerning the nontransparency and lack of fairness of selection. There was a general perception that the MSPAS was selecting NGOs that were "friendly" toward the government. Thus, there was no competition among NGOs in the start-up phase, and there was limited information about NGO quality and ser- vice capability. Largely as a result of political pressures to show results as well as initial program suc- cess and the large unmet health needs in the country, the GOG made a decision, after only six months of operation, to move from a pilot basis to a nationwide program. At this point, in early 1998, the government turned to a process of convening potential NGO contrac- Health System Innovations in Central America 33 tors to present the proposed services and jurisdictions to be covered, along with the asso- ciated payments. With the national roll-out came national conferences for a much larger group of NGOs, in which proposals were developed. By this time, there was considerably more demand from the NGO community, in part because many of the original fears about nonpayment and corruption had been allayed. The MSPAS continued to apply an open contract or long-list strategy in 1998 and 1999, in which any NGO presenting an acceptable proposal and demonstrating a minimal degree of capacity were contracted. A simple, two-step assessment process was developed in early 1998. The first step consisted in demonstrating at least one year's experience in ser- vice provision (for a PSS) or project administration (for an ADMSS), legal personality, and financial management structure. The second step assessed the NGOs on their capacity to organize, (financially) manage, and deliver the PBS. Organizational and financial man- agement capacities were the key selection criteria. This second level of review resulted in an overall score for each PSS and ADMSS based on a point system. Although selection criteria were still ambiguous and led to continuous complaints of favoritism from rejected NGOs, and NGOs were still invited to apply (and selected) in a less than transparent manner, this new, two-step process was a vast improvement over the preselection process of a year earlier. Many NGOs recognized that the MSPAS was mov- ing toward setting clears rules. The open contract strategy enabled a rapid selection and contracting process and, therefore, accelerated extension of SIAS coverage.30 Skepticism gave way to eagerness on the part of most of the NGOs to participate in this new program. The number of NGOs wishing to sign agreements grew dramatically as well. Figure 1.1 illustrates the rapid growth of contracted NGOs and signed agreements, with 74 NGOs signing 110 agreements in 1998. The MSPAS continued to improve in the selection process. The MSPAS was sensitive about complaints of nontransparency and favoritism, but the market was also becoming Figure 1.1. Growth in NGOs and Contracts, 1997­2002 180 160 140 120 100 No. of NGOs 80 60 no. of 40 Contracts 20 0 97 98 99 00 01 02 19 19 19 20 20 20 Source: MSPAS/SIAS 30. This strategy did not provide any incentive to the NGOs to exceed the minimal requirements spec- ified by the MSPAS when they were developing their proposals. 34 World Bank Working Paper more competitive as more NGOs sought participation and NGOs that already had contracts sought to increase market share. The MSPAS had to find a way to select the best NGOs among many applicants. Beginning in 2000, a more sophisticated process was established and supported by the concurrent development of the HACyA system (described in step 4). New selection criteria and a corresponding point system were developed to evaluate NGO proposals, but unlike the previous process, all requirements and procedures were made transparent. Potential NGO applicants were fully informed of the requirements, proposal due dates, and procedures. Preapplication conferences were announced and held to fully inform potential applicants of the process.31 Again, selection consisted of a two-step process: The first step consisted of licensing of NGOs as either a provider (PSS) or administrator (ADMSS). To obtain a license, applicants had to present infor- mation on legal status, location (preferably in rural areas), bank accounts, adminis- trative and financial personnel, infrastructure, office equipment, vehicles, and (for PSSs) evidence of experience in providing basic care. An MSPAS representative made a site visit to each applicant to verify information provided in the licens- ing application. In the second step, the MSPAS opened a competition among licensed NGOs, inviting technical bids through an RFP to licensed NGOs. Terms of reference were specified in proposal guidelines developed by the SIAS program. Because the price paid by the MSPAS is fixed, NGOs competed on the technical merits of their proposals. Technical criteria were established to assess the proposal, and the final decision was made by a committee consisting of central- and regional-level MSPAS staff. This step would be followed by signing the contract and assigning the specific catchment areas to the winning NGOs. As described in step 4, the HACyA system also assessed and certified NGOs that already had contracts. Contract renewal depended on successful certification based on perfor- mance indicators related to community organization, human resource supply, service coverage, and financial management benchmarks. Introduction of the HACyA process in 2000 was a big step toward more selective contracting Table 1.6. Results of Contract Renewal and NGO of NGOs. Table 1.6 presents Selection Processes, 2000­02 summary data on the results Status of agreements 2000 2001 2002 of the HACyA process on contract renewal and issuing Not reneweda 24 13 4 of contracts between 2000 Conditionally renewed 40 8 17 and 2002. Renewed without conditions 71 114 129 It is noteworthy that over Not assessed 0 19 11 this period, 41 contracts were Total 135 154 161 cancelled because of failure to a. Contracts were tendered to other NGOs. meet minimum performance Source: MSPAS, SIAS. 31. The MSPAS published an announcement of upcoming preapplication conferences in the local press. Health System Innovations in Central America 35 standards (see Table 1.6). Also, contracts were tendered and later signed with other NGOs applying a more rigorous, transparent, and apparently competitive selection process. The creation of clearer "rules of engagement," transparency, open communication, and setting of performance standards for issuing and renewal of contracts improved the quality of the requests for proposals and their responses. These factors resulted in a more competitive environment. In fact, by 2001, there were more NGOs seeking agreements with the MSPAS than there was demand. The government did not implement a full bidding process, as outlined in the framework section above. Because the MSPAS fixed the payment, NGOs did not compete on price. Many NGOs rejected competition and considered it more appropriate for the for-profit sector. In fact, most NGOs would not have participated in SIAS under price competition. Also, because an NGO had to be licensed for specific municipalities where it had an established presence, it was unclear how contestable the contracting process was. The licens- ing process required that each applicant demonstrate a presence in the municipality where it was seeking to provide services. In effect, the NGOs are licensed for specific municipalities. The extent to which NGOs competed head to head for a specific municipality--or, alter- natively, NGOs located in neighboring municipalities competed for contracts out of their geographic jurisdiction--is unknown. However, MSPAS staff who participated in the selec- tion committee claim that for most catchment areas that were bid (because of expansion or cancellation of contract) after 1999, there were at least two candidates. In less remote catch- ment areas, up to eight NGOs applied. Step 7: Prepare and write the Contract Conceptual Framework The final contract is the product of negotiations over the proposal submitted by a particular provider. If the RFP is well prepared and services are clearly specified, then the contract can draw heavily upon it. Once the type of contract and the process by which it will be developed have been established, the individual contract components need to be specified. Most terms are placed in three main chapters: respon- sibilities of the contractor, responsibilities Box 1.8. Prepare and Write the Contract of the purchaser, and general terms and conditions. Many of the contract compo- Decide how much collaboration will be nents have been previously mentioned in involved in development of the contract. this chapter. This section highlights key Define contract terms. provisions (see Box 1.8). First, how long will the contract be in effect? Are there provisions for renewing the contract? Will there be open bidding at the end of the contract period, or can there be a contract extension dependent on good performance? Setting performance standards as a condition for extending the contract can be a strong incen- tive for the contractor to perform well because it can eliminate the need to go through the sometimes arduous contracting process again. Second, the contract must specify what services will be provided, including frequency, geographic accessibility, and record keeping. The contract should also clearly define the 36 World Bank Working Paper population to be served. As discussed in a previous section, it is best that the contract specify the desired outputs and results, including how, when, and by whom success will be assessed. Third, the payment section of the contract addresses method and timing. Timing can be regular or paid in discrete sums (for example, quarterly, semiannually, or annually). Many contractors may not have access to the credit needed to meet expenses on a regular basis if their payments come in large discrete sums. This section of the contract also needs to specify if some payments are linked to performance, and if so, how meeting targets will be verified. Finally, the contract must define the conditions under which either party can termi- nate the contract as well specify dispute resolutions mechanisms. Contractors need some predictability in their contracts against unnecessary financial risk, but provisions are also necessary for suspending or discontinuing the contract for cause. Such causes can include failure to perform, fraud, or other forms of corruption. The Guatemalan Experience The MSPAS leadership faced an important decision at the outset of the contracting process. Previous to SIAS, there was only one form of contracting, the Public Contracting Law (Ley de Contrataciones del Estado [LCE]), which regulated all commercial contracts. Typical of contracting laws elsewhere, the LCE specified standard procurement procedures through competitive bidding, including prequalification of suppliers, bid solicitation, bid evalua- tion, and contractor selection. The MSPAS considered the LCE onerous, more appropri- ate for contracting with for-profit commercial enterprises (to provide services such as communications, electricity, construction, and transportation), and inappropriate for social services, especially contracted through nonprofit providers. Complying with LCE mandates would involve prohibitively high transaction costs for the NGOs and therefore severely limit market supply. For example, most NGOs could not afford LCE mandates related to application fees (estimated at US$9,000) or secure financial guarantees repre- senting 15 percent of contract value. Applying the LCE provisions would favor for-profit firms that could afford the upfront fees. Although the extent to which any for-profit firms would be interested in providing basic care to rural populations is unknown, the MSPAS sought to exclude the for-profit sector from SIAS.32 The LCE also required the contractor to pay taxes on revenues received from the MSPAS, yet the NGOs possessed nonprofit status. To enable contracting, the MSPAS had to modify the legal framework as well as cre- ate a contracting instrument that was less burdensome that that specified in the LCE. First, the MSPAS moved to include an article (number 24) in the new Health Law that was approved by the Guatemalan Congress in early 1998, permitting the contracting out of health services.33 This was a huge political victory for the MSPAS and essentially provided a green light to scale up the fledgling contracting effort. 32. As mentioned, the MSPAS was very sensitive to criticisms from both inside and outside govern- ment regarding the "privatization" of health care. Authorities also feared that many NGOs would not par- ticipate in a competitive bidding process, particularly if for-profit firms also participated. 33. According to article 24 of the Health Law, "For the purposes of coordinating the provision of health services, the Health Ministry may enter into agreements and contracts with the institutions com- prising the Sector and other institutions related therewith. Likewise, the Sector's public institutions may, by means of agreements or other legal instruments, enter into service delivery agreements with each other and with private organizations." Health System Innovations in Central America 37 Second, given the potential bottlenecks, lag times, and lack of flexibility associated with the LCE combined with the urgency for quickly developing written relationships with NGOs, the MSPAS crafted a new instrument, the convenios, that in effect represented a new regulatory type of contract. Outwardly, the convenio itself appeared similar to the standard contract. However, through the use of the convenios, the MSPAS sidestepped processes and procedures relevant to contracting that are specified in the LCE. This did not go unnoticed. The convenios were challenged in court for being out of compliance with the LCE. The MSPAS defended the legality of the instrument, claiming that it was supported in the 1998 Health Law. A 1999 court decision favored the MSPAS position. Although the legality of the convenios remains an open question, the MSPAS was permitted to continue applying this instrument. Although they were still in use in 2004, their long-term viability remains to be seen, as did the possibility of challenges by the GOG to convert to the more formal contracts under the LCE. Table 1.7 compares elements of a stylized contract and contracting process per the LCE with comparative features of the convenio. The convenio defines the specific services to be cov- ered, the population (juris- dicción) to be served, and the Table 1.7. Comparison of Processes and Provisions financial amount of the con- Related to an "Ideal" Contract and a Convenio tract. The contract also in- "Ideal" Contracting and Contract Convenio cludes sections that address Provider selection terms of payment from the Competitive bidding No MSPAS and conditions for Prequalification of providers Yes (after 2000) payment. With the exception Publication: bid notice or call Yes (after 2000) of features related to com- for proposals petitive bidding, posting of Posting of contract performance bond No bonds,andperformancemea- Specification clauses (for contractor) sures, most of the elements Beneficiaries Yes of a stylized contract were Services By service category incorporated into the conve- Personnel Yes (after 2000) nio by 2000. Performance requirements Partly (after 2000) After 2000, especially Incentives No with the onset of the HACyA Reporting requirements Yes system and as performance Responsibilities of the MSPAS indicators became more in Payment and payment procedures Yes line with overall program ob- Performance review procedures Yes (after 2000) jectives, contract terms be- came more explicit. Specific Monitoring procedures Yes, but not specific expectations for results were Provisions for contract change developed and incorporated Dispute resolution Yes into contracting. However, Unanticipated work or events Yes as of 2002, the contracts were Provision for contract duration Yes lacking positive incentives Provisions for contract noncompliance Yes for the highest-performing Provisions for contract termination Yes ADMSSs and PSSs to receive Authorized signatures Yes financial remuneration for Source: Adapted from Nieves and LaForgia (2000, p. 6, table 1). outstanding performance. 38 World Bank Working Paper Rather, performance indicators were used to exclude, renew, or add new ADMSSs and PSSs. An important test of the durability of these convenios occurred in 2000, when there was a leadership change in government. Normally, in Guatemala during a regime change, it would not be uncommon for existing contracts to be nullified, and the convenios' ambigu- ous status made that possibility more probable. However, despite a complete turnover in leadership, all the convenios remained intact and the system continues to be refined. Finally, the SIAS program coordinator, with central MSPAS staff, developed the con- tract. Involvement with potential contractors remained minimal because the expertise to prioritize and cost out services was seen as a capability that existed among the MSPAS and supporting technical experts. Step 8: Monitoring compliance and evaluating contract performance The Conceptual Framework The decision to contract involves both a new set of operational capabilities and a philo- sophical shift (see Box 1.9). Instead of micromanaging the direct delivery of services (as is common in the ministries of health of many developing countries), the contracting agency is managing through attention to results. This leaves significant discretion to the contrac- tor(s) in deciding how best to achieve these results. With this new perspective comes a need for new kinds of analysis. Instead of analyz- ing the best way to produce services, the contracting agency has to be able to analyze the market, evaluate potential contractors, and monitor contract compliance and perfor- mance. Within the overall contract management function, there needs to be the capability to monitor results, deploy on-the-ground evaluation teams, and develop protocols for identifying early warning signs of contract noncompliance, including necessary corrective actions. Also, prompt, accurate payment to contractors is essential for the function to be cred- ible and sustainable. A separate fiduciary entity, either within the contracting department or through a third party, needs to track contract costs and payments for each and every Box 1.9. Monitor Compliance and Evaluate contract and work closely with Contract Performance those monitoring contract com- pliance to ensure that payments Make payments to the contractor in a timely manner, as specified in the contract. linked to performance are made appropriately. Reward excellent performance. Establish formal moni- toring systems. How the contract will ulti- Communicate regularly with the contractor to solve mately be managed depends on problems early and maintain rapport. the information that comes Develop an agreed-upon progression of management from monitoring the contract, activities to address inadequate performance. responses to poor performance, Develop a contracting organization. and the internal capabilities that Evaluate performance. the contracting organization Health System Innovations in Central America 39 must possess or acquire to be successful. Some combination of the following monitoring instruments can be applied: (a) application of tracking tools to monitor progress and schedules, (b) a reporting and information system to monitor technical performance and costs, (c) periodic independent assessments to assess specific performance components (such as client satisfaction surveys), (d) on-site inspections, (e) client complaint reporting and management system, and (f) purchaser-contractor monitoring conferences. It follows that effective contract oversight requires effective reporting and information systems that integrate inputs drawn from application of the above instruments. Some form of internal information system is critical to track contracts: their duration, budget, perfor- mance measures, payment system, and payment schedule. This is an area in which techni- cal assistance or the sharing of best practices from other contracting organizations can be essential to avoid unnecessary duplication of efforts. Finally, managing a contract when the contractor is performing inadequately is not an all-or-nothing activity. Rather, the purchaser needs to be proactive. The purchaser needs to identify risks of noncompliance with performance targets early on to collaborate in solv- ing problems when possible. Only when this fails should the purchaser move toward apply- ing the sanctions set out in the contract or, ultimately, terminate the contract if there is sufficient cause. The consequences in each case should be spelled out as clearly as possible in the contract itself. In cases where there are relatively few contractors, the contracting agency may prefer to work more closely with the provider, giving targeted technical assis- tance. Alternatively, the contracting agency may organize opportunities for similar con- tractors to share experiences and learn from each other's best practices. The Guatemalan Experience Monitoring and evaluating contract compliance has been a challenge to SIAS since its onset. During the early years, monitoring was sporadic at best. Fearing charges of corrup- tion, initially at least, oversight focused on financial reporting and tracking of use of funds by NGO contractors. For this purpose, in 1998, a financial management unit was estab- lished inside the MSPAS and financial officers were hired for each district where NGO con- tractors were providing services. The unit developed clear guidelines (with manuals and corresponding forms) for financial reporting requirements, trained NGO administrative personnel, and conducted site visits. Although site visits were irregular because of a small number of staff available, the only aspect of the overall monitoring system that worked consistently was financial oversight. The NGOs also had a strong incentive to comply with financial-reporting mandates because these made up the criterion was linked to future payments. For example, ADMSSs and PSSs had to submit quarterly financial reports and provide evidence that at least 75 per- cent of the previous transfer payment had been spent before they could receive another allo- cation. Compliance was high, though some NGOs were denied payments because of delays in submitting the financial reports or because of inconsistencies in the information reported. In all cases, these inconsistencies were due to neglect and low capacity rather than fraud. Monitoring of service delivery and coverage was much less developed, particularly early on. Monitoring was originally delegated to the district level. Specifically, an area facil- itator (facilitador del área), who in theory reported to the DAS, was responsible for the ongo- ing monitoring of ADMSSs and PSSs. In the beginning, the facilitators used primarily 40 World Bank Working Paper process measures from the medical guidelines in their assessment. They had the authority to highlight problem areas and, in the most extreme cases, recommend contract cancellation. The facilitators were inconsistent in their monitoring responsibilities, and the facilitator- based system--as well as the position of facilitator--were gradually eliminated. Although MSPAS area and district officials were responsible for monitoring the NGOs, the facilita- tors tended to be more responsive to the central MSPAS. At the same time, they were to serve as consultants to the NGOs to help them improve the quality of services they pro- vided. In reality, these roles proved to be in conflict, and resulted in weak monitoring of the NGOs. Although "on paper" facilitators did have the authority to progressively mon- itor performance (up to recommending that a contract be terminated), it is not clear how well defined the steps in this process were. Also, all facilitators were physicians and focused their efforts on supporting the NGOs with medical care delivery; most were uninterested in administrative aspects of the monitoring process. Another weak link in the monitoring process was the low capacity of district and area officials to consolidate and analyze the monthly and quarterly statements sent by the PSSs and ADMSSs.34 In sum, monitoring was more informal than systematic. From 1997 to 1999, the monitoring system resulted in lit- tle knowledge of real population coverage of PBS activities delivered by the NGOs. The MSPAS was informed by the facilitators that few NGOs were delivering the entire PBS, but it did not have the data to determine the extent of the situation. With the implementation of the HACyA process for contract renewal in 2000 (described in step 4), the certification process became the main feature of the monitoring system. NGO contractors had to supply the MSPAS with reports on technical performance and resource allocation and use (see Table 1.5). Conducted annually, the assessments con- sisted of a review of data provided by the NGOs and site visits by certification teams. A set of indicators was defined for both technical and financial management aspects, and each set was scored on a 100-point scale. If the total average (unweighted) score fell below 40 for either set of indicators, the contract was not renewed. For NGOs receiving a score between 40 and 79, a conditional renewal was granted in which the MSPAS indicated spe- cific areas of improvement as stipulated in a letter of understanding. The NGOs were given one year to address the issues raised by the MSPAS. Failure to do so would mean automatic cancellation of the contract. To support the monitoring and recertification process, the MSPAS created a moni- toring unit to consolidate and analyze HACyA-mandated data on an annual basis. In an effort to incorporate data from both public and contracted programs by district, a situa- tion room (sala de situación) was also established in each district in 2000. The situation room was essentially an informal process that depended on the commitment of MSPAS district authorities. They were intended as coordinating points for data collection of both publicly and NGO-administered health systems, with the goal of creating a network of basic statistical data for an entire district. However, as of the most recent internal program evaluation, the effectiveness of this effort had not been established. Problems continued in the timeliness of payment, despite the intervention of MoF and direct bank transfers to the NGOs. This caused frequent, if intermittent, operating prob- 34. The monitoring system was ad hoc and caused documentation burdens for district officials. NGOs were mandated to supply, monthly or quarterly, at least eight separate forms on service production, mor- tality, activities, and financial management. Health System Innovations in Central America 41 lems for the contractors, and limited availability of medicines, supplies, and equipment at certain times during the year. A contracting organization continues to be under development. Strong capacity existed at the program coordinator level to develop such an organization, but more staff are required to fully implement contract management and monitoring functions. The area of contracting infrastructure--including market analysis, financial management, and information technology--continued to require development. Capacity at the district level, where decisions concerning provider selection and retention were increasingly made, was more uneven. Step 9: Encouraging competition in the long term The Conceptual Framework A well-designed contract and corre- sponding monitoring system will not Box 1.10. Encourage Competition in the necessarily lead to continuing bene- Long Term fits and savings (see Box 1.10). Com- petition is another key element for Create incentives to organize the supply side of services in the NGO and private sectors. the long-term success of contracting. In isolated markets, consider the creation of con- Costs shift, new suppliers may enter testable units of care provision. the market while others leave, and Anticipate transitional difficulties. new technologies may become avail- Establish contingency plans to reduce the length able. The purchasing organization of service interruptions. Reduce the risk of cor- should continuously explore the mar- ruption and fraud. ket environment for these changes. Establish formal policies for contracting. Competition drives contractors to Prevent contractor inertia or fatigue. search for better ways to improve quality, increase client satisfaction, and lower costs. Stimulating a competitive contracting process is no easy task--maintaining it can be even more challenging. Supplier unavailability, transitional difficulties resulting from switching suppliers, service interruption, corruption, and supplier inertia can inhibit the functioning of the competitive contracting process. Table 1.8 presents recommended solu- tions to these obstacles to ensuring competition. Purchasing agencies should regularly assess the state of the competitive environment. This can be accomplished through two measures. The first is a performance review in which a single contractor's performance is compared with others. The second is a bidding test. Rebidding contracts from time to time provides reliable feedback on whether the pur- chasing agency is paying the best price and provides information on the number of poten- tial suppliers in the market. The Guatemalan Experience SIAS's progressive movement from early 1997, when there was reluctance and skepticism about contracting with the government, to 2002, when NGOs were clamoring for partici- 42 World Bank Working Paper pation, reflects the maturation of both Table 1.8. Techniques for Removing the market for services as well as the gov- Obstacles to Competition ernment as purchaser. Specific activities Obstacle Techniques for reducing obstacle implemented in late 1999 and 2000 that Supplier Use of short-term contracts promoted long-term competition were a availability Advertise widely largely open, transparent, and competi- Open up the contracting process tive bidding process for new contracts; a Reduce start-up costs contract renewal process than resulted Encourage new firms in contract cancellations and market exit; Pay bills on time ongoing communication between the gov- Transitional Retain partial government service ernment (at the district level) and the con- difficulties Use multiple contractors tractors; and the creation of performance Evaluate each contract standards that are clearly linked to contract Service Require rebidding renewal (and program goals). Increasingly, interruption Use short-term contracts these rules are understood by all. Open up the contracting process Themarkethasbecomecompetitive-- or at least contestable--in many areas. By Evaluate each contract 2001, two or more NGOs were compet- Political Remove exclusionary practices issues ing for nearly all new contracts, even in Use open processes remote areas. As previously mentioned, Evaluate each contract in some close-in areas, five or more NGOs Formulate legislative policy may compete for a contract. The market Inertia Use short-term contracts is also consolidating. NGOs that already Open up the contracting process have at least one contract are bidding for Evaluate each contract and winning additional contracts, rather Source: Adapted from Marlin (1984, p. 102, than new market entrants. For example, exhibit 11.1). the number of NGO contracts increased by 24 (18 percent) between 2000 and 2002, but the number of NGOs involved in the program decreased by 1. Over this same period, 66 contracts were conditionally renewed. These NGOs were given one year to achieve uncondi- tional renewal or risk cancellation. The market will likely consolidate even more over the short and medium terms, espe- cially if the MSPAS does not adjust the per capita payment. Without an upward adjustment in the capitation payment, only relatively large NGOs with additional, usually external, sources of financing to cover the alleged cost gap would remain. Very limited information exists on whether the MSPAS developed contingency plans for the possibility of a large number of NGOs leaving the market. Preventing contractor fatigue is another issue that has yet to be addressed. Another issue relates to the incomplete regulatory framework govern- ing the convenios. Until the government promulgates regulations governing the use of con- venios and the selection of providers, continuous legal challenges loom. The MSPAS has yet to consider expanding the scale in terms of population coverage per contract, or opening the process to price competition. The current modus operandi-- a large number of single contracts, each covering an estimated 10,000 people with a fixed per capita payment--was in part an artifact of the original pilot design. Given the lack of information on NGO capacity, poor access to rural communities, and the general distrust between the MSPAS and NGOs, the MSPAS (in 1996­97) considered 10,000 people "man- Health System Innovations in Central America 43 ageable" for most NGOs. The catchment area of 10,000 residents remains an arbitrary benchmark. Although by 2002, most providers already served significantly more people,35 the MSPAS prefers to award multiple contracts to a single provider rather than increase contract scale. For example, in 2002, several NGOs had been awarded three or four con- tracts. Increasing scale is a politically sensitive issue. MSPAS officials express concern that contracts covering a larger number of people would favor larger NGOs with international affiliates and would be detrimental to cultivating a national market. Finally, MSPAS offi- cials suggest that most NGOs would reject price competition and claim that few have infor- mation on costs to develop a price proposal. However, as mentioned above, NGOs are already competing for new contracts, albeit for an MSPAS-fixed price. Apparently, MSPAS authorities have been reluctant to consider price competition, fearing that it may result in significantly higher capitation rates that the MSPAS can ill afford to finance. Conclusion and Lessons Learned Contracting goes beyond a simple relationship between a purchaser (in this case, a gov- ernment agency) and a provider. In theory, if done strategically as a means of improving the quality, access, or both of services to the population, it can ultimately transform the market for services, creating greater efficiency and rationality in the distribution of resources and services. Where a sophisticated market is not present, legal and regulatory frameworks are ambiguous, and contract management and monitoring capacity are nonexistent, accomplishing these goals will take time. These were the conditions under which Guatemala boldly launched a large-scale contracting-out effort. More specifically, when the MSPAS began its contracting program for basic services in rural areas, it faced a plethora of obstacles. First, contracting for health services was an untried process in a system that previously did only direct provision of care. Second, much of the regulatory framework to support this type of contracting was absent, and a new model, the convenio, was created to offer a contract that was timelier and better adapted to the needs of both the purchaser and the contractor. Third, a large number of health NGOs already existed in Guatemala, and many worked in rural areas, but the MSPAS was un- familiar with the market, particularly their willingness and capacity to supply the health services specified by the MSPAS. Fourth, there was resistance from MSPAS staff and health worker unions who favored inhouse provision. Fifth, there was estrangement between the government and the NGOs, who were initially suspicious of working with the government. Finally, there was significant distrust of government intentions from the indigenous people whom the program largely served. Recognizing these unfavorable conditions, the original design called for a small pilot project. Scaling up would occur during a subsequent phase based on the results of a planned evaluation. However, what began as a small-scale contracting initiative in 1997, with three departments and seven NGOs, rapidly ballooned into a large-scale contracting effort in 1998--before learning, program modifications, and other early course corrections could be incorporated. With elections due in late 1999, there was little time or concern for 35. For example, a provider survey found that PSSs served an average of 19,400 people (N = 21) in three departments (provinces), and the ADMSSs served an average of 12,500 (N = 12) (see Danel and La Forgia in this volume). 44 World Bank Working Paper following best practices in contracting out health services. Driven by political necessities related to compliance with the Peace Accords and their implications for the political cycle, an often ad hoc process was initiated in which many shortcuts were taken to rapidly implement the large-scale effort. Nevertheless, a case can be made that if the MSPAS had followed a more stringent and methodological process of first establishing all the institu- tional conditions supporting the purchasing function and setting up and evaluating small demonstration projects, as originally planned in the PMSS pilot, the change of government in 2000 might have thwarted the scaling-up process. In the wisdom of hindsight, by 1999, the contracting-out experience had created sufficient momentum, scope, and buy-in by NGOs and donors that the newly elected government (in 2000) had little choice but to sup- port the program.36 Contracting has yielded benefits even though capacity was weak and only under devel- opment. First, a basic package of services for women, infants, and children was developed, based on epidemiological information and targeted particularly toward the poor, rural population. This was in itself a major accomplishment. Second, as discussed in the com- panion chapter in this volume, services were extended to unserved and underserved pop- ulations, though not to the full extent anticipated. Third, an initially unsophisticated system of contracting and provider selection (for example, the MSPAS's major and direct involvement in selection of the initial seven NGOs for the pilot) evolved into a more trans- parent, performance-oriented, and decentralized system of decision making for new and existing contracts that appears to have been largely successful. Finally, the program demon- strated that at a tumultuous time in the country's political and social history, a public- private partnership could be forged that would address serious unmet needs for many poor Guatemalans. As reported in the companion chapter in this volume, the contract-based models per- formed as well as--and often better than--traditional providers (health posts) in terms of such performance measures as population coverage, patient satisfaction, production, pro- ductivity, and coverage for immunizations, prenatal care, and use of ORS. However, the contracting models, particularly direct contracting, displayed significantly higher unit costs. How the process of contracting described here affected performance is a difficult call. In general, the design was successful in meeting program objectives: basic care coverage was extended in a relatively short period (four years) to a large number of distant and dis- persed communities that previously had no or little access to care. However, production and coverage rates for priority PBS interventions such as immunizations, prenatal care, growth monitoring, ARI care, and use of ORS in the home were lower than anticipated for most NGO (and traditional) providers (but probably higher than before the program's onset). Providers were inconsistent in delivering the full package of PBS interventions, often emphasizing vaccinations while failing to regularly provide other priority PBS inter- ventions. Greater specification of these priority services in the contract, together with per- formance indicators as well as more vigorous monitoring by the MSPAS, could have contributed to standardization of services according to MSPAS priorities as well as higher coverage rates for these services. Also, some NGOs claimed that the per capita payments were insufficient to cover the full costs of the PBS coverage, which led them to restrict ser- 36. The new government originally criticized the PMSS, suggesting that it would withdraw support. Health System Innovations in Central America 45 vice. Failure to conduct a baseline survey to gather health and utilization data handicapped the capacity of the MSPAS to monitor, evaluate, and compare provider performance. Turning to the elements of the contracting framework (see box I.1), two distinct phases can be discerned: (1) a developmental, trial-and-error phase between from 1997 through 1999 and (2) a phase of correction and consolidation phase between from 2000 through 2002.37 Table 1.9 presents a scorecard rating MSPAS performance on each of the steps of the framework. Table 1.9. Rating of Guatemalan Contracting Experience According to the Conceptual Framework Steps in the contracting framework 1997­99 2000­02 1. Assess feasibility Moderate N/A 2. Gain political and institutional support Moderate High 3. Define service specifications Low Moderate 4. Select performance measures Low High 5. Define payment methods Moderate Moderate 6. Select providers, maximize competition Low Moderate 7. Negotiate and write the contract High High 8. Monitor and evaluate the contract Low Moderate 9. Encourage competition Low Moderate N/A not available. The MSPAS receives good marks for garnering political and institutional support for the contracting effort and mitigating the effects of health workers' resistance. It also quickly recognized and filled the gaps in the legal and regulatory framework, preparing and ush- ering through Congress a new Health Law that permitted the contracting out of services. The development of robust financial payment, management, and oversight systems to enable the government to budget funds, make payments, and track financial flows was another factor enabling implementation and expansion. There has been no evidence of corruption or diversion of funds by the NGOs or the MSPAS. The major challenge in the Guatemalan experience was related to the rapid move from small to large scale. The expansion stressed the still developing contracting capabilities of SIAS and required prompt and, in some cases, improvisational response. It limited the abil- ity to institutionalize the contracting process or allow for adequate learning by NGOs, local districts, and the contracting arm itself. During the early phase of the PMSS, a number of shortcomings became evident in the fledgling contracting process. Before the decision to scale up, little work was done in assess- ing the NGO market or consulting with them on program design. Although the MSPAS developed and costed out the PBS, the costing model was based on MSPAS rather than NGO costs. The "one size fits all" package of services was probably inappropriate for the 37. This second phase had more to do with the maturation of the contracting process than with the change of government. 46 World Bank Working Paper diverse health conditions evident in Guatemala, and it did not account for the many years of experience of a number of health NGOs in rural areas. The lack of clear performance indicators linked to program goals resulted in confusion over what was expected and con- tributed to problems of PSSs and ADMSSs providing the full range of services. There was no baseline against which to evaluate the performance of PSSs and ADMSSs. The non- transparent selection of NGOs created early doubts about the intentions of the government among the NGO community and the noncompetitive, open contracting process resulted in the selection of several poorly qualified NGOs. Early efforts to monitor compliance focused on micromanaging process efforts in the organization and delivery of care that often had little to do with the outputs or results desired. By 1999, ambition was giving way to reality. MSPAS staff realized that many mis- takes were being made as a result of the ad hoc nature of decision making and dealing with problems. The PMSS started to put in place a number of changes that improved the contracting process in several critical areas: a shift to an open, transparent, and compet- itive selection process; the development of meaningful performance indicators (linked to desired epidemiological outcomes) that become points of accountability in evaluat- ing new contracts or renewal of existing contracts; development of the contract man- agement function inside the MSPAS; and systematic performance monitoring. Although the PMSS was originally implemented without competition, competition was gradually brought into contacting process. Shortcomings remain. Failure to make adjustments in the per capita payment has already resulted in market exit of two qualified NGOs. In-depth analysis of NGO costs is needed to create a standardized cost structure that is more in line with actual costs. Sys- tematic evaluation of performance has yet to be built into the program. The PBS requires assessment to determine the effectiveness of the defined interventions. The capacity to learn and to adapt is one of the important lessons from the Guatemalan experience. Failing to abide by all best contracting practices is not a determinant of failure. The ability to learn from the mistakes and build upon strengths appears to be the key deter- minant of success. Bibliography Abramson, Wendy. 1999. "Partnerships between the Public Sector and Non-governmental Organizations: Contracting for Primary Health Care Services." Abt Associates Inc., Bethesda, Md., financed by the U.S. Agency for International Development. Annis, Sheldon. 1981. "Physical Access and Utilization of Health Services in Rural Guatemala." Social Science and Medicine 15(4):515­23. Bennett, Sara. 1997. "Private Health Care and Public Policy Objectives." In C. Colclough, ed., Marketizing Education and Health in Developing Countries: Miracle or Mirage? Oxford: Clarenden Press. Bennett, Sara, and Anne Mills. 1998. "Government Capacity to Contract: Health Sector Experience and Lessons." Public Administrative and Development 18:307­26. Brushan, Indu, Sherryl Keller, and Brad Schwartz. 2002. "Achieving the Twin Objectives of Efficiency and Equity: Contracting Health Services in Cambodia." ERD Policy Brief 6. Asian Development Bank, Economic and Research Department, Manila. Health System Innovations in Central America 47 Cercone, James, and Magdalene Rosenmöller. 2000. "Costa Rica--a New Hospital for Heredia--a Public-Private Partnership for Health Care." In Magdalene Rosenmöller, ed., Challenges of Health Reform. Barcelona: IESE Business School, University of Navarra. Cominsky, Sheila. 1987. "Women and Health Care on a Guatemalan Plantation." Social Science and Medicine. 25(10):1163­73. Deman, Hedi. 2000. "Informe final de la coordinadora del subprograma III: Ampliación y reorientación de los servicios de salud programa mejoramiento." MSPAS, Guatemala City. Domberger, Simon. 1998. The Contracting Organization--a Strategic Guide to Outsourc- ing. New York: Oxford University Press. Eichler, R., P. Auxila, and J. Pollock. 2001. "Performance-Based Payment to Improve the Impact of Health Services: Evidence from Haiti." The Manager 10(2). Available at http://www1.worldbank.org/hnp/hsd/Contracting.asp (accessed November 2004). Gestión y Tecnología en Salud y Desarrollo. (GETSA), S.A. 2001. Estudio de caso: El pro- ceso de contratación a gran escala para la prestación de servicios básicos de salud: La expe- riencia de Guatemala. Consultant's report, Guatemala.City. Gilson, Lucy, Priti Dave Sen, Shirin Mohammed, and Phare Mujinja. 1994. "The Potential of Health Sector Non-governmental Organizations: Policy Options." Health Policy and Planning 9(1):14­24. Goldman, Noreen, and Anne Pebley. 1994. "Childhood Immunization and Pregnancy- Related Services in Guatemala." Health Transition Review 4(1):29­44. Gwatkin, Davidson R., Shea Rustein, Kiersten Johnson, Rohini P. Pande, and Adam Wagstaff. 2000. "Socio-Economic Differences in Health, Nutrition, and Population in Guatemala." HNP-Poverty Thematic Group of the World Bank. Washington, D.C. Processed. Instituto Nacional de Estadística (INE). 1999. Guatemala--Encuesta Nacional de Salud Materno Infantil 1998/99. Guatemala City. Loevinsohn, Benjamin. 2002. "Contracting for the Delivery of Primary Care in Cambodia: Design and Initial Experience from a Pilot Test." World Bank, Washington, D.C. Available at http://www1.worldbank.org/hnp/hsd/Contracting.asp (accessed November 2004). Marek, Tonia, Issakha Diallo, Biram Ndiaye, and Jean Rakotosalama. 1999. "Successful Contracting of Prevention Services: Fighting Malnutrition in Senegal and Madagas- car." Health Policy and Planning. World Bank, Washington, D.C. Marlin, John Tepper. 1984. Contracting Municipal Services--a Guide for Purchase from the Private Sector. New York: John Wiley & Sons. McPake, Barbara, and Elias E. Nglande Banda. 1994. "Contracting out of Health Services in Developing Countries." Health Policy and Planning 9(1):25­30. Mills, Anne. 1998. "To Contract or Not to Contract? Issues for Low- and Middle-Income Countries." Health Policy and Planning 13(1):32­40. Mills, Anne, and Jonathan Broomberg. 1998. "Experiences of Contracting: An Overview of the Literature." Macroeconomics, Health and Development Series, Technical Paper 33. World Health Organization, Geneva. Mills, Anne, Natasha Palmer, Lucy Gilson, Di McIntyre, Helen Schneider, Edina Sinanovic, and Haroon Wadee. 2004. "The Performance of Different Models of Primary Care Provision in Southern Africa." Social Science and Medicine 99:931­43. 48 World Bank Working Paper Mintz, Patricia, Gerard Martin La Forgia, and William Savedoff. 2001. "Contracting Health Services: Getting from Here to There." World Bank, Washington, D.C. MSPAS. 1998. Guia para la formulación de propuestas para la prestación y administración de servicios básicos de salud. Guatemala.City. ------. 2000. Habilitación, adjudicación, certificación y acreditación. Guatemala City. Nieves, Isabel, and Gerard Martin La Forgia. 2000. "Guatemala--Large-Scale Government Contracting of NGOs to Extend Basic Health Services to Poor Populations in Guatemala--a Public-Private Partnership for Health Care." In Magdalene Rosen- möller, ed., Challenges of Health Reform. Barcelona: IESE Business School, University of Navarra. Pan American Health Organization. 1999. Epidemiological Bulletin--Guatemala City. 12(3). Available at http://www.paho.org/English/SHA/prflgut.htm (accessed October 2003). Pebley, Anne R., and Noreen Goldman. 1992. "Family, Community, Ethnic Identity and the Use of Formal Health Care Services in Guatemala." Office of Population Research Working Paper 92-12. World Bank, Washington, D.C. Rosen, James E. 2000. "Contracting for Reproductive Health Care--a Guide." HNP- Population and Reproductive Health Working Group, World Bank, Washington, D.C. Shulman, Martha A. 1982. "Alternative Approaches for Delivering Public Services." Urban Data Reports. International City Management Association, Washington, D.C. Souters, R., and F. Griffiths. 2003. "Improving Government Health Services through Con- tract Management: A Case from Cambodia." Health Policy and Planning 18(1):74­83. Taylor, Robert F. 2003. "Contracting for Health Services." In Private Participation in Health Services. Washington, D.C.: World Bank. Vining, Aidan, and Steven Globerman. 1999. "Contracting-out Health Care Services: A Conceptual Framework." Health Policy and Planning 46:77­96. World Bank. 1994. "Brazil: The Organization, Delivery, and Financing of Health Care in Brazil: Agenda for the 90s." Report 12655-BR. World Bank, Washington, D.C. ------. 2003. "Making Services Work for Poor People." In World Development Report 2004. Washington, D.C.: World Bank. CHAPTER 2 Contracting for Basic Health Care in Rural Guatemala-- Comparison of the Performance of Three Delivery Models38 Isabella Danel and Gerard La Forgia Abstract Beginning in 1997, the government of Guatemala extended coverage of a pack- age of basic health services to 3 million people living in rural, impoverished, and primarily indigenous communities through three different delivery models: con- tracting of nongovernmental organizations (NGOs) to directly provide services (direct model), contracting of NGOs to act as financial managers for the Minis- terio de Salud y Assistencia (Ministry of Health and Social Protection [MSPAS]) health service providers (mixed model), and strengthening of health posts operated by the MSPAS (traditional model). This chapter presents the results of household and provider surveys carried out in 2001. The household survey found that women and children in catchment areas served by the mixed model had significantly better results for many key health indicators when compared to the traditional model. Results for those served by the direct model were sim- ilar to the traditional model although these communities had only recently begun to receive services and were located much farther from health facilities. Users tended to report greater satisfaction with the new models. The provider survey found that the NGO-based models are generally more productive than the tradi- tional model; however, they are more costly, and results on economic efficiency were mixed. We conclude that mixed and direct provider models are good alter- natives to traditional health services for providing basic services in Guatemala. Contracting basic health care may be an effective strategy to reduce health inequities in other country contexts where access to health services is inadequate. 38. This chapter is based on field research conducted in 2001. It draws on field research and statistical analyses commissioned to GETSA (Gestión y Tecnología en Salud y Desarrollo), S.A. A companion chapter by La Forgia, Mintz, and Cerezo in this volume examines the contracting-out process in Guatemala. 49 50 World Bank Working Paper Introduction Governments in developing countries are seeking ways to deliver health services effectively and efficiently to populations whose current access to health care is inadequate. There is a growing body of experience in developing countries with contracting a variety of health ser- vices through nongovernmental organizations (NGOs) that suggests it may be an alternative for improving health outcomes (Bhushan, Keller, and Schwartz 2002; Eichler, Auxila, and Pollock 2001; Marek and others 1999; Palmer 2000; Slack and Savedoff 2001). NGOs have been contracted to deliver an array of health services including nutrition services, HIV/AIDS services, and primary health care. However, evidence showing improved performance of these alternative models in delivering primary health care services is still weak. Only a lim- ited number of studies have examined the impact and performance of contracting NGOs to deliver primary health care. Studies from Cambodia and Haiti demonstrated improvements in some key health indicators particularly immunization rates. (Bhushan, Keller, and Schwartz 2002; Eichler, Auxila, and Pollock 2001). Studies comparing NGOs and public providers of primary care show that the former are generally more productive and the results on efficiency are mixed (Bitrán 1995; Levin and others 1999; Lavadenz 2001; Bhushan, Keller, and Schwartz 2002). In 1997, following the signing of Peace Accords ending decades of civil war, the govern- ment of Guatemala considered options for rapidly scaling up health services in rural, impov- erished, primarily ethnic communities. It decided that contracting NGOs to deliver primary health care was the mechanism most likely to succeed on the scale and timeframe required. This chapter presents findings from two surveys that assessed the success of that effort by comparing the performance of the new and traditional models for delivering basic health ser- vices, as well as the allocative and technical efficiency of these models. The chapter presents background on the three models of care and some of the ques- tions leading to the study that were raised after the contracting process had been under way for a number of years. It follows with the objectives of the two surveys carried out, their methods, and their findings. It concludes with a discussion of the findings and their policy implications. Background In 1997, the government of Guatemala (GOG) embarked on an ambitious effort to extend basic health coverage to impoverished rural and indigenous populations through the con- tracting of private, nongovernmental organizations (NGOs). By 2002, the GOG had crafted 161 convenios (agreements) with 88 NGOs to reach more than 3 million beneficiaries. Before this effort, the GOG (through the health ministry--the Ministerio de Salud y Assistencia (Ministry of Health and Social Protection [MSPAS]) had provided health services only through inhouse providers and had not entered into service contracts of any kind with pri- vate providers.39 The coverage extension effort was part of a broader health reform policy that sought to modernize and decentralize the MSPAS, improve the efficiency and quality of health services, strengthen management and governance functions of the MSPAS, and 39. The contracting process is analyzed in chapter I in this volume. Health System Innovations in Central America 51 increase financial resources of the sector. This set of initiatives was known as the Integrated Health Care System (Sistema Integrado de Atención en Salud [SIAS]). The program's main objective was to extend a paquete básico de servicios (package of basic services [PBS]) to rural, indigenous populations with heretofore little or no access to health services, giving priority to prevention, maternal and child care, and basic cura- tive care (see Box 2.1). Contract- ing was introduced as a means to rapidly extend services while Box 2.1. Basic Health Care Package making the service delivery system more responsive to clients and the Maternal care MSPAS.40 1. Prenatal care The design involved three 2. Delivery care financial and organizational 3. Tetanus toxoid arrangements to provide the PBS. 4. Iron and folate supplementation during pregnancy These are schematically presented 5. Postpartum care in Figure 2.1 and the main features 6. Child spacing: education and referral are summarized in Table 2.1. The first two arrangements, direct and Infant and child care mixed contracting of NGOs, were 7. Immunizations introduced through the SIAS 8. Care of acute respiratory infections program. The third arrangement, 9. Care of diarrhea, cholera MSPAS health post (HP)-based 10. Prevention, treatment of nutritional deficiencies delivery, already existed. Direct 11. Growth and development monitoring of children contracting entailed the contract- younger than two years ing of NGOs, as PSSs (proveedora Care of illnesses and emergencies de servicios de salud [health ser- 12. Cholera vices provider]), to provide the 13. Dengue PBS directly through outreach ser- 14. Malaria vices to predetermined geographic 15. Tuberculosis catchment areas (jurisdicciones) 16. Rabies containing about 10,000 inhabi- 17. Sexually transmitted diseases, HIV/AIDS tants. The PSSs received a capitated 18. Emergencies (fractures, burns, and so on) payment and were responsible for the purchase of all inputs (except Environmental health for vaccines, which were sup- 19. Vector control plied by the MSPAS). The second 20. Zoonosis control arrangement, a mixed contracting 21. Sanitation model, involved the contracting of 22. Water quality surveillance NGOs as health services admin- 23. Food hygiene istrators (administradora de ser- 24. Improved household sanitary conditions. vicios de salud [ADMSSs]), which 40. The Peace Accords of 1996, which ended decades of civil strife, were a main driver of the cover- age extension effort. The accords contained provisions for the improvement of health in the country, par- ticularly in underserved rural areas. 52 World Bank Working Paper Figure 2.1. Delivery Models: Direct, Mixed, and Traditional B a s i c P S S s e r v i c e s B P e r c a p i t a E v a c c i n e s B a s i c N A D M S S E P e r c a p i t a s e r v i c e s F D r u g s P e r s o n n e l I M S P A S v e h i c l e s s u p p l i e s C I V a c c i n e s M S P A S A p e r s o n n e l t e a m s R I E P e r s o n n e l B a s i c S s u p p l i e s M S P A S d r u g s H P s s e r v i c e s v e h i c l e s operated exclusively as financial managers for services delivered by MSPAS teams. They also received a capitated payment. The ADMSSs hired and paid additional personnel, mainly physicians and administrative staff, to complement MSPAS staff, purchased supplies, and managed the books for the MSPAS. ADMSSs did not supervise or provide technical over- sight; this was the responsibility of MSPAS area offices.41 Both models were designed to enable MSPAS teams to expand coverage rapidly by facilitating the acquisition of inputs through bypassing the notoriously slow and cum- bersome government personnel hiring and materials procurement systems. About 60 per- cent of the convenios have been awarded to NGOs that operate as ADMSSs. Several NGOs received contracts as both ADMSSs and PSSs. Some NGOs hold both ADMSS and PSS contracts. Thethirdarrangementconsistedofthetraditionalmodelofpubliclyfinancedservicedeliv- ery in rural areas. Under this arrangement, personnel assigned to MSPAS HPs were responsi- ble for delivering the PBS within the facilities' catchment area. Similar to other facilities and services operated directly by the MSPAS, the HPs were financed through budgetary allocations managed at the area level. For the most part, PSS and ADMSS providers were assigned catchment areas where there waslittleornoMSPASpresenceandwherecommunitieshadnoregularaccesstohealthcare.42 By design, PSSs were generally assigned catchment areas in remote regions with deficient 41. The mixed model resulted from a political compromise between MSPAS authorities and local health workers. Originally, the MSPAS was interested only in direct contracting. However, resistance from area and district MSPAS personnel threatened the viability of this proposal. The mixed model emerged as a negotiated solution to secure their support for direct contracting. 42. The direct and mixed models were seen as more rapid means to extend services than the tradi- tional MSPAS. The MSPAS was constrained by onerous hiring practices, weak logistical systems, and anti- quated and protracted procurement practices. Further, recruiting and maintaining professional personnel to provide care in remote areas had proven difficult. Health System Innovations in Central America 53 Table 2.1. Features of SIAS Delivery Models Features Direct PSS-NGO Mixed ADMSS Traditional MSPAS HP Assigned Mainly remote and dis Usually communities Communities closer to catchment area persed rural hamlets closer to main road or main road or market market towns towns Input All inputs except All inputs except vaccines All inputs management vaccines and some personnel Delivery PSS-NGO Mixed teams consisting of MSPAS HP personnel responsibility MSPAS and ADMSS-hired personnel Team Physician Physician Auxiliary nurse composition Institutional facilitator Institutional facilitator Volunteer Cuban Community facilitatora Community facilitatora physician (if available) Community health pro Community health pro- Community health moters and TBAsa moters and TBAsa promoters and TBAsa Place of Outreach community Outreach community HPs delivery centers centers Periodicity of Monthly visits to com Monthly visits to commu- Daily, Monday through service delivery munity centers nity centers Friday Financing Capitation payments Capitation payments Budgetary allocations mechanisms In kind (vaccines) Budgetary allocations In kind (vaccines) (MSPAS personnel) In kind (vaccines) Number of 88 NGOs, 161 signed conveniosb 1,289c NGO providers (national, 2002 a. Volunteer workers. b. A number of NGOs had multiple contracts; 60 percent of the convenios were for PSSs; and some NGOs possessed both PSS and ADMSS contracts. c. Not all participated in SIAS. road networks and where most communities consisted of dispersed hamlets, and ADMSSs' catchment areas tended to be closer to roads and market towns. Before the onset of SIAS, HPs existed throughout the country and usually were located in concentrated rural com- munities near market towns and main roads. Inability to recruit and maintain physicians to provide services in rural areas led the GOG to enter into an agreement with Cuba in 1998 in which the latter sent approximately 300 volunteer physicians to staff rural HPs and, to a lesser extent, serve on itinerant teams operated by ADMSS and PSS providers that were unable to recruit Guatemalan physicians. The MSPAS crafted contract-like convenios (agreements) to engage the NGOs under the direct and mixed models. The convenio defined the services to be covered, the popula- tion ( jurisdicción) to be served, and the financial amount of the contract. The contract also included sections addressing payment terms and conditions of payment. MSPAS norms, annexed to the convenios, specified the team composition and the locale of PBS provision. In theory, PSS and ADMSS teams had to consist of a physician (or regis- tered nurse) and an institutional facilitator (rural health technician or nurse) who were 54 World Bank Working Paper responsible for providing the PBS.43 Their teams also contained up to eight community facilitators--volunteer workers who lived in the communities, received a small stipend, and assisted the institutional facilitator and medical teams, but were not responsible for service provision. The community facilitator played a key liaison role between the outreach teams and the communities, apprising the community of the pending arrival of the outreach teams and organizing the work of other volunteer health workers (see next paragraph)). The work of the community facilitators, combined with local knowledge of the presence of a physi- cian on the outreach teams, stimulated demand for PSS and ADMSS services. The PSSs and ADMSSs were also responsible for recruiting and paying a small stipend to community-based traditional birth attendants (TBAs) and community health promot- ers, as well as training all personnel. Providers also were to establish one or more commu- nity centers where itinerant teams were to provide services on predetermined dates. These centers were existing structures made available by community groups44 and located within one hour's walk from the communities they served. Teams were expected to provide care at the centers at least once a month, and volunteer health workers were responsible for mak- ing monthly household visits. Nurse auxiliaries and, when available, volunteer Cuban physi- cians were responsible for providing the PBS at MSPAS HPs. As discussed in the next section, HPs represented a facility-based delivery model, compared to the outreach-based model of the PSSs and ADMSSs. The NGOs received capitated payments in four yearly installments. The amounts received varied depending on the type of contract (ADMSS or PSS), number of inhabitants in their assigned geographic area, and distance considerations. ADMSSs received a lower cap- itation rate than the PSSs for two reasons. First, as described earlier, an ADMSS contracted with a reduced number of personal compared to a PSS. Second, because most PSSs were assigned to remote areas, they received higher rates to account for higher transportation costs. The capitation payment covered the direct cost of the basic package plus administrative expenses, as well as expenditures related to institutional strengthening. In 2001, the average per capita payment was approximately US$6.25. The total per capita cost was about US$8.00, including the cost of vaccines that the Ministry of Health supplied directly to all providers.45 In 2001, the MSPAS spent US$14.2 million on capitated transfers to ADMSSs and PSSs. However, we estimate that the MSPAS spent an additional US$7.1 million on SIAS for administration, in-kind transfers (vaccines), supervision, HPs, and training of MSPAS per- sonnel. Taken together, in that same year, the MSPAS directed about 13 percent of its total budget to coverage extension through SIAS. PSS and ADMSS providers were financed through capitation payments and in-kind payments for vaccines. However, consonant with the nature of the mixed model, ADMSSs benefited from budgetary allocations that paid for MSPAS personnel assigned to their provider teams. HPs were funded through line-item budgetary allocations managed by area administrative units. 43. Rural health technicians were specially trained health workers who provided preventive activities such as well-baby checkups, vaccinations, and community health promotion. If physicians were unavail- able, NGOs hired registered nurses. 44. In some cases, a building, usually a house, was rented, but generally a house was provided free of charge. No facilities were constructed. 45. The MSPAS also purchased and supplied drugs for the treatment of tuberculosis and malaria for NGOs demonstrating need in their catchment areas. However, some NGOs reported purchasing these drugs directly from suppliers when the MSPAS was unable to supply them in a timely manner. Health System Innovations in Central America 55 The MSPAS developed and expanded the program by trial -and error. In addition to designing the technical content of the extension program, the MSPAS had to create the man- agerial capacity to select providers, specify contractual provisions, negotiate and make con- tracts, issue payments, and monitor performance. These functions were first performed at the central level but gradually transferred to area offices. The MSPAS also had to negotiate the creation of a budgetary line item with the Ministry of Finance, as well as procedures and processes to guide the transfer of funds to the providers. There were modifications in pro- gram and contracts management between 1997 and 2002.46 Table 2.2 shows the growth of contracts and beneficiaries. The original design called for a three-year pilot period in which a limited number of contracts would be issued. Pressed by the need to comply with the Peace Accords and the government's mandate to show results before the next round of elections (in 1999), the decision was made to rapidly scale up the coverage extension effort. Public spending on direct transfers to NGOs increased rapidly, from US$8.0 million in 1998 to US$15.6 million in 2002. The population that NGOs were contracted to serve reached a total of 3.2 million in 2002; all were rural residents, and most of them were from impoverished indigenous groups. Table 2.2 Summary Statistics on SIAS Implementation Agreements, NGOs, and Estimated Coverage, 1997­2002 No. of signed No. of NGOs Estimated Year agreements under contract population covered 1997 14 7 460,000 1998 110 74 2,180,000 1999 119 78 2,516,000 2000 136 89 2,940,000 2001 154 89 3,000,000 2002 161 88 3,200,000 Source: MPSAS/SIAS annual reports, various years. Issues In 2001, as a result of the lack of quantifiable information and the short implementation period, program performance had yet to be assessed. No baseline data were collected at pro- gram onset. In 2001, the MSPAS began collecting data on NGO performance through an analysis of NGO registers, postconsultation interviews with patients, and community focus groups in part to establish baseline data to facilitate monitoring and evaluation. However, no information existed on the comparative performance of the three delivery models. This study aimed to complement the MSPAS's efforts and methodologies by obtaining robust information on the performance of NGO and MSPAS providers as a means to strengthen the quality, efficiency, and coverage of basic health services as well as consolidate the development 46. Contracting procedures and processes are described in detail in chapter I in this volume. 56 World Bank Working Paper of MSPAS institutional capacity to manage contracts, monitor performance, and evaluate results. According to unpublished MSPAS documents (MSPAS 2000a, 2000b), the program's main success was in increased coverage of basic health services, especially immunization coverage. For example, they cite that the national child immunization coverage rates for measles for children 12 to 23 months of age increased from 69 percent in 1996 to 83 percent in 1998.47 Anecdotal evidence suggests that immunization coverage increased significantly in remote regions. However, it is unknown whether these increases can be attributed solely to the program or whether the MSPAS-NGO interface through direct and mixed delivery models is cost effective for providing PBS coverage on a regular basis. Others claimed that the PBS contracted out to NGOs has been only partially implemented. Further, there is no data comparing the performance of the direct and mixed delivery models with the tradi- tional MSPAS system. Questions have been raised about the PBS. The MSPAS considered the PBS non- negotiable. However, supplementary services could be provided at the discretion of the NGOs if alternative financing existed. The PBS was developed without local input from MSPAS line personnel or the NGOs. The latter criticized SIAS for forcing a "one size fits all" care model, claiming that the PBS should include frequently seen pathologies such as skin infections. Also, MSPAS norms allowed providers little flexibility in determining how best to deliver the PBS. All providers had to adopt an MSPAS-mandated personnel structure, train- ing program, materials, and equipment. Some NGOs considered the norms a "straitjacket model" that was inappropriate for some situations. Most NGOs relied solely on MSPAS capitation payments.48 Charging user fees was not permitted, nor did our research uncover any evidence of such charges. As of 2002, the per capita rate applied in 1997 had not been adjusted for inflation, estimated at nearly 30 per- cent over this period. The failure to adjust the payments contributed to five NGOs with- drawing from SIAS in 2000 and 2001. Others coped through reducing their range of activities. Interviews with NGO providers suggest that most provided a subset of PBS- mandated interventions.49 Important differences were observed among the provider models in terms of the team composition, population served, and catchment area location. These contributed to differ- ences in coverage, production, productivity, and costs described elsewhere in this chapter. First, as described previously, by design most PSSs were assigned catchment areas in distant regions where the road network was wanting. Using census data and maps, local MSPAS area and district officials and central-level SIAS personnel delineated geographic areas where the MSPAS had little or no presence. The areas farthest from a health center (usually located in a market town)50 and possessing a poor road network were generally assigned to PSSs. Anec- dotal evidence suggests that the distance between villages of PSS catchment areas was much 47. National Health Survey data show that measles vaccine coverage for children aged 12­23 months increased from 75 percent in 1995 to 81 percent in 1998­99. Full vaccine coverage increased from 43 per- cent to 60 percent (ENSMI [Encuesta Nacional de Salud Materno-Infantil] 1998­99). 48. Anecdotal evidence suggests that a minority of NGOs had access to additional sources, but it is unclear whether such funds were used to complement SIAS-financed services. 49. Production data collected from the NGOs appear to confirm this statement. 50. This was confirmed in the household survey (see Table 2.5). Health System Innovations in Central America 57 greater than for the other providers. Conversely, ADMSSs were usually assigned catchment areas closer to towns or near main roads. Also, most HPs were already located on main roads. According to SIAS officials, assigning catchment areas to ADMSSs that were closer to towns was meant to accommodate MSPAS personnel who resided in these towns and were assigned to ADMSS provider teams.51 Also, SIAS officials considered that the PSS model would be more robust to expand services to areas of difficult access. Second, the average nominal population assigned to the PSSs was larger than those served by ADMSSs and HPs. For example, in the provider sample, PSSs (N = 21) served 19,400 people on average, compared to 12,500 (N = 12) per ADMSS and 9,500 (N = 22) per HP. Whether the magnitude of the differences observed in the sample held for providers else- where is unknown. Third, unlike the outreach orientation of the direct and mixed providers, the HPs remained a facility-based delivery model. This resulted in part from catchment area residents' relatively easy access to the HPs, which were located on main roads and within or near more concentrated communities. SIAS used the existing HP network, and no new facilities were constructed.52 Table 2.3. Average Number of Staff by Provider Type and Profession Nurse or Institutional Provider type Total Physician nurse auxiliary facilitatora Direct (N = 21) 2.3 1.0 0.1 1.2 Mixed (N = 12) 2.0 1.0 0.4 0.8 Traditional (N = 22) 1.5 0.5 1.0 0 a. Nurse or trained rural health technician. Source: provider sample. Fourth, staffing volume, composition, and hiring responsibility varied among the mod- els (see table 2.3). PSS and ADMSS staffing consisted of three types of health workers: physi- cians, rural health technicians, and community facilitators. In contrast, nurse auxiliaries staffed the HPs. Because the HPs performed few outreach activities, community facilitators were not assigned to them. Nevertheless, HPs benefited from SIAS through the assignment of volunteer Cuban physicians to complement nurse auxiliaries. The MSPAS prioritized SIAS areas for the assignment of the volunteers, but there were many more HPs than avail- able volunteers.53 To a lesser extent, these volunteers were also assigned to PSS and ADMSS 51. Originally, the MSPAS sought to apply the direct PSS model only. However, responding to pres- sure and criticism from MSPAS personnel, the mixed model emerged in which "administrative" NGOs-- ADMSSs--would purchase supplies and hire additional personnel to complement MSPAS personnel. MSPAS personnel assigned to mixed providers, together with area and district authorities, were allowed to identify the jurisdictions for which they were responsible. According to MSPAS personnel, they gener- ally chose uncovered areas with relatively easy access to towns where they resided. 52. Outreach activities were not performed by the nurse auxiliaries who staffed HPs, but by volunteer Cuban physicians who visited communities inside the HP's catchment area on an irregular basis. 53. About half of the HPs in the sample were staffed by volunteer physicians. 58 World Bank Working Paper teams unable to recruit local physicians. HPs were also responsible for training and work- ing with community-based, volunteer health promoters and TBAs.54 PSSs recruited and hired all staff. ADMSSs did the same except for the rural health tech- nicians, who were salaried MSPAS staff with several years of experience in rural health ser- vices. If the catchment areas were found to contain more residents than originally estimated, the MSPAS financed the hiring of additional staff on a case-by-case basis. About half the PSSs in the sample hired additional staff to cover their catchment areas, compared to only 2 of 12 ADMSSs.55 Also, several PSSs reported paying higher wages to attract personnel to work in remote areas. All providers were responsible for identifying and training community workers and TBAs, as well as organizing community groups to support health promotion and disease prevention activities. However, unlike the PSSs and ADMSSs, HPs did not pay stipends to volunteer TBAs and community health promoters because of the near-absence of outreach activities. These workers were also seen as part of an already established MSPAS network of community volunteers linked to the HPs. The MSPAS area units were responsible for managing the ADMSS provider teams and HPs. Together with central-level SIAS, area units were also responsible for supervising all providers. Objectives of the Study The overall objective was twofold: (1) to assess the performance of Guatemala's program to extend coverage of basic health services by comparing key indicators in communities served by three delivery models--NGOs as direct providers, NGOS as resource administrators for MSPAS providers, and traditional MSPAS providers; and (2) to determine the relative eco- nomic efficiency of NGO and MSPAS providers in delivering basic health services. Household Survey Based on a household survey of rural communities served by NGO and MSPAS providers in four departments (Alta Verapaz, Quiché, Totonicapán, and Huehuetenango), comparisons were made of (a) indicators of the provision of priority PBS interventions, including vaccine coverage, prenatal care (PNC), vaccination against tetanus, iron and folate supplementation for pregnant women, knowledge and use of oral rehydration solution (ORS) among women whose children have diarrhea, growth-monitoring checkups, and use of carnets (record cards) for infants; and (b) perceived quality of and satisfaction with care. 54. All three models of the provider sample did not have records of community health agents or TBAs in their catchment areas. Accounting ledgers maintained by the providers lumped together stipend pay- ments for all volunteers, including community facilitators. 55. Recruitment and stability of personnel have been problematic for the PSSs since the program's onset. At the time of the survey, four PSSs were without physicians. They hired registered nurses as sub- stitutes. Health System Innovations in Central America 59 Provider Survey Based on a survey of production inputs, input prices, and outputs of providers serving the communities selected for the household survey, the study compared the efficiency of the three provider models. An indirect measure of allocative efficiency is gauged through com- paring the service mix of each provider and determining to what extent providers are deliv- ering PBS activities, particularly the high-priority interventions related to immunization, acute respiratory illness (ARI) care, diarrhea care, prenatal care, and well-baby checkups. Technical efficiency was estimated by comparing production and labor productivity of the provider models. Economic efficiency was analyzed by comparing unit and per capita costs of providing PBS, non-PBS, and priority interventions. Methods Household Survey A household survey was carried out in four departments that had benefited from large- scale contracting: Alta Verapaz, Quiché, Huehuetenango, and Totonicapán. All of these departments have a high concentration of ethnic communities and the first three in par- ticular are largely rural, with dispersed populations living in communities that are far from health services. Sample size and selection. Sample size was calculated to permit identification of statisti- cal significance with differences in immunization rates for children 12­59 months old of 5 percent or greater. One hundred fifty communities were randomly selected to participate in the survey, 50 from each of the three service-provider-type catchment areas (PSS, ADMSS, MSPAS HPs). Selection criteria for communities included: (a) those with a pop- ulation of fewer than 10,000, (b) those where services had been provided for at least two years, and (c) those without a health center (with a physician permanently on staff). For the traditional model, only communities with an HP or those in an HP's catchment area were included. All eligible communities in the four departments and their populations were listed by service provider type. Fifty communities from each provider type were randomly selected by use of a population proportional to size selection method. Each of the communities was visited during one day. Twenty households were randomly selected in each community, and all women in the household between the ages of 15 and 44 were interviewed. If an eligible woman was not available, the interviewer returned once later the same day. No other return visits were made. Questionnaire. Questions used were as similar as possible to those used in the 1998­99 and 2002 ENSMI surveys so that comparisons could be made with those results. Questions were included to assess vaccine coverage for children 12 to 59 months of age, use and qual- ity of prenatal care, use of care and satisfaction with care received for diarrhea, respiratory illness, and growth monitoring for children younger than 2 years. 60 World Bank Working Paper Interviews and data entry. Interviewers were selected for their previous experience with household surveys and their language skills. In most of the households, the interviews took place using the language primarily spoken, either Spanish or Mayan. In a few cases where a less common dialect was spoken, translators were required. Data were entered twice and analyzed by use of Epi-Info software. Data analysis. The descriptive analysis was carried out by use of Epi-Info.56 A logis- tic regression was carried out by use of LogXact to adjust for significant variations among the populations served by the different providers. Variables included in the regression were distance from health facility and community size. In the final model, results are strat- ified by department and adjusted values presented because, for some variables, results var- ied considerably by department. Provider Survey This study assessed the costs and productivity of 57 providers: 13 ADMSSs-MSPAS, 22 PSSs, and 22 HPs.57 These were the providers responsible for the catchment areas sampled in the household survey. The universe of SIAS providers working in the four departments from which the household survey was drawn consisted of 42 PSSs, 24ADMSSs, and 204 HPs (see annex table 2.1.1). Data collection. The study used three methods to obtain total cost and production data. First, providers were asked to keep accurate records of all services and activities delivered, as well as all inputs used to deliver such services during a three-month period.58This method was used to record the daily use of variable inputs such as fuel, supplies, per diem payments, training, and vaccines. Contracted surveyors who made regular visits to the NGOs in situ pro- vided assistance in completing the input survey forms. Prices for inputs were obtained from provider ledgers and receipts obtained from the NGOs. In the case of vaccines, which were supplied by the MSPAS, input prices were supplied by the central ministry. Prices of supplies for the HPs were obtained by MSPAS departmental offices.59 The consistency of a subset of the data was checked by comparing the completed forms with quarterly financial and pro- duction reports provided to the MSPAS by NGOs. Second, fixed costs were not measured directly through the survey. These included personnel costs as well as depreciation. Information about spending on salaries and bene- fits was collected from accounting ledgers and divided into two categories: direct provision (for example, doctors, nurses, and so on) and administration (for example, accountants, administrators, and so forth). Depreciation was based on an inventory of equipment, vehi- cles, and buildings. Prices of all equipment and vehicles were obtained from the NGOs and 56. This study provides averages for NGOs and traditional providers that individually may perform at widely varying levels. Improving the performance of NGO and traditional providers requires individ- ual assessments of their performance. 57. One PSS and one ADMSS were eliminated because of insufficient production data. 58. Because of variations in drug and medical supply purchases, records were kept over a six-month period. 59. The research did not obtain data from MSPAS health centers. Health System Innovations in Central America 61 the MSPAS (for HPs), and the cost of the HPs was estimated for the year of construction. Standard straight-line methods were applied, assuming a 5-year useful life for vehicles, equipment, and furniture and 20 years for buildings (according to Guatemalan account- ing standards), with no salvage value. A monthly depreciable cost was assigned for each item in the inventory. Finally, indirect MSPAS costs related to supervision, monitoring, oversight, and finan- cial management of the coverage extension program were estimated and apportioned to each provider in the sample. Spending data (executed budgets) were collected from three levels of the MSPAS: central institutional,60 central SIAS administration, and area (departmental) pro- gram administration. Based on interviews with MSPAS and program staff, allocation per- centages were crafted to distribute a proportion of total costs to the coverage extension program. Final cost distribution to the providers was based on the funds received by each provider as a percent of total provider financing. About 13 percent of area expenditures (of the four departments constituting the sample) and 3 percent of central-level spending were distributed to providers participating in the program. Data analysis. Weighted average costs were developed to compare specific types of interventions. The study assumed that unit costs vary for the different outputs for which the providers were responsible. For example, it is reasonable to assume that the unit cost of ARI care, which typically involves a physician consultation and the use of antibiotics, is higher than the unit cost of a well-baby checkup, which may be administered by a nurse and does not involve pharmaceutical inputs. However, observing the specific inputs used for each type activity or service was beyond the scope of the study and available financial resources. Rather, the research team gathered a panel of NGO and MSPAS physicians, nurses, and rural health technicians with ample experience in the program to determine the variable inputs for a subset of outputs. Variable inputs--such as drugs, medical supplies, and the time of each member of a provider team--were prorated among the set of nine outputs. Weights were derived and applied to each provider's direct costs.61 These weights are presented in Table 2.4. Because MSPAS and HP providers constitute a different personnel mix, separate weights were developed. For example, the average weighted costs for PSS providers were computed as follows: TC 21 PSWi PSS j ACi Pss = j =1 N 21 PSS j,1 j =1 where ACPSS = average cost of output i for PSS provider type, i TCPSS = total direct cost of PSS provider j, j WPSS = variable unit cost weight W for output i delivered by PSS provider type, and i N PSS = number of i outputs provided by PSS provider j. j,i 60. Eleven of 35 budgetary activities were identified as supporting SIAS. 61. This excludes both provider administrative costs and MSPAS indirect costs. 62 World Bank Working Paper Table 2.4. Weights for Selected Outputs PSS and Traditional Outputa ADMSS MSPAS HP ARI treatment 0.115 0.109 Diarrhea treatment 0.083 0.079 Prenatal consultation 0.122 0.124 Iron and folate supplementation 0.050 0.051 Well-baby checkup 0.03 0.002 Child vaccination 0.091 0.089 Vaccination of pregnant women 0.037 0.044 Other PBS interventions 0.307 0.329 Non-PBS interventions 0.188 0.174 Total 1.0 1.0 a. Output is a contact between a patient and the provider for a particular health problem or activity in which an intervention was performed. For the comparative unit and per capita cost analyses, we performed data adjustments. For HP and ADMSS providers with volunteer physicians,62 we applied the average salary and stipend cost for a physician paid by ADMSSs-MSPAS in the sample. For PSS providers with volunteer physicians, we applied the average salary and stipend costs for physicians paid by PSS providers without volunteers. Training and transportation costs for the health posts were not discernible from MSPAS accounting ledgers. Because both HPs and ADMSSs were assigned catchment areas close to towns and main roads, we applied ADMSSs' aver- age costs of these inputs to the HPs. The nonparametric Kruskal-Wallis test was used to compare the data from the three groups of providers. This test compares the means of the three groups and assumes that the shapes of the three distributions are similar, but does not assume normality.63 The hypotheses for the comparison of the three independent groups are: Ho--the samples come from identical provider groups; Ha--the samples come from different provider groups. If Ho was rejected, we applied the multiple comparisons test to determine the difference between pairs of provider groups. Findings Household Survey This section presents findings from the household survey comparing performance of the three provider types based on key indicators of the provision of PBS among people living in 62. Ten (of 22) health posts, 3 (of 12) ADMSSs, and 3 (of 21) PSSs reported having volunteer physi- cians at the time of the survey. 63. The test statistic is a function of the ranks of the observations in the combined sample. Given the small samples for each provider, and that tests indicated the data structures violated parametric assumption of inde- pendent observations and normality, the nonparametric Kruskal-Wallis is the more appropriate test. Health System Innovations in Central America 63 each of their catchment areas. The indicators studied include immunization rates, prenatal care uptake, use of tetanus toxoid, iron and folate supplementation during pregnancy, preva- lence and care of children with diarrhea and respiratory illnesses, growth monitoring, and satisfaction with care. Simple descriptive findings are presented first, followed by a summary table showing odds ratios adjusted for key factors. Comparability of the three samples. We looked at a number of variables to assess whether the three samples were comparable with each other. Table 2.5 presents general sample char- acteristics. Household response rates were more than 90 percent for each of the three catch- ment areas. They were slightly lower for the traditional health services sample. Certain households did not participate because of refusal (1.0 percent), no women between 15 and 44 years of age in the household (1.9 percent), and no one at home (4.5 percent). The inter- view rate for eligible women identified in the households that participated was similar in the three groups. Individual refusal rates were also low (less than 1 percent), and the main rea- son that an eligible woman did not participate was that she was not at home. The average number of women per household, their average age, and the average number of children per woman interviewed were similar for the three groups. The distribution of eligible women by department of residence varied. Traditional and direct catchment areas were sampled in all four departments, but mixed catchment areas were sampled only in Alta Verapaz and Quiché because the departments of Huehuetenango and Totonicapán do not have ADMSS providers. Table 2.5. General Sample Characteristics Catchment area Traditional Sample characteristic MSPAS HP Mixed ADMSS Direct PSS Number of households 939 1,005 1,026 in sample Number of households 854 (91%) 933 (93%) 964 (94%) interviewed Eligible women 1,114 1,287 1,357 Number of completed 1,060 (95%) 1,217 (95%) 1,278 (94%) interviews Average number of 1.3 1.3 1.4 women per household Average age of women 28.2 27.6 27.6 interviewed Average number of 1.2 1.2 1.2 children younger than five years per woman Department of residence Distribution of women interviewed (%) Alta Verapaz 30.2 57.4 38.0 Quiché 20.0 42.6 22.8 Huehuetenango 33.2 0 19.0 Totonicapán 16.5 0 20.1 64 World Bank Working Paper Sample community characteristics. The number of inhabitants in the communities sam- pled in the three catchment areas was similar (table 2.6). However, communities served by direct providers tended to be located considerably farther from health facilities (HP or health center) compared to communities served by mixed and traditional providers. Only 10 per- cent of direct provider communities were located within 3 kilometers of a health facility, compared to 22 percent and 35 percent, respectively, of mixed and traditional provider com- munities. Conversely, 56 percent of direct provider communities were more than 10 kilo- meters away (compared to 30 percent and 20 percent among mixed and traditional), and 24 percent were more than 20 kilometers away (data not shown). The quality of the roads leading to health facilities was also considerably worse in direct provider communities. All the communities sampled had 95 percent or more Mayan populations. Access to water was similar within the three groups, with communities in the traditional services catch- ment areas having slightly better access to potable water and those in direct provider services catchment having slightly worse access. Illiteracy rates are for the entire municipality in which each community lies and are similar for all three groups. Taken together, tables 2.4 and 2.5 demonstrate that, with the exception of distance from health facility and the quality of access roads, the women and their communities were rea- sonably comparable with each other based on the data collected. This is consistent with the SIAS design features, outlined in section 2.2, in which direct providers were generally assigned to more distant communities than mixed and traditional providers. Table 2.6. Number of Communities Sampled with Given Characteristic, by Provider Type Provider catchment area Traditional Mixed Direct Community characteristic MSPAS HP ADMSS PSS Number of inhabitants 0­600 17 18 17 601­1,200 12 19 13 >1200 19 15 19 Distance to health facility (km) 0­3 17 11 4 4­10 19 31 15 >10 12 10 29 Type of access (roads) Regular or good 44 46 37 Difficult or bad 4 6 11 Other characteristics Mayan population (%) 96 95 98 Households with water (%) 77 73 68 Illiteracy (%)a 60 65 62 Total number of communities 48 52 48 a. Illiteracy rates are for the municipality in which each community lies. Health System Innovations in Central America 65 Immunization rates. Immunization rates varied considerably among the four depart- ments, with Totonicapán having significantly worse rates (data not shown). Because mixed provider catchment areas were sampled only in Quiché and Alta Verapaz, immunization rates are shown for catchment areas served by all three provider types in those two depart- ments only (Table 2.7).64 Immunization rates for all four departments together are shown for catchment areas for direct and traditional providers because they both included Totonicapán. For the departments of Quiché and Alta Verapaz, children in mixed provider catchment areas consistently have the best immunization rates. Children who live in direct provider catch- ment areas in Quiché and Alta Verapaz consistently have the lowest immunization rates. However, direct provider and traditional areas have very similar rates when data from all four departments are used. Table 2.7. Immunization Rates (%) among Children 12­59 Months of Agea Type of immunization All four departments Quiché and Alta Verapaz only and other Traditional Mixed Direct Traditional Mixed Direct characteristics MSPAS HPs ADMSS PSS MSPAS HPs ADMSS PSS All vaccines 73.0 NA 74.2 82.0 84.3 74.5 No vaccine 5.2 NA 5.7 3.4 2.7 4.1 DPT3 80.1 NA 80.7 86.5 89.3 81.6 Polio 3 81.0 NA 81.7 86.8 89.5 82.4 Measles 81.7 NA 82.0 86.5 89.6 83.5 BCG 93.0 NA 87.7 93.4 93.0 88.8 Percent with carnet 59.2 NA 53.8 57.1 57.3 51.4 Number of children 860 NA 1,080 396 1,069 686 BCG Calmette-Guérin bacillus. DPT diphtheria-pertussis-tetanus. NA not applicable. a. Information obtained from the carnet or the mother. Prenatal care. PNC rates did not vary as much by department as did immunization rates. PNC rates (at least one visit) were similar and reasonably high in traditional and mixed provider catchment areas (86 and 87 percent, respectively) but were lower in direct provider areas (75 percent; see Table 2.8). First trimester care rates were slightly higher for women in mixed provider areas. Among women who had at least one PNC visit, the average number of visits during the pregnancy was similar in the three groups ranging from 2.8 (direct provider areas) to 3.0 (mixed provider areas). The quality of PNC received, as measured by tetanus toxoid vaccination and iron and folate supplementation, was also highest in mixed 64. Results for the other variables do not vary substantially by department and are therefore shown in the aggregate for all four departments by provider type (catchment area). Apparently there is a long history of distrust in the safety of immunizations among the ethnic groups who live in Totonicapán, and rates there are consistently lower than in other departments. 66 World Bank Working Paper Table 2.8. PNC Use and Services Provided Catchment area Traditional Mixed Direct Characteristic MSPAS HP ADMSS PSS Percentage and number of 1999 63.8 68.7 63.4 women who have given birth since July (545) (687) (672) Women who gave birth Percentage with at 75.0 86.9 78.0 least 1 PNC visit Average number of PNC visits 2.9 3.0 2.8 Percentage who received 61.9 63.8 55.1 tetanus toxoid Percentage who received iron 57.7 68.0 57.8 Percentage who received folate 52.0 62.4 52.6 Women with more than 1 PNC visit Percentage with 1st trimester PNC 32.3 35.6 29.1 provider catchment areas. Rates for these indicators were higher in traditional catchment areas compared to direct provider areas. Diarrhea and respiratory illness among children less than two years old. Prevalence rates for diarrhea and respiratory disease in the three months prior to the survey are shown in Table 2.9. Diarrheal prevalence was highest in mixed provider catchment areas and lowest in traditional catchment areas. Women in mixed provider communities were more likely to have used ORS during their child's last episode of diarrhea and more likely to say they knew how to use ORS. Compared to traditional provider communities, women in direct provider Table 2.9. Diarrhea and Respiratory Illness Prevalence (in the Last 3 Months), Health Service Use, Knowledge and Use of ORS Catchment area Traditional Mixed Direct Characteristic MSPAS HP ADMSS PSS Number of children < 2 years old 515 595 616 Prevalence of diarrhea in the last 3 months (%) 45.2 55.8 49.7 With diarrhea and received ORS (%) 42.1 51.8 44.8 With diarrhea and received medical care (%) 48.0 53.8 46.0 Prevalence of respiratory illness in the last 3 months (%) 35.5 22.7 33.4 With respiratory illness and received medical care (%) 74.3 77.0 68.4 Women with children < 2 years who say (%) 41.8 54.5 45.6 they know how to use ORS Women with children < 2 years who have ORS in the home (%) 12.6 19.8 12.9 Health System Innovations in Central America 67 communities were slightly more likely to have used ORS during their child's last episode of diarrhea and to say they knew how to use it. They were just as likely to have it in the house. The percentage of women with children under two years of age who had ORS in the home was low (less than 20 percent) in all three types of communities. Use of medical care for diar- rhea and respiratory diseases in children was high in all three groups of communities, vary- ing from 46 percent to 54 percent for diarrhea and from 68 percent to 77 percent for respiratory illness. Children in mixed provider and traditional communities were more likely to have received medical care during an episode of illness than children in direct provider communities. Women were more likely to seek medical care for infants with respiratory dis- ease than for those with diarrhea. More than two-thirds of mothers sought care for infants and children with respiratory illness. Growth monitoring. MSPAS guidelines in 2001 called for growth (weight and height) monitoring every two months among children younger than two years of age. However, only a small percentage of children under two in each of the three groups had received growth monitoring in the three months prior to the survey (3­12 percent), as shown in table 2.10. Among those who did, nearly all were weighed, though children from traditional and direct provider communities were more likely to be weighed than those in mixed provider com- munities (93 percent and 100 percent compared to 86 percent). Children in direct provider communities were more likely to have their weight noted on their carnet (data not shown). However, fewer children had their height measured during their last checkup (68­81 per- cent). Very few children had their height noted on their carnet. Table 2.10. Growth Monitoring in the Last 3 Months Catchment area Traditional Mixed Direct Characteristic MSPAS HP ADMSS PSS Number of children < 2 years of age 544 639 663 Growth monitored in the last 3 months (%) 3.1 8.3 12.4 Weighed during last checkup (%) 100.0 86.3 96.3 Height measured during last checkup (%) 70.6 68.2 81.0 Adjusted odds ratios. Estimated adjusted odds ratios from the final logistic regres- sion model are summarized in Table 2.11.65 The model includes distance from health center and size of community. Age of mother did not significantly change results and was not included in the final model. Results are stratified by department to adjust for any differences. The model adjusted for differences in outcomes associated with dis- tance from health facility and community size. The farther women lived from a health facility, the worse their health indicators tended to be, whether they received care from a mixed, direct, or traditional provider. Because the majority of women who lived far- 65. Crude odds ratio data are presented Annex Table 2.1.2. 68 World Bank Working Paper Table 2.11. Adjusted Odds Ratio for Key Health Indicators: Mixed and Direct Provider Communities Compared to Traditional Communities Adjusted odds ratioa Traditional Mixed Direct Characteristics MSPAS HP ADMSS PSS Children aged 12­59 months with vaccineb BCG 1.00 1.46c 0.94 DPT3 1.00 1.76c 1.05 Polio3 1.00 1.72c 1.05 Measles 1.00 1.55c 1.05 All vaccines 1.00 1.70c 1.11 Women with history pregnanciesb At least 1 PNC visit 1.00 1.69c 1.28 First trimester PNC visit 1.00 0.95 0.93 Received tetanus toxoid 1.00 1.53c 0.96 Received iron 1.00 1.73c 1.14 Received folate 1.00 1.43c 1.01 Children younger than 2 years old Prevalence of diarrhea in the last 3 months 1.00 1.58c 1.21 Diarrhea in the last 3 months and received ORS 1.00 1.55c± 1.23 Diarrhea and received medical care 1.00 1.29 1.05 Prevalence of respiratory illness in the last 3 months 1.00 0.71c 0.99 Respiratory illness and received medical care 1.00 1.14 0.89 Growth monitored in the last 3 months 1.00 3.10c 5.49± Weighed during last checkup 1.00 0.28 1.84 Height measured during last checkup 1.00 1.04 0.62 Women with children younger than 2 years old Say they know how to use ORS 1.00 1.67c 1.31c Have ORS in the home 1.00 1.06 1.06 a. Adjusted for woman's age, number of pregnancies in the last 5 years (proxy for fecundity), distance from health center, and community population. b. Information from the carnet or from the mother when the carnet was not available c. P<.05 compared to traditional MSPAS HP. ther than 10 kilometers from a health facility had a direct provider, it was necessary to adjust the results to make a fair comparison. The results were also stratified by depart- ment because of the differences, particularly in immunization coverage, noted by department. The fact that the mixed NGOs did not provide care in Totonicapán, where immunization rates are particularly low, contributed to higher coverage; stratifying by department adjusted for this difference. Table 2.11 summarizes the adjusted odds ratios, comparing indicators from mixed and direct provider communities to those in traditional provider communities. The odds of receiving each of the vaccines as well as all of the vaccines were significantly better for Health System Innovations in Central America 69 children in mixed provider catchment areas compared to traditional areas. The odds of being vaccinated for children in direct provider catchment areas were very similar to, and not significantly different from, results in traditional areas. The odds of PNC use were also greater among women in mixed provider areas compared to traditional areas. The likelihood of first trimester care was not different for women in the three types of catchment areas. However, women in mixed provider areas were also signifi- cantly more likely to have received tetanus toxoid, iron, and folate during their pregnancy. Neither the odds of receiving prenatal care, nor of receiving tetanus toxoid or iron and folate supplementation, were significantly different for women in direct and traditional provider catchment areas. Diarrhea and respiratory disease prevalence was significantly higher among children in mixed provider communities. Women in mixed provider areas were more likely to use ORS than were those in traditional areas, and women in mixed and direct provider areas were more likely to say they knew how to use ORS than women in traditional areas. However, there were no significant differences in possession of ORS at the time of the interview among women in the three types of catchment areas. Likewise, there was no significant difference in the odds of using medical care for children less than two years of age who had diarrheal dis- ease or a respiratory illness. Children in mixed and direct provider areas were significantly more likely to have had their growth monitored compared to those in traditional areas (although, as noted, coverage was low for all three provider types). Choice of health services. There was a striking degree of cross-utilization of services by mothers in each of the three groups using services not specifically assigned to them (table 2.12). This was particularly true for mothers in traditional health service catchment areas, where only 25 percent of pregnant women, 36 percent of mothers whose children had diar- rhea, 27 percent whose children had respiratory illness, and 24 percent who took children for routine growth monitoring went to their assigned HP for care. Instead, 23 percent of preg- nant women, 38 percent of children with diarrhea, 42 percent with respiratory illness, and 6 percent who went for growth monitoring received care in community centers and 16, 20, 28, and 47 percent, respectively, in health centers. Cross-utilization also occurred in direct provider communities, but to a much lesser extent, with 38 percent of pregnant women, 67 percent of mothers whose children had diar- rhea, 61 percent of those with respiratory illness, and 83 percent who took children for growth monitoring receiving care in a community center. Mothers in mixed provider communities used community centers and HPs nearly equally. Mothers in mixed provider communities and direct provider communities also took themselves and their children to health centers for pregnancy care (22 and 20 percent, respectively), diarrhea (33 and 14 percent), respira- tory illness (19 and 11 percent), and growth monitoring (20 and 2 percent). The study also found that an important percentage of women continue to choose TBAs for their PNC, and this percentage was highest for women in HP provider areas (35 percent) and lowest (21 per- cent)for women in mixed provider areas. Satisfaction with care. Satisfaction with care was measured using four indicators (prob- lem resolution; provision of medicines; waiting more than one hour; and friendliness of services), as shown in figure 2.2. More than 70 percent of women responded positively to 70 World Bank Working Paper Table 2.12. Choice of Health Services for Women and Children Catchment area Traditional Mixed Direct Type of health service used MSPAS HP ADMSS PSS Women who had prenatal care In community center (%) 23 35 38 In HP (%) 25 25 10 In health center (%) 16 22 20 At home--TBA (%) 35 21 29 Children who received medical care for diarrhea In community center (%) 37.7 29.6 67.2 In HP (%) 36.0 27.3 12.7 In health center (%) 20.2 33.0 14.2 Children who received medical care for respiratory illness In community center (%) 41.5 38.4 61.4 In HP (%) 26.6 28.3 15.8 In health center (%) 27.7 19.2 11.4 Children who were weighed In community center (%) 5.9 39.2 82.9 In HP (%) 23.5 33.3 6.1 In health center (%) 47.1 19.6 2.4 these questions for all three types of providers. The exceptions were answers to questions about traditional providers' waiting times and friendliness of care for children with respira- tory illnesses. Community centers in ADMSS communities generally received the highest scores for satisfaction. Community centers in both mixed and direct provider communities scored particularly well for friendliness of care and waiting times compared to traditional HPs. Mixed provider services were identified as more likely to resolve problems then the other provider types. However, perception of problem resolution was generally high for all three providers. By and large, women also said they received the medicine they needed from all provider types. Although satisfaction rates were generally high, they tended to be better for care provided for diarrhea than for care provided for respiratory illness. Figure 2.2 shows only satisfaction with care received from the assigned provider. Data not shown reveals that women from traditional areas who cross-utilized community cen- ters consistently rated satisfaction with care received higher than from their assigned health posts. Women in all three types of catchment areas rated satisfaction with care higher for community centers for all the indicators except for receipt of needed medicines for diar- rhea from direct providers where women rated health posts slightly more positively. Provider Survey This section presents findings comparing measures of technical and economic efficiency, including labor productivity and average unit and per capita costs, for each type of provider. Health System Innovations in Central America 71 Figure 2.2. Percent of Women Satisfied with Care Received for Their Children under 2, by Catchment Area of Residence and Type of Illness 66 88 Friendly 98 100 care 98 99 61 Waited 72 <= 1 77 hour 81 80 80 Traditional Rec'd 78 medicine 96 90 Mixed they 95 82 needed 84 Direct 81 Problem 92 98 was 97 84 resolved 91 150 100 50 0 0 50 100 150 Respiratory Illness Diarrhea Note: Only satisfaction with care received from the catchment area's assigned provider is shown. Depending on the analysis, the findings are classified by types of services provided: (a) all ser- vices and activities,66 (b) all services and activities excluding vaccinations,67 (c) PBS inter- ventions (see Annex 2.1), and (d) MSPAS priority interventions (that is, care of ARI, care of diarrhea, prenatal care, and well-baby checkups) included in the PBS. Data on total produc- tion and spending by provider and service type are presented in Annex Table 2.1.3. Service mix, production, and productivity. Table 2.13 displays the distribution of service volume by selected interventions reported by the sampled providers. The distribution of ser- vice production is relatively consistent among providers, but some differences are note- worthy, suggesting a need for greater standardization of the PBS. All providers concentrated on the basic service package, including vaccinations, which accounted for more than three- quarters of provision. Vaccinations represented about 40 percent of all outputs for each provider, though for the PSSs, vaccinations accounted for a lower share of total outputs than the other providers. Differences among the groups were significant for the vaccinations and "other PBS" categories. For both these categories, HPs were significantly different from the other providers. Non-PBS services are higher in the HPs (22 percent of total volume, com- pared to 14 percent for the PSSs and ADMSSs), suggesting that they cater to spontaneous morbidity demand of the population whereas the PSSs and ADMSSs tend to focus on PBS activities. Although all providers display low levels of outputs related to the four priority interventions (care of ARI, care of diarrhea, PNC, and well-baby checkups), taken together 66. Costs for the all services ("all") category are unweighted, and all other cost data are based on weights presented in Table 2.3. 67. Vaccinations were not part of the routine service provision by the sampled providers, but they were also given during MSPAS-organized campaigns in which all providers participated (see Annex 2.2). 72 World Bank Working Paper Table 2.13. Percent Distribution of Total Monthly Outputs,a by Provider and Service Type, 2001b Type of service of Direct PSS Mixed ADMSS Traditional HP activity (N = 21) (N = 12) (N = 22) Priority PBS Care of ARI 9.2 9.2 7.3 Care of diarrhea 3.3 1.7 2.5 PNC 3.6 2.5 2.2 Well-baby checkupc 9.4 1.5 3.4 Vaccinationsd 37.9 38.6 43.2 Other PBS benefitd 22.4 32.5 19.2 Non-PBS benefit 14.2 14 22.2 Total (100%) 28,429 14,231 17,348 a. Contacts between a provider and patient or community member in which a specific service was provided or activity performed. b. January­March 2001. c. Growth and development monitoring of children under 2 years old. d. P<05 (difference between PSSs, ADMSSs, and HPs). these activities represent about 25 percent of PSS output volume compared to 15 percent for ADMSSs and HPs. Figures 2.3 and 2.4 compare average monthly production and labor productivity for each provider model for three categories of interventions: "all services," PBS, and "all ser- vices" except vaccinations.68 Figure 2.3 displays the results for all sampled providers. For each category, the differences among the groups and between each pair of providers are sig- nificant (P<.05) in terms of production (average monthly volume of outputs per provider). PSSs demonstrate markedly higher production than the HPs and slightly, but still signifi- cantly, higher production than the ADMSSs. Although the results are similar for produc- tivity (average monthly provider volume per health worker providing care),69 differences are significant for only the "all services" and PBS categories. PSS personnel provided an aver- age of 568 interventions per month, 10 percent more productive than the HPs. However, PSS personnel were 18 percent more productive for PBS interventions and for interventions excluding vaccinations. The differences between ADMSS and HP providers are less marked: the former are 7, 17, and 14 percent more productive for "all," PBS, and nonvaccination cat- egories, respectively. These differences were found to be significant for the two former cat- egories of interventions. The presence of a physician on a provider team can be an important determinant of demand and therefore drive production. Most PSS and ADMSS teams in the sample include physicians, and about half of the HPs were staffed with volunteer Cuban physicians.70 68. Because vaccinations represented about 40 percent of all outputs for each provider, and were pro- vided during national campaigns as well during regular visits, removing vaccination outputs provides a better estimate of services regularly provided in the communities constituting providers' catchment areas. 69. M.D., nurse, auxiliary nurse, and rural health technician. 70. Ten of 12 ADMSSs, 16 of 21 PSSs, and 10 of 22 HPs reported having physicians on the delivery teams. Health System Innovations in Central America 73 Figure 2.3. Average Monthly Production and Productivity By Output Category and Provider Type, 2001 568 1,354 547 All* 1,186 510 789 Direct 486 1,162 471 PBS* 1,020 Mixed 397 614 352 Without 840 Traditional 336 vacci- 728 290 nation* 447 600 400 200 0 0 500 1,000 1,500 Productivity Production (outputs per worker) (outputs per provider) *P<.05 (across groups and between each pair of providers for production). P<.05 (across groups and between each pair of providers for productivity). Figure 2.4 compares performance for providers with physicians on the delivery team. Unlike the comparisons of Figure 2.3, differences were not significant for each cat- egory of interventions presented in Figure 2.4. Although average production and pro- ductivity of PSS personnel are higher than those of the other two provider types, dif- ferences between the ADMSSs and HPs are negligible. In fact, productivity of HP personnel (538) for the "all interventions" category is actually greater than for ADMSS personnel (494). Comparing the data in Figure 2.3 (full sample of providers) with that of Figure 2.4 (subset of providers with M.D. degrees), the presence of a physician on the HP team increases production and productivity by more than 40 percent. (PSSs register a marginal increase in production and productivity, and ADMSSs report a slight decrease.)71 For example, the average monthly production per HP is approximately 50 percent higher in HPs with physicians for all categories of interventions ("all services," PBS, and without vaccines). These results are self-evident to the extent that a physician represented an addi- tional member of the team for HPs and was expected to perform outreach visits to com- munities in the HP catchment areas.72 However, the presence of a physician probably stimulated demand--especially in communities that previously had little or no access to physician services. Comparisons of productivity showed mixed results. HP teams con- taining M.D.s are 7 and 2 percent more productive for "all services" and PBS categories, but 9 percent less productive for interventions if vaccines are excluded. This suggests that having a physician on staff may generate greater demand for services at HPs, but this does not necessarily result in greater productivity. 71. Because nearly all PSSs and ADMSSs had physicians, it was not possible to compare those providers with and without physicians (see note 33). 72. In the case of PSSs and ADMSSs, nurses or rural health technicians were hired if the providers were unable to recruit physicians. 74 World Bank Working Paper Figure 2.4. Average Monthly Production and Productivity by Output Category and Provider Type with Physicians, 2001 561 1,437 All 494 1,144 538 1,132 Direct 484 1,241 420 PBS 966 Mixed 411 864 Traditional 339 890 Without 307 705 vacci- 290 nation 609 600 400 200 0 0 500 1,000 1,500 2,000 Productivity Production (output per worker) (output per provider) Expenditures, unit and per capita costs. This section presents both observed and adjusted costs derived from the provider survey. Observed costs should be interpreted with caution. First, they do not account for the economic costs of volunteer physicians.73 Second, MSPAS ledgers did not permit the identification of training and vehicle maintenance costs or stipend payments to volunteers. Third, 2 of 12 ADMSSs and 2 of 21 PSSs were without physicians at the time of the survey. Because of the small sample of 12 ADMSSs, the absence of physicians in 2 of them and the use of Cuban volunteers in another 2 skewed average personnel costs downward for this category of providers. We performed adjustments to correct some of these data gaps for unit, per provider, and per capita costs.74 Cost structures varied among the provider models. Average (unadjusted), per provider, monthly expenditures by the PSSs were about three times greater than those of the ADMSSs, but more than five times higher than those of HP providers, as shown in Table 2.14. Indirect MSPAS cost allocation represents a greater share of total costs for ADMSSs and HPs because the MSPAS hired personnel and was responsible for technical oversight of the delivery teams for these providers. In relation to direct spending (excluding provider overhead and MSPAS indirect costs), Table 2.14 also shows that about one-fifth of PSS and ADMSS spending went for training and vehicle maintenance. Higher levels of spending for transportation by PSSs resulted from more difficult access and greater travel distance to and among the communi- ties in their assigned catchment areas. Also, PSSs and ADMSSs had to train a larger number of new personnel and provide for a wider range of training than other providers. According to NGO informants, high rotation of personnel resulted in elevated training costs, even after three years of implementation. ADMSS training costs were less than those of the PSSs because the former tapped experienced and tenured MSPAS rural health technicians. As mentioned, HPs' costs related to direct administration, training, and vehicle maintenance could not be ascertained from MSPAS spending ledgers. 73. Ten of 22 HPs, 3 of 21 PSSs, and 2 of 12ADMSSs reported volunteers as their staff physicians. 74. Adjustments consisted of attributing: (a) the average cost of contracted Guatemalan doctors reported by the ADMSSs and PSSs to the volunteer Cuban physicians reported for all providers and (b) average costs of ADMSS training and vehicle maintenance to the HPs. Health System Innovations in Central America 75 Table 2.14. Percent Distribution of Total and Average Monthly Expenditures (Observed) by Spending Category and Provider Type, January­March 2001 Direct PSS Mixed ADMSSS Traditional HP Expense category (N = 21) (N = 12) (N = 22) Total spending Q 1,132,079 Q 398,486 Q 206,869 Average per provider spending Q 53,909 Q 33,207 Q 9,403 Distribution (%) 100 100 100 Indirect Provider overhead 3 4 0a Central MSPAS overhead 7 13 20 Area MSPAS overhead 3 5 7 Direct Salaries and stipends 49 34 39 Depreciation 3 2 <1 Training 19 16 0b Drugs, medical supplies 11 23 32 Utilities, rent <1 <1 2 Vehicle maintenance 5 3 0b Note: US$1 = Q 7.9. a. Included in indirect costs of area and central MSPAS. b. Spending for these items was not specified in MSPAS ledgers. Average monthly PSS personnel costs alone dwarf total monthly spending by HP providers and are markedly higher than ADMSS personnel spending. We found two reasons for this disparity. First, given the larger assigned population,75 PSS teams consisted of more professional personnel than other providers. For example, PSSs reported teams consisting of 2.3 nonadministrative personnel on average, compared 2.0 for ADMSSs and 1.5 for the HPs (see table 2.3). Second, PSSs recruited more volunteer community facilitators, incurring higher outlays for travel stipends than ADMSSs. The former averaged Q13,770 per month, compared to the latter's monthly expense of Q6,550. We found no evidence of payments to community volunteers working in HP catchment areas.76 PSSs and ADMSSs reported paying above-market salaries and travel stipends to attract personnel to work in distant communities, particularly physicians. On average, ADMSSs paid Q6,000 per month for salaries and stipends, compared to Q5,850 for PSSs. In comparison, senior-level MSPAS area and district Directors earned approximately Q6,000 and Q4,500, respectively. Higher PSS and ADMSS spending on drugs and supplies was related to higher levels of production, which in turn probably was linked to the larger populations residing in their catchment areas. Regarding the use of drugs and supplies, the ADMSSs were the most resource intensive, averaging Q7,600 per provider per month, compared to Q6,000 and Q3,000 for PSSs and HPs, respectively. 75. As reported in the "Issues" section of this chapter, according to provider registers, PSSs were nom- inally responsible on average for 19,400 people, compared to 12,500 per ADMSS and 9,500 per HP. 76. Nor did we find evidence of ADMSS or PSS payments for other community workers, such as com- munity health promoters or TBAs. However, these providers did finance training for them. 76 World Bank Working Paper Figure 2.5 compares Figure 2.5 Average per Provider Costs: Observed and average, monthly costs per Adjusted (in Q$) provider by use of observed and adjusted spending data. The adjustments result in Direct a doubling of HP per pro- vider costs (from Q9,400 Mixed to Q18,800), and PSS and ADMSS costs increased Traditional slightly. Adjusting for the use of volunteer physicians Q0 Q20,000 Q40,000 Q60,000 in nearly half the HPs con- tributed to the increase in Observed Adjusted costs. After adjustments, HP costs per provider are about one-third and one-half of those for PSS and ADMSS providers, respectively. Survey findings demonstrate large and mostly consistent differences in average unit (per intervention) costs. Table 2.15 presents the average observed and adjusted unit costs by activity category and provider type. For all three categories of interventions, the observed (unadjusted) unit cost differences between PSSs and HPs were significant (P<.05). PSSs have the highest observed unit costs, and HPs have the lowest. For the total and PBS categories, PSSs have triple the units costs of HPs, but are six times more expen- sive for non-PBS interventions. ADMSSs occupied an intermediate range between the PSSs and HPs, but differences were not statistically significant. Average unit costs for specific activities follow this trend, except for well-baby checkups.77 Table 2.15 Average Monthly Unit Costsa by Activity and Type of Provider (in Quetzales), 2001 Direct PSS Mixed ADMSS Traditional HPs Activity Observed Adjusteda Observed Adjusteda Observed Adjusteda All services 40 40 28 30 12 24 PBS* 38 38 27 28 13 25 Nonbasic package* 53 54 38 40 9 19 Selected activities Care of ARI* 50 51 35 37 18 35 Care of diarrhea* 102 103 141 151 38 76 PNCb,* 192 195 194 208 93 185 Well-baby checkup* 1 1 5 5 1 2 Vaccinations* 14 14 10 10 4 7 Note: US$1 = Q 7.9. a. Adjusted for volunteer physicians (all providers) and for training and vehicle costs (HPs only). b. Includes cost of iron supplementation. *P.05 (differences between each pair of providers for observed and adjusted unit costs). 77. The relatively lower volume of output for PNC and well-baby checkups for ADMSS providers drove the higher unit costs. Health System Innovations in Central America 77 The adjustments reduced the differences among providers, but differences remained sig- nificant (P<.05) and the tendency remained: PSSs displayed significantly higher unit costs for each category of intervention, and HPs operated with the lowest unit costs. However, PSSs displayed significantly lower adjusted unit costs for care for diarrhea, PNC, and well-baby checkups than their ADMSS counterparts, in part because of the low levels of output observed for the latter. Per capita costs of population coverage by provider type are presented in Figure 2.6. Data on nominal coverage are drawn from PSS and ADMSS contracts and HP records that were derived from rapid household surveys that all providers are mandated to con- duct annually. Consonant with the unit cost findings, HPs are the lowest cost provider (P<.05).78 Figure 2.6 Cost per Capita by Provider Type, Nominal and Effective Population Coverage, (Adjusted Costs) 2.8 Nominal 2.8 coverage 2.0 Direct Mixed 5.7 Effective Traditional 8.5 coverage 7.6 0 2 4 6 Cost per capita (Q)* *P = <0.05 (differences between each pair of providers for both nominal and effective coverage). Adjusting for effective coverage,79 PSSs become the least-cost provider (Q5.7 per capita), followed by HPs (Q7.6) and ADMSSs (Q8.5). The differences among the pairs of providers are statistically significant (P<.05). Nevertheless, because of the lower effective population coverage by HPs, they provide an average of 4.0 interventions per capita, compared to 2.2 for the ADMSSs and 2.0 for the PSSs. In short, for the popula- tion that the HPs effectively cover, they provide more services per capita than the other provider types. 78. Also, although not shown in figure II.6, HPs provide more monthly outputs per capita (1.8), com- pared to PSSs (1.5) and ADMSSs (1.1). These results should be interpreted with caution because they are based on nominal coverage. 79. Drawn from the household survey, adjustments are based on the observed demand for priority interventions (ARI care, diarrhea care, PNC, and well-baby checkups) by residents of a specific catch- ment area from the provider assigned to that catchment area. Vaccinations were not included because they were provided generally through vertical campaigns and were not part of the regular delivery system (see Box 2.1). Effective population coverage for PSSs was 52 percent, compared to 35 and 26 percent, respectively, for the ADMSSs and HPs. 78 World Bank Working Paper Discussion The new direct and mixed provider delivery models performed as well as--and, in the case of mixed providers, often better than--traditional health care services in delivering a basic health package to rural, indigenous communities that previously had no or limited access to health care services. These provider models are generally more productive and display higher labor productivity than the traditional model. Users also report higher demand for and sat- isfaction with the care received. Results on economic efficiency are mixed, however. Direct and mixed model providers have significantly higher unit and per provider costs than their traditional counterparts. Yet direct providers display lower per capita costs than the other providers when accounting for effective population coverage and the economic costs of volunteer physicians, training, and transportation. Mixed providers generally register inter- mediate performance on unit costs, production, and productivity compared to the other provider models. These findings suggest that mixed and direct provider models are good alternatives to traditional health care services for providing a basic package; the findings also support the models' continued extension to other rural areas of Guatemala that still lack health service coverage. These models are viable alternatives as long as the government con- tinues to commit the additional funding required to underwrite their higher costs. Findings The household survey found that communities served by mixed providers tended to have better performance indicators than those in communities served by traditional and direct providers for immunizations, PNC use, quality of PNC services, and use of ORS during chil- dren's episodes of diarrhea. Mixed provider communities had significantly better immu- nization and PNC coverage than communities with traditional providers.80 Mixed provider communities had better outcomes even though these areas had previously not been covered by regular health services. Both mixed and traditional providers' catchment areas were com- parable in the sense that they were located close to main roads and towns, and residents had access to providers outside their immediate areas. After controlling for distance from a health facility, the performance of direct providers was found to be very similar to that of traditional providers for nearly all indicators measured, which provides evidence that this model is a good alternative to traditional health services for the basic health care package studied. This is a key finding, considering the more difficult con- ditions of delivering care in the PSS catchment areas where communities were more dis- persed and access more difficult. Although this study did not benefit from baseline data, it is likely that baseline indicators were worse at program onset (in 1997) in areas served by PSSs (and to a lesser extent, ADMSSs) than those in areas served by traditional health services. Women from all of the three community groups are generally more satisfied with care provided in community centers of ADMSS and PSS providers than care provided in HPs. Care is considered friendlier, and reported waiting periods are shorter. Problem resolution and receiving needed medicine is similar to or better than the situation in HPs. However, 80. Some PSSs attributed low levels of vaccination coverage to the MSPAS supply system. They reported breakdown of logistics and communications between MSPAS districts and PSSs for the distrib- ution and storage of vaccines. Annex 2.2 describes problems in the distribution chain. Health System Innovations in Central America 79 these findings may be related to lower expectations among the women who previously had no access to care. Distances presumably hindered women from direct provider communities from using HPs and health centers and may have contributed to increased perceptions of sat- isfaction. Nevertheless, the study also finds that many women served by traditional providers "voted with their feet" and sought services in community centers in catchment areas served by mixed and direct providers. Furthermore, women from traditional provider areas rated community centers higher on every indicator of satisfaction with care that was measured. The provider survey found that PSSs demonstrate the highest costs as well as the highest production and labor productivity. They also appear more efficient in terms of effective population coverage. Contrarily, traditional providers show the lowest costs and levels of production and productivity, and the mixed providers display generally intermediate levels. Having a physician on staff significantly raised the production of HPs, probably in response to increased demand. However, productivity (output per worker) remained nearly equal to that of HPs with only a nurse auxiliary. The PSSs consistently outperformed the ADMSSs on production and productivity mea- sures (despite having a similar team composition and delivery models) while providing ser- vices in more distant, dispersed, and hard-to-access communities. They also did a better of job in delivering priority PBS interventions. Several explanations can be proffered regarding PSS performance. First, the higher average production for the PSSs could relate to the greater number of people residing in their catchment areas. Second, it is likely that there was pent- up demand among residents of PSS catchment areas, given the difficult access to other providers located in or near towns. Third, on average, PSSs had slightly more care-giving per- sonnel than ADMSSs (2.3 versus 2.0). A final explanation contributing to observed produc- tion and productivity may relate to the incentive system inherent in the contract.81 Unlike the ADMSSs and HPs, the PSSs were fully responsible for nearly all inputs (except vaccines) as well as the planning, management, technical oversight, and delivery of contracted services. In contrast, the ADMSSs relied on MSPAS staff, particularly rural health technicians, as well as MSPAS technical management and oversight.82 The HPs were fully dependent on the MSPAS for supervision as well as all inputs and support activities. Higher productivity may relate to superior technical supervision provided by PSS management compared to super- vision of the mixed and traditional providers provided by MSPAS districts. Also, PSSs may have been more exacting in registering outputs than other providers. Nonetheless, the PSSs were facing additional pressures. As explained in chapter I in this volume, in 2000, the gov- ernment introduced elements of competition, particularly among direct model NGOs. Con- tract renewal depended on a performance review, and all new contracts were competitively bid. For example, 24 NGO contracts were not renewed in 2000 because providers failed to meet minimum performance standards and reporting requirements set by the MSPAS. Another 40 received conditional renewals. The MSPAS also reported that upon tendering the cancelled contracts in 2001, more NGOs submitted bids than there were available contracts. In sum, PSSs faced strong incentives to deliver the PBS to all communities in their catchment areas as well as maintain production and financial records. 81. The higher observed productivity is consistent with other studies comparing the efficiency of pub- lic and private providers (Bitrán 1995; Levin and others 1999; Lavadenz 2001). 82. The ADMSSs were essentially input managers, hiring additional staff and purchasing and distrib- uting other inputs. 80 World Bank Working Paper Direct providers are costly compared to the mixed and traditional alternatives. Serving distant villages increased transportation costs and contributed to higher personnel costs because PSSs paid above-market salaries as well as travel stipends to attract and retain per- sonnel. For example, on average, PSSs spent nearly as much on transportation as the HPs spent on drugs and supplies. Further, all personnel hired by the PSSs were new to the MSPAS system, which contributed to higher training costs. The typical PSS provider spent nearly 50 percent more on training alone than traditional providers spent on all inputs. Supervision costs were also high because PSSs often had to field and two or more teams to cover the com- munities in their assigned catchment areas. Nevertheless, when accounting for effective pop- ulation coverage and adjusting for the economic costs of volunteer doctors, training, and transportation (not included in spending ledgers of traditional providers), PSSs are more effi- cient than both traditional and direct providers. In fact, the mixed providers have the highest per capita costs (for effective coverage), but the differences from traditional providers are not significant. Findings Compared to National Data National immunization rates from ENSMI 2002 were only slightly improved over immu- nization rates from ENSMI 1998­99, with full coverage increasing from 60 to 63 percent. However, these national averages mask substantial improvements during this period for indigenous children, whose full immunization coverage increased from 51 to 61 percent and nearly equaled that of ladino children, whose full coverage rates actually decreased slightly from 65 to 63 percent. One of the questions raised is whether and how much extension of coverage contributed to these improved outcomes for indigenous children. Compared to immunization rates provided by ENSMI 2002, mixed providers performed better than the national average for every vaccine. Direct providers performed better for DPT3, polio3, and full coverage but slightly worse for BCG and measles. Compared to national data for indige- nous children, all three groups did better than the national average. This suggests that exten- sion of coverage contributed to the increase in indigenous vaccination rates. Compared to the national PNC coverage rate of 90 percent, coverage rates were slightly lower for women in mixed provider communities and much lower in direct provider com- munities. However, mixed and traditional communities had better PNC coverage when com- pared with indigenous women nationally (81 percent). First trimester care was lower for all providers compared to the national average (51 percent) and to indigenous women nation- ally (38 percent). Tetanus toxoid coverage was similar among women who received PNC in mixed provider communities compared with the national average (64 percent) and better than for indigenous women nationally (60 percent). Context and Caveats Similar to the findings of Mills et al. (2004), interpreting the findings of contracting mod- els requires clarity regarding the contextual conditions related to model features, provider and location selection, and other factors affecting supply and demand. Provider selection, location, and population coverage varied across models. First, providers were not assigned randomly to the catchment areas. Direct service providers were the MSPAS's model of choice for communities located farther from health services and were Health System Innovations in Central America 81 assigned a priori to more difficult situations in terms of distance and road access.83 In most cases, the ADMSS catchment areas were chosen by MSPAS district managers in consultation with the rural health technicians assigned to ADMSS teams. According to SIAS authorities, there was a tendency to choose catchment areas with which the rural health technicians were somewhat familiar and where they had easy access to towns where many resided. Second, the catchment areas of the direct providers were on average geographically larger and more dis- persed, and they contained more population than those assigned to other providers. The gov- ernment decided it could not scale up its provider network in terms of staffing, purchasing and distribution of drugs and supplies, transportation, and training as rapidly as it could through direct contracting of service provision to NGOs (PSSs). Third, the PSSs had to start from scratch in terms of recruiting and hiring personnel, training, and knowledge of MSPAS norms and procedures. Because the ADMSS and HP models used MSPAS personnel with ample experience in rural health delivery who were clearly knowledgeable about the MSPAS system, these providers had a clear advantage, at least at the outset. Finally, there were no baseline measures to assess the degree of improvement that resulted from service provision via the new models. It is likely that the more distant, isolated communities served by direct provider NGOs had less access, less coverage, and consequently worse baseline measures at onset than the ADMSS catchment areas located near main roads and towns. The intrinsic differences in the delivery models and team composition contributed to differences in performance. The HPs applied a facility-based model rather than the outreach- based model of the other providers. Production was highly dependent on the staffing of highly motivated volunteer physicians from Cuba. The HPs performed admirably under the conditions of regular supplies and use of volunteer physicians. It is important that, even accounting for the economic costs of volunteer physicians, HPs were the least-cost provider of the group and provided PBS coverage similar to that of the PSSs. Low production was related to a relatively limited number of assigned staff (1.5 per HP) and evidence from the household survey that many residents in the catchment area had access to providers else- where. Few outreach activities were performed (except for vaccinations), thus the HPs did not benefit from community facilitators who served as a liaison between delivery teams and the communities for the mixed and direct providers. Though the HPs made use of volunteer community health workers and TBAs, the latter were not paid a stipend, as was the case for their counterparts working with PSS and ADMSS providers. Again, the stipend was an input applied to the outreach model and meant to cover expenses for travel among dispersed rural hamlets. HPs without physicians displayed very low levels of production and productivity. Given the inability of the MSPAS to recruit and maintain Guatemalan physicians, filling the gap with volunteer foreign doctors was a short-term, but probably unsustainable, solution. One of the objectives of the household study was to assess the demand for care from populations residing in the catchment areas of each of the three delivery models. Resi- dents of traditional and mixed provider catchments had choices unavailable to residents of direct provider catchment areas. The results demonstrate a considerable amount of cross-utilization of services such that performance in mixed and traditional provider communities cannot necessarily be attributed solely to the assigned provider. Women 83. Although we tried to control for the greater difficulties in obtaining services at greater distances by controlling for this in our regression analysis, there may be other factors associated with access besides distance that we did not control for. 82 World Bank Working Paper from traditional provider (HP) catchment areas often used MSPAS health centers as well as direct and mixed provider-operated community centers to which they have access. Com- munity centers, where care is provided by either direct or mixed providers, are not generally located near traditional provider areas, and women had to actively seek them out during the monthly visits by the itinerant teams. Likewise, women in mixed provider communities demanded care from both HPs and health centers to which they had access. As noted, the location of mixed model (ADMSS) community centers near main roads and towns facilitates use of a range of providers. As expected, women residing in PSS catchment areas displayed significantly greater use of PSS-operated community centers. Travel time and distance pro- hibits easy access to other providers. In short, they were less likely to make the effort to go to an HP or health center to fill in needs between monthly provider visits. Limitations The research faced several limitations that impeded any definitive statement on identifying the superior delivery model. The data were cross-sectional, and there were no before and after data available. Although the communities appeared similar on many counts and we tried to control for important differences, we were limited by the variables collected. There may be other explanations for the differences observed that we did not measure. These include basic health determinants such as income and education, and determinants of quality such as the training of the provider (physician, nurse, and so on). It is worth repeating that the substantial cross-utilization of services by the women from the communities studied makes interpretation of the findings difficult, particularly for mixed and traditional providers. Results for communities from one model's catchment area cannot be attributed solely to the assigned provider. The results are likely to be diluted, with differ- ences tending toward nil. However, these operational research findings reflect the reality of most communities and consumers of health care, and they should be interpreted in this light. Conclusions and Policy Implications The findings are important for situations like Guatemala's, where there is a large population without coverage and a commitment to expanding basic health services quickly and apply- ing the financing to do so. The SIAS model is limited by the types of services it can provide. For example, it is not an appropriate model for curative services of acute illnesses, given that visits are made only once a month. However, its focus on preventive services is attractive because of the greater cost-effectiveness and benefits of many preventive services and the assurance such a model could provide that services are reaching dispersed or marginalized populations. All providers produced a low volume of PBS priority interventions except for vaccinations, which represented about 40 percent of all outputs for each provider84 while accounting for about 13 percent of costs. Clearly, a greater effort is needed to strengthen pro- vision of other PBS interventions across all providers. The overall low volume of outputs observed for other PBS and priority interventions may relate to the weak monitoring and oversight by the MSPAS.85 84. In 2001, the MSPAS placed high priority on vaccination coverage. 85. Monitoring by the MSPAS was more informal than systematic, as described in chapter I in this volume. Health System Innovations in Central America 83 Nevertheless, extension through NGO contracting is expensive. The MSPAS reported spending US$61 million on contracts with direct and mixed providers between 1998 and 2002. Accounting for overhead costs, coverage extension represented approximately 13 per- cent of the MSPAS budget in 2002. Supported by two loans from the Inter-American Devel- opment Bank, as well as grants from several bilateral agencies, financing has been sustained for more than eight years. Financial commitment of this magnitude may not be forthcom- ing in the future. Direct providers are the most productive and expensive, displaying adequate levels of cov- eragebuthighlevelsofusersatisfaction.Thehigherproductivityandusersatisfactionobserved in Guatemala is consistent with NGO-contracting experiences elsewhere (Loevinsohn and Harding 2004). The evidence suggests that the mixed and traditional models are also effective models yet display lower costs. However, they are unproven alternatives under the conditions of the catchment areas where direct providers operated. Traditional providers were also highly dependent on volunteer physicians. In the past, the MSPAS has had difficulties in recruiting, managing, and retaining physicians for rural health services. Whether mixed and traditional models could extend coverage in distant areas more efficiently or with equal or higher levels of coverage and output than direct providers remains untested. With the exception of immunizations, only a small portion of some priority PBS inter- ventions is actually being delivered by the three models. The lack of provision of growth mon- itoring is striking, particularly given the high prevalence of chronic (70 percent) and global (30 percent) malnutrition among indigenous children younger than five years old (ENSMI 2002). Growth monitoring is an important primary care activity to identify infants and chil- dren at risk. It is not clear how much of the environmental health part of the PBS is provided by community health workers and rural health technicians. There is very limited provision of the "care of illnesses and emergencies" part of the package, although with proper supervision and training, diseases such as malaria and tuberculosis could be identified and treated in the community.Nevertheless,mostofthediseaseslistedunder"illnessesandemergencies,"aswell as acute diarrheal and respiratory illnesses among infants and children, often require imme- diate attention and cannot await the appearance of the outreach team. Obtaining better out- comesfortheseillnesseswillrequireimprovementsinthereferralandcounter-referralsystems between communities and health services. This is a challenge for any type of outreach model. Our findings are different from those in other studies in Cambodia and Bangladesh, which showed clear outcome improvements in using NGO services to provide basic pack- ages of care over those provided by traditional providers (Bhushan, Keller, and Schwartz 2002; Mahmud 2002). There are many possible explanations, some of which have already been noted in this chapter: this is not a randomized trial; we are comparing findings at the end of a period of service provision and not before and after; and direct provider NGOs were specifically selected to provide care in an area where the government did not think it could rapidly scale up services. In addition, compared to the two countries where other studies have been carried out, Guatemala has a relatively well-developed and relatively long-standing health care delivery system (as do most countries in Latin America). Guatemala spends more money on health care, has a higher number of health professionals, and had much better baseline coverage rates for immunization and PNC. For instance, the national immunization and PNC coverage rates for Cambodia before the intervention were 39 percent and 21 per- cent, respectively. The best rates achieved after the intervention were 40 percent. This com- 84 World Bank Working Paper pares to national baseline coverage in Guatemala (ENSMI 1998­99) of 60 percent for immunization and 58 percent for PNC. The traditional health services generally perform better in Latin America and the Caribbean than in other parts of the world (better immu- nization rates, higher PNC coverage, lower infant mortality, and so forth) so the bar is set higher for alternative models. Nevertheless, inequities in Latin America and the Caribbean are among the largest in the world, particularly in Guatemala. The added value that mixed providers supplied was clearly observed, and direct provider NGOs did nearly as well as the other providers in areas with difficult access where the MSPAS had little or no historical presence. Thus, our study suggests that contracting out services in areas and for population subgroups whose health outcomes lag behind those of the general population is an effective strategy to reduce inequities. Annex 2.1 Annex 2.2 Vaccinations: Problems in the Distribution Chain The comparatively low performance of PSSs in Quiché and Alta Verapaz (before controlling for distance) related to vaccination coverage merits special attention. Guatemala used two approaches for its vaccination programs. One, a vertical approach used a centralized, top- down system of determining total demand for various vaccines for pregnant women, infants, and young children. The MSPAS conducted three to four annual vaccination campaigns, which were run centrally. Under this National Immunization Program, the MSPAS was responsible for purchasing--on a large scale and a few times a year--all the vaccines required for that year. It developed its requirements in part based on information provided by the area directors, who had some latitude to modify their requirements based on local circumstances. These vaccines were in turn delivered to each district. Each district, in a cascading pattern, provided vaccines to the local health centers, which then passed on the required quantities to HPs. Districts were also responsible for providing vaccines to PSS and ADMSS providers according to their projected requirements. District administrators, however, disputed the value of the quantities requested by the PSSs. They contended that there were many un- accounted for losses of vaccines and that it was difficult to monitor their appropriate use. PSSs claimed that the district administrators favored the MSPAS teams, whether affiliated with ADMSSs or HPs, and did not regularly communicate to the PSSs when supplies were Annex Table 2.1.1. Number of Providers: National, Departments, and Sample, 2001­02 Provider type National In 4 sampled departments Sample PSSa 154b 42 21 ADMSSa 24 12 HP 1,289 204 22 Total 1,443 270 55 a. Number of convenios (several NGOs had more than one contract). b. Number of signed convenios among 89 NGOs (approximately 60 percent were PSSs). Source: MSPAS-SIAS; provider survey. Health System Innovations in Central America 85 Annex Table 2.1.2. Odds Ratios for Receiving Elements of the PBS, by Type of Provider (Crude Data) Crude OR Indicatora MSPAS ADMSS PSS Children aged 12­59 months with vaccineb TARJETA (P33) 1.00 1.04 0.84* BCG 1.00 1.81* 0.70* DPT3 1.00 2.02* 0.73* Polio3 1.00 1.94* 0.74* Measles 1.00 1.90* 0.74* All vaccines 1.00 1.93* 0.75* Women with history of pregnanciesc At least 1 PNC visit 1.00 2.02* 0.80 First trimester PNC visit 1.00 1.26* 0.79* Received tetanus toxoid 1.00 1.27* 0.72* Received iron supplementation 1.00 1.55* 0.78* Received folate supplementation 1.00 1.51* 0.80* CHILDREN YOUNGER THAN 2 YEARS Prevalence of diarrhea in the last 3 months 1.00 1.39* 0.95 Diarrhea and received ORS 1.00 1.39* 0.89 Diarrhea and received medical care 1.00 1.32* 0.80 Prevalence of respiratory illness in the last 3 months 1.00 0.56* 1.25* Respiratory illness and received medical care 1.00 1.36 0.70 Growth checkup in the last 3 months 1.00 1.01 2.24* Weighed during checkup in the last 3 months 1.00 0.20* 3.02 Height measured during checkup in the last 3 months 1.00 0.56 1.93 WOMEN WITH CHILDREN YOUNGER THAN 2 YEARS Say they know how to use ORS 1.00 1.53* 0.89 Have ORS in the home 1.00 1.68* 0.75* a. Adjusted for distance from health center and community population. b. Information from carnet or from the mother when carnet was not available. c. Women who have given birth since July 1999. *p < .05. available. Further, some of the areas served by these facilities were remote and without elec- tricity, making refrigeration and proper storage of these vaccines almost impossible. Concurrently, a horizontal approach to vaccinations was used, by both the ADMSSs and the PSSs, and to a lesser extent by the health centers and HPs. In this approach, vaccines were provided to mothers, infants, and children throughout the year as part of PNC and well-baby and child monitoring. However, as a result of logistical problems, particularly in the case of the PSSs (mostly about ensuring the quality of vaccines held for long periods), it appears that the campaign-based vertical approach remained predominant. Again, in interviews PSS rep- resentatives claim that communication regarding the arrival of vaccines was inadequate. 86 World Bank Annex Table 2.1.3 Average Monthly Production and Spending,a by Provider and Service Type, 2001 Working Direct (N = 21) Indirect (N = 12) Traditional (N = 22) Category and Unadjusted Adjusted Unadjusted Adjusted Unadjusted Adjusted Paper type of expenditures expenditurese expenditures expenditurese expenditures expenditurese intervetion Outputs (Q) (Q) Outputs (Q) (Q) Outputs (Q) (Q) PBS 24,391 919,271 932,642 12,234 323,579 347,128 13,506 170,985 341,012 Non-PBS 4,038 212,808 215,903 1,997 74,908 80,359 3,842 35,884 71,567 ARI care 2,612 129,896 131,786 1,314 45,723 49,050 1,273 22,452 44,778 Diarrhea care 927 94,020 95,387 235 33,095 35,503 427 16,284 32,476 PNCb 1,010 194,182 197,006 352 68,351 73,216 389 36,131 72,059 Well-baby checkupsc 2,658 3,058 3,103 218 1,076 1,155 594 466 930 Vaccinationsd 10,784 150,832 153,026 5,471 53,093 56,956 7,500 27,674 55,193 Other PBS services 6,400 347,284 352,335 4,616 122,242 131,139 3,322 67,989 135,576 All 28,429 1,132,079 1,148,545 14,231 398,487 427,487 17,348 206,869 412,578 Note: US$1 = Q7.9. a. Spending includes direct, overhead, and MSPAS administrative costs. Direct costs are weighted according to weights presented in table 2.3, except for the "all" category. b. Includes spending on iron fortification. c. Growth and development monitoring of children younger than 2 years old. d. Vaccinations for children and pregnant women (tetanus toxoid). e. Costs adjusted for volunteer physicians (all providers), training, and vehicle and administrative costs (HPs). Source: provider survey. Health System Innovations in Central America 87 There were an undetermined number of instances in which providers had mobilized their communities but the supplies never arrived--or contrarily, supplies arrived but the PSSs were not informed. Bibliography Bhushan, I., S. Keller, and B. Schwartz. 2002. "Achieving the Twin Objectives of Efficiency and Equity Contracting Health Services in Cambodia." Economics and Research Department Policy Brief Series 6. Asian Development Bank, Manila. Bitrán, Ricardo. 1995. "Efficiency and Quality in the Public and Private Sectors in Senegal." Health Policy and Planning 10(3):271­83. Eichler, R., P. Auxila, and J. Pollock. 2001. "Output-Based Health Care: Paying for Per- formance in Haiti." Private Sector and Infrastructure Network Note 236. World Bank, Washington, D.C. ENSMI. 1998­99. Encuesta Nacional de Salud Materno-Infantil 1998­1999. Guatemala City. ------. 2002. Encuesta Nacional de Salud Materno-Infantil 2002. Guatemala City. Lavadenz, Fernando. 2001. "Redes públicas, descentralizadas y comunitarias de salud en Bolivia." Pan American Journal of Public Health 9(3):182­89. Levin, Ann, Mark McEuren, Vito Tanzi, Gerry Van Dyck, and Nino Sekyere-Boakye. 1999. "Costs of Maternal Health Care Services in South Kwahn District, Ghana." Special Ini- tiative Report 20. ABT Associates, Bethesda, MD. Loevinsohn, Benjamin, and April Harding. 2004. "Contracting for the Delivery of Com- munity Services: A Review of Global Experience." HNP Discussion Paper. World Bank, Washington, D.C. Mahmud, H. 2002. "Mid-Term Health Facility Survey--Urban Primary Health Care Pro- ject. Bangladesh." Mitra and Associates, Manila. Marek, T., I. Diallo, B. Ndiaye, and J. Rakotosalama. 1999. "Successful Contracting of Pre- vention Services: Fighting Malnutrition in Senegal and Madagascar." Health and Policy and Planning 14(4):382­89. Mills, A., N. Palmer, L. Gilson, D. McIntyre, H. Schneider, E. Snanovic, and H. Wadee. 2004. "The Performance of Different Models of Primary Care Provision in Southern Africa." Social Science & Medicine 59:931­43. Ministerio de Salud y Assistencia (Ministry of Health and Social Protection [MSPAS]). 2000a. "Sistema Integral de Atención a la Salud." Guatemala City. Processed. ------. 2000b. "La modernización de la salud pública en Guatemala." Guatemala City. Processed. Palmer, N. 2000. "The Use of Private-Sector Contracts for Primary Health Care: Theory, Evidence and Lessons for Low-Income and Middle-Income Countries." Bulletin of the World Health Organization 78(6):821­29. Slack, K., and W. Savedoff. 2001. "Public Purchaser-Private Provider Contracting for Health Services: Examples from Latin America and the Caribbean." Sustainable Devel- opment Department Technical Papers Series. Inter-American Development Bank, Washington, D.C. Soderlun, N., P. Mendoza-Aran, and J. Goudge, eds. 2003. The New Public/Private Mix in Health: Exploring the Changing Landscape. Geneva: The Global Forum for Health Research. CHAPTER 3 The San Miguelito Hospital Reform in Panama-- Evaluation and Lessons Ricardo Bitrán, Cecilia Má, and Paulina Gómez Abstract Located in Panama's San Miguel region, the newly opened Hospital Integrado San Miguel de Arcángel (Hospital San Miguelito, also known as HISMA), has brought several innovations to that country's government health sector. A specially created purchasing entity, CONSALUD, pools the regional financial resources of the Ministry of Health (MOH) and the Social Security Institute (SSI) and purchases, or contracts out, services from HISMA using a simple but effective prospective payment system. HISMA's board, in turn, purchases or outsources all medical and support services from outside private firms owned by doctors, other health professionals, managers, accountants, cleaning and catering services, and so on. Thus, HISMA does not have any staff other than its board and the small group of experts who monitor contractor compliance by outside providers. In 2000, one year into its operation, the MOH and the World Bank sponsored an evaluation of HISMA. This paper presents the findings from that effort. The performance of HISMA, measured in terms of technical and economic efficiency, and in terms of quality of care, was compared with that of two control hospitals run with the traditional approach of historic budgets and fixed staff. These were Chepo Hospital, property of the SSI, and Solano Hospital, property of the MOH. The evaluation found that HISMA outperforms the control hospitals in nearly every aspect: its unit costs of production are for the most part lower than those of Chepo and Solano, the productivity of its medical contractors is higher than that of the medical staff of the control hospitals, and the quality of care is considerably superior to that of the controls. Thus, the evaluation of the HISMA model is highly favorable. Unfortunately, the strong resistance seen in Panama to changing the status quo and expanding HISMA's model attests to the difficulties, faced in this country and elsewhere in Latin America, of promoting policies that favor beneficiaries in an environment dominated by interest groups. 89 90 World Bank Working Paper Introduction and Objectives The management model adopted in 2000 by Panama's Hospital Integrado San Miguel de Arcángel (Hospital San Miguelito [HISMA]) is a revolutionary public policy initiative; draw- inglessonsfromthisexperiencecanbeusefultoPanamaandtootherlow-andmiddle-income countries.LocatedinPanama'sSanMiguelregion,HISMAcombinesfourinnovativefeatures. First, in San Miguel, the Ministry of Health (MOH) and the Social Security Institute (SSI) pool their financing into CONSALUD, a public entity set up with the specific aim of enabling public payers of health care to buy services from public and private providers by means of innovative purchasing strategies. In turn, CONSALUD purchases hospital ser- vices from HISMA on behalf of the MOH and SSI, and thus pays for the hospital's financ- ing and recurrent costs.86 Through this arrangement, HISMA became the single common provider of hospital services for the beneficiary populations of these two entities in the region. HISMA's model contrasts sharply with that found in most developing countries, where institutions such as the MOH and the SSI possess separate health systems for their respective beneficiary populations. For efficiency reasons, many such countries aspire to integrate both systems. In Peru, for example, the independent operation of these institu- tions' systems implies major inefficiencies, arising from duplication and underutilization of physical and human resources. In Panama, outside the San Miguel region, the MOH and SSI continue to run their parallel health systems, duplicating and often underusing hospi- tals and ambulatory health care facilities. Second, CONSALUD purchases hospital services from HISMA on the basis of a prospective payment system, paying HISMA according to previously agreed prices for each hospital discharge, outpatient surgery, and emergency visit. Thus, HISMA receives from CONSALUD an amount of revenue that is directly linked to the types and volumes of ser- vices it provides. This financing system is in stark contrast with the system in place in most public hospitals in Latin America. Those providers typically receive a fixed budget that varies only marginally year after year, and which is only weakly linked to the volume and types of services actually produced. Third, HISMA is headed by the HISMA board, an administrative entity whose main responsibility is to purchase management, medical, and other types of services from exter- nal service companies on behalf of HISMA. Indeed, as opposed to traditional public hospi- tals, HISMA has practically no medical or administrative employees. The services HISMA needs to function, such as accounting, cleaning, diagnostic procedures, and the diverse medical care services, are not produced by internal personnel, but are instead contracted out by HISMA to third parties. These are external private companies, made up of medical and nonmedical professionals that supply HISMA with the services it needs. This out- sourcing model solves a common problem faced by health ministries and social security institutions in Latin America: excess personnel that cannot be fired and personnel needs that cannot be met because of restrictive civil service and other laws. Fourth, HISMA comanages a network of referrals and counter-referrals with the health centers and clinics in its catchment area. This too is an innovation in the regional context, 86. To date, CONSALUD has been used only in the context of the HISMA initiative. Recently there have been proposals to contract out primary health care services through CONSALUD, but they have not materialized. Health System Innovations in Central America 91 where formal referrals between ambulatory facilities and hospitals in the public sector are rare or lacking altogether. Each of these innovations deserves a performance analysis--and so much the more, all of them together--because they constitute a unique experience. This chapter presents the findings of an assessment of HISMA's experience at the end of its first year of opera- tions. The assessment was part of a more comprehensive reform program financed by Panama's MOH and a loan from the World Bank. The chapter is organized as follows. The second section presents a brief review of the lit- erature on the issue of hospital reforms and contracting for health services in the public sec- tor of developing countries. The third section contributes further information on the genesis of the CONSALUD-HISMA project and on this model's characteristics. The fourth section describes the study's methodology. The fifth section presents the main findings, and the sixth section proposes public policy conclusions. All the information presented on the HISMA model's assessment has been drawn from Bitrán and others (2001) unless otherwise indicated. International Experience The HISMA model comprises three distinct innovations addressed in the literature on health sector reform. The first, hospital autonomization or corporatization, is the policy by which public hospitals are granted greater independence from a central government agency (a ministry of health or a social security institute) and can therefore exert more con- trol over matters such as hiring and firing of health personnel and purchasing and selling of other production inputs and medical services. The second, purchaser-provider split, denotes the policy by which the payer and provider of health services, traditionally merged as one sole entity in the public sector, are separated into two different entities, giving rise to a contract between purchaser and provider, and a payment system. The third, contract- ing out, denotes two possible situations: one where the public payer, operating under a purchaser-provider split arrangement, decides to purchase hospital or primary care ser- vices from a private provider; and another, outsourcing, where a provider of health ser- vices seeking to improve economic efficiency chooses to buy some of the services it needs from outside sources, rather than produce them internally. In developing countries, the reform of public hospitals lies at the center of the public policy debate on health. In their review of experiences in hospital autonomization, Govin- daraj and Chawla (1996) identified two main justifications for this policy. First, public hos- pitals consume an important part of government health budgets. This limits government's ability to reallocate additional financial resources to pay for other services, such as primary care and preventive maternity and baby care, which are more cost effective than most hos- pital care and for which there is unmet demand. Second, there is concrete evidence that public hospitals in these countries operate at low efficiency levels, therefore a significant part of their expenses constitutes a waste of scarce public resources. These authors' arguments coincide to some extent with those presented by Over and Watanabe (2003). The latter characterize the political context in which the debate on hos- pital reform takes place (first column of Table 3.1) and recommend politically sensitive areas that should be tracked through monitoring so as to evaluate hospital performance (second column). 92 World Bank Working Paper Table 3.1. Hospital Reform in Developing Countries: Political Context and Politically Sensitive Issues to Be Tracked through Monitoring and Evaluation Politically sensitive issues that any Four main contextual facts in the political context hospital monitoring and evalua surrounding the debate about hospital reform system should track 1 Government hospitals are generating large budget Financial performance of hospitals deficits that governments cannot or are no longer willing to afford. 2 The population, particularly the vulnerable, fear Access to and out-of-pocket payments for that hospital reform will result in constrained hospital care, particularly for the vulnera- access to care--a major departure from ble populations constitutional guarantees of free health care for all or from the tradition of universal provision of hospital services at no direct charge to patients. 3 Reform opponents fear that reform will hamper Fulfillment of public duties (teaching, the ability of public hospitals to perform various research) activities beyond direct patient care, such as research and teaching, which benefit the entire health care system, not just hospital patients. 4 Medical unions are typically influential reform Hospital personnel management (number opponents because they anticipate that the and types of posts, salary levels, other staff reform will result in job losses in public hospitals. benefits) Source: Adapted by the authors from Over and Watanabe (2003). In a book published by the World Bank, Preker and Harding (2003) analyze in detail the situation of public hospitals in developing countries and examine a possible solution referred to as hospital corporatization. Corporatization embraces a group of organizational reform options for public hospitals and constitutes an intermediate alternative between two extremes: on the one hand, the traditional model of public ownership operating in a monopoly situation and, on the other hand, the privatization of public hospitals, operat- ing under private management in a competitive environment. Outsourcing of support services is one of several possible strategies that public hospi- tals can adopt under corporatization. The United States and the United Kingdom have the greatest experience in contracting out services for public hospitals. An assessment of these initiatives by Mills and Bloomberg (1998) indicates that for services such as cleaning and catering, savings are between 20 percent and 30 percent. Yet it appears that these savings dwindle with the passing of time, when the competitive environment that prevailed dur- ing the initial public tender disappears. With respect to the contracting out of clinical ser- vices, an assessment is more difficult. Limited information on such contracting in the United States points toward the establishment of long-term relations based on trust, an ensuing decline of competitive tenders, and the decreasing importance of prices after the first round of contracting. Initial contracting is competitive, but renegotiation is not. Infor- mation from the United Kingdom points in the same direction. In their review of the literature on contracting experiences, Mills and Bloomberg (1998) conclude that the introduction of market mechanisms such as outsourcing will become increasingly important in the sphere of public health services. They state that there is a lack Health System Innovations in Central America 93 of information on the concrete impact of outsourcing on costs, efficiency, quality, and other dimensions of performance, and they point out that the literature has concentrated more on the promotion of contracting policies than on examining them in detail first. Despite the existence of successful experiences in contracting, Mills and Bloomberg note that valid crit- icisms persist of the possible impacts of such policies--for example, in terms of health sys- tem equity. Finally, they also note that the existing literature contains little discussion of how to carry out an assessment of these initiatives. The authors consider that the key variable in evaluating contracting experiences in public hospitals is the marginal cost of contracted ser- vices and its comparison to the marginal cost of direct production. However, they recognize that the measurement of marginal costs in hospitals constitutes a methodological challenge. In developing countries, the evidence is still scanty and assessments incipient. However, Govindaraj and Chawla (1996) conclude that the outcomes obtained to date, although scarce and preliminary, are not encouraging: Success with autonomy in public sector hospitals in developing countries has been limited, and there have been few gains in terms of efficiency, quality of care, and public accountability. . . . it is as much the confused and erroneous ideas of autonomy, as the poor implementation of the autonomy measures, which have been responsible for the relative lack of success of the autonomy initiative. (p. 35) In recent years, the contracting out of primary health care from private providers has become widespread in the developing world. Loevinsohn (2001) documented the experi- ence of Cambodia, a country that developed a large-scale pilot plan to assess the advan- tages of different modalities for the contracting out of primary care. The preliminary lessons drawn from this experience suggest that rewarding providers according to results and management autonomy, based on a clearly defined contract with explicit service deliv- ery targets, is an efficient tool for increasing efficiency, effectiveness, and quality. Eichler, Auxila, and Pollack (2001) reported on their findings from a study about the contracting out of primary health care by nongovernmental organizations (NGOs) in Haiti. The Haitian experience is important because it exposes problems arising from inad- equate contract design. For example, if the providers are funded to pay for operating costs, they have no incentives to improve productivity, management, and operations, or to increase coverage and enhance quality. Eichler and her colleagues demonstrate that mod- est economic incentives, equal to approximately 5 percent of the NGO contracts, can exert a significant and positive impact on the behavior and performance of these entities. Fur- thermore, the authors warn against "beginners' mistakes." For example, these include not specifying in the contract what will happen if the specified goals are not attained, or allow- ing the creation of monopolies through contracts handing over geographic zones to only one NGO for several years. Finally, they point out the minimum conditions for contract- ing out, such as knowing the beneficiary population and ensuring that the indicators deter- mining payment are relevant, verifiable, and under the NGO's control. Genesis of HISMA and Description of the Model In 1996, the government of Panama started the Health Sector Reform Project, the first component of which established a new management, organizational, and service provision 94 World Bank Working Paper Box 3.1. Outsourcing of Services by HISMA HISMA does not produce services of any kind, whether administrative, medical, or support. All these services are purchased by the HISMA board from third parties. In order to do this, it has established controlled purchase contracts with payment conditioned on the outcomes of periodic performance assessments of each provider contracted. The services contracted out by HISMA are listed below. Services Contracted Out by HISMA Final medical services Intermediary medical General services (purchased from 6 companies services (purchased (purchased from of doctors and nurses) from 5 companies) 3 companies) Outpatient visits Laboratory Nutrition Inpatient Diagnostic imaging Dietitian Vaginal and caesarean deliveries Special procedures Accounting Surgical interventions Pharmacya Cleaning Emergency servicesb Security a. The pharmacy service was assumed by the HISMA board as of April 2000. b. The general emergencies service was assumed by the HISMA board as of January 2001. model in the San Miguel Health Region. The project's objectives were to modernize hos- pital management, improve integration of the public network of health care providers in the region (HISMA and ambulatory health care facilities in its catchment area), and over- come efficiency and service quality problems. A strategic alliance between the SSI and the MOH was established as part of the model to finance hospital services, and the necessary legal entities were created for purchasing (CONSALUD, a private institution) and provid- ing (the HISMA board). As already noted, a central element in the new hospital management model adopted by HISMA is contracting out or outsourcing intermediary and final services (see Box 3.1 and Figure 3.1). Every three years, HISMA launches a competitive bidding process to con- tract out clinical and ancillary services. Within HISMA, the entity responsible for manag- ing external contracts is the HISMA board. Box 3.1 describes the services contracted out by the HISMA board. Through external contracting, HISMA embarked on an operating regime wherein its service providers were financed according to their real production, as was HISMA when it received financing from CONSALUD according to its production. HISMA began operations in December 1998. For the first time in Panama, MOH health services were integrated with those of the SSI. Also for the first time, a public hos- pital in Panama adopted referral and counter-referral systems in coordination with the ambulatory health care facilities under its jurisdiction. In 1999, the MOH and SSI endowed CONSALUD with US$13,500,000.87 CONSALUD finances HISMA through a regulatory contract controlling the quantity, costs, and quality 87. Panama's official currency is the balboa (B). The exchange rate in Panama during the period involving the assessment (2000) was (and continues to be as of April 2005) B 1 to US$1. All figures in this chapter are in 2002 dollars. Figure 3.1. The HISMA Model Social Security Institute Ministry of Health Financing of Financing of CONSALUD through CONSALUD through regular budget regular budget transfers transfers CONSALUD Financing of HISMA Foundation through payment based on Basic Health Service Units, including (a) Fee-for-service (FFS) hospital discharges, (b) purchasing of clinical ambulatory surgeries, and (c) and support services emergencies from external service System providers Private producers of final clinical services Service provision Innovations HISMA Board Private producers of intermediate clinical services in Central San Miguel de Arcángel Private producers of (HISMA) Hospital general services America 95 96 World Bank Working Paper of the services that HISMA produces for CONSALUD. The latter entity finances its own expenses through the interest on endowments from the MOH and SSI. These amounted to roughly US$100,000 in 1999. CONSALUD paid HISMA prospectively on the basis of three broad categories of ser- vices: hospital discharges or hospitalizations, outpatient surgeries, and emergency visits. Prospective payment means that the price paid by CONSALUD to HISMA was established prior to provision on the basis of cost studies that yielded the following prices: US$319 per hospital discharge, US$199 per outpatient surgery, and US$40 per emergency visit. To cap spending, CONSALUD established annual production ceilings for each service category, as follows: 21,630 hospitalizations, 3,996 outpatient surgeries and 85,512 emer- gency consultations. In addition, to finance fixed costs, CONSALUD paid the HISMA board US$700,000 per month. When production ceilings are trespassed, CONSALUD still reimburses HISMA but on the basis of a lower unit cost to limit incentives for HISMA to induce excessive demand. Table 3.2. HISMA Production Ceilings for 1999 (US$) Output Output ceilings Cost Global budget Hospitalizations 21,630 319 6,899,970 Outpatient surgeries 3,996 199 795,204 Emergency visits 85,512 40 3,420,480 Total -- -- 11,115,654 -- Output ceilings were defined separately for the three categories of output, but not for the total. Source: Contract signed between CONSALUD and HISMA (1999). CONSALUD audits the bills submitted by HISMA before authorizing payments. What is more, the HISMA board's purchases require the previous authorization of the Comp- troller of the Republic. The external companies contracted by the HISMA board started their contracts in October 1998; by 2001, only the nutrition companies had been replaced. Payment of these companies was conditional on a quality assessment of the services delivered according to two criteria: technical quality and user satisfaction. HISMA hired experts to carry out med- ical and other service audits. These auditors were responsible for conducting periodic assessments of compliance by outside providers with previously defined service specifica- tions, including health care treatment protocols. User satisfaction was measured by means of monthly surveys of the users of the various services under outsourcing. If the assess- ments showed noncompliance with service specifications or user dissatisfaction, then pay- ments to the contractor were withheld or reduced. Methodology Two hospitals similar to HISMA--Chepo Hospital, owned by the SSI, and Nicolás Solano Hospital, the property of the MOH--were selected as a basis of comparison to examine HISMA's relative performance (see Table 3.3 and Figure 3.2). Both hospitals operate under Health System Innovations in Central America 97 Table 3.3. General Characteristics of HISMA, Chepo, and Solano Hospitals (1999) Characteristic HISMA Chepo Solano Population in 497,000 91,786 252,465 catchment area Beds 216 87 inpatient 172 inpatient (24 observation) (26 observation) Doctors 107 67 Nurses 98 55 Nurse assistants 79 62 Doctor hours 816 511 460 Nurse hours 778 520 440 Hospital discharges 12,755 5,270 8,938 Bed occupancy rate 48 72 68 Average length of 3.2 4.0 4.7 stay in days Clinical services Medicine Medicine Medicine Surgery Surgery Surgery OB/GYN OB/GYN OB/GYN Pediatrics Pediatrics Pediatrics Dentistry Dentistry Medical specialties General surgery Internal medicine Internal medicine Cardiology Cardiology Trauma Trauma Orthopedics Orthopedics Orthopedics Ophthalmology Ophthalmology Ophthalmology Ear, nose, and throat Ear, nose, and throat Eye, ear, and nose Maxillofacial surgery Maxillofacial surgery Pneumology Pneumology Psychiatry Psychiatry Urology Plastic surgery Hematology Oncology Dermatology Allergy-immunology Gastroenterology Physiatry Occupational health OB/GYN obstetrics and gynecology. 98 World Figure 3.2. Health Care Networks, HISMA, and Control Hospitals Bank care Working of Santo National referral ArnulfoArias Children's Psychiatric Tomás Cancer Hospital Complex Hospital Hospital Hospital Hospital level (SSI) (MOH) (MOH) Paper (MOH) (MOH) hospitals) Third (national Referral and counter-referral system San Miguel Health Eastern Panama Western Panama Region Health Region Health Region care of HISMA hospitals) Chepo level Hospital San NicolásSolano Regional Miguel Arcángel Hospital (MOH) Hospital (SSI) (MOH and SSI) Second (regional 5 health care 2 polyclinics 3 polyclinics health centers of 2 local 9 health 2 local 8 health 11 sub- primary care centers 1 polyclinic primary care centers health units 1 polyclinic (SSI) units (MOH) level centers centers) ("ULAPS") (MOH) ("ULAPS") 33 health (MOH) (SSI) posts (MOH) First (ambulatory Health System Innovations in Central America 99 the traditional management model common to all public hospitals in Panama. This means that they are subject to historical, supply-side budgets that vary only marginally from year to year and are not adjusted to reflect output; they have medical and other staff on their pay- roll, who in practice they cannot fire or motivate financially to improve quality or produc- tivity, and who produce internally all of the medical services the hospital delivers as well as the nonmedical services it consumes (for example, accounting, laundry, nutrition). All three hospitals deliver more or less the same kinds of health services to similar populations, and they are also comparable in terms of complexity and service mix. Small differences in the medical specialties available in the hospitals were deemed marginal in this comparative analysis of performance: most of the specialties that differ from hospital to hospital are inter- nal medicine subspecialties that do not involve inpatient care. The presence of an oncolo- gist at Solano Hospital could constitute a major difference in service mix, although tumors do not constitute one of the top 10 discharge categories. The three hospitals differ in size, and this was taken into account in the analysis. HISMA is the largest of the three hospitals, with 216 beds and 6 operating rooms; Solano Hospital is second in size, with 172 beds and 5 operating rooms; Chepo Hospital is the smallest, with only 87 beds and 2 operating rooms. Other things being equal, HISMA's larger size could result in relatively lower average costs, because it could spread its fixed costs over a larger volume of output than Chepo or Solano. The study methodology was a comparative analysis between HISMA's performance and that of the two control hospitals. Primary information was collected for the analysis by means of surveys, interviews, and data gathered in the three hospitals as well as in the ambulatory facilities in their catchment area. Several secondary sources of information were also resorted to. This research addressed two main questions: (1) Does HISMA out- perform the other hospitals? (2) If so, why? To answer the first question, the researchers constructed the performance indicators shown in Table 3.4 and compared them among the three hospitals. These indicators were grouped into three major categories: output, efficiency, and quality of care. Output indi- cators comprise measures of the volume of production of inpatient and outpatient services in the hospitals (for example, annual number of surgical interventions performed). Effi- ciency indicators relate output data with information on the resources required to produce that output (for example, annual number of hospital discharges per bed). This research assessed three dimensions of quality. Clinical management-related qual- ity was assessed using several indicators, applying an accreditation guideline, applying a user satisfaction survey, and measuring waiting times and rejection rates. Technical quality is gen- erally measured with two indicators: nosocomial infection rate (NIR) and percentage of com- plications; however, this was not possible because only HISMA kept NIR records for 1999 and none of the three hospitals registered readmissions and reinterventions for that year. Quality in terms of user satisfaction was assessed through surveys to explore general stan- dards of care, access, and timeliness; availability of drugs; staff manners with patients; com- fort and hotel amenities; medical quality tracers; and the patient's overall perception of care. Findings Study findings are presented separately for the three performance indicator groups--output, efficiency, and quality. 100 World Bank Working Paper Table 3.4. Selected Performance Indicators Output Efficiency Quality Discharges Discharges per bed Average length of stay Emergency visits Bed occupancy rate % caesarean of total births Outpatient visits Bed turnover rate % live births of total births Surgical interventions Major surgical interventions Emergency surgery/planned Emergency visits per per operating room surgery index discharge Average length of stay in days Death rate Outpatient visits per Ancillary tests per discharge Compliance with accreditation discharge Inpatient ancillary tests per standards for Surgical interventions discharge management per discharge Imaging exams per Compliance with accreditation discharge standards for support services Drug prescriptions per Compliance with accreditation discharge standards for clinical services Cost per outpatient visit Cost per OB/GYN visit Cost per pediatric visit Cost per surgical visit Cost per surgical discharge Cost per medical discharge Cost per OB/GYN discharge Cost per pediatric discharge Cost per vaginal delivery Cost per caesarean delivery Output HISMA's per capita hospitalization rate of 2.6 percent was the lowest of the three hospi- tals, a finding that may be explained by HISMA's relatively short life (see output perfor- mance indicators in table 3.5). A comparison of per capita outpatient visit rates among the Table 3.5. Summary of Output Performance Indicators Output indicators HISMA Chepo Hospital Solano Hospital Discharges 12,755 5,270 8,938 Discharges per hundred population 2.6 5.7 3.5 Emergency visits 112,539 16,805 64,352 Emergency visits per person per year 0.22 0.18 0.25 Outpatient visits 20,652 26,209 17,090 Surgical interventions 6.927 2,256 3,506 Emergency visits per discharge 8.8 3.2 7.2 Outpatient visits per discharge 1.6 5.0 1.9 Surgical interventions per discharge 0.5 0.4 0.4 Health System Innovations in Central America 101 three hospitals is not appropriate because of differences in their ambulatory output struc- ture. In particular, Chepo Hospital has a first-class outpatient clinic. HISMA's emergency visit rate was 9 percent higher than that of Solano Hospital and 30 percent above that of Chepo Hospital. The first-year emergency visit target set for HISMA was 85,521--therefore, the actual number, 112,539, was nearly one-third higher. Although it is not entirely out of line with the rates observed in the other two hospitals, it was unclear to the researchers why such a high rate took place. HISMA's output structure was more surgery intensive than that of the reference hospitals, probably because of HISMA's greater ability to perform certain kinds of outpatient surgical procedures. Efficiency Technical efficiency. HISMA's doctor and nurse productivity were the highest of the three hospitals for emergency visits and surgeries and the second highest for discharges (see Table 3.6). In contrast, HISMA's total doctor productivity for outpatient visits was the lowest, possibly reflecting HISMA's mandated focus on inpatient care. In fact, HISMA does not provide outpatient visits on demand, but only to patients before or after their hospitalization. Table 3.6. Physician and Nurse Productivity Output-to-input ratios HISMA Chepo Hospital Solano Hospital Discharges per physician hour 15.6 11.5 17.5 Discharges per nurse hour 16.4 12.0 17.2 Emergency visits per physician hour 138.0 37.0 126.0 Outpatient visits per physician hour 25.0 57.0 33.0 Total visits per physician hour 163.0 94.0 159.0 HISMA generally exhibited higher technical efficiency, as measured by output-to-input ratios, than the control hospitals (Table 3.7). Yet at the same time, HISMA's bed occupancy rate was well below that of the control hospitals, partly because of HISMA's younger age (see above) and partly also because of this hospital's relatively shorter hospital stays of 3.2 days (versus 4.0 in Chepo and 4.7 in Solano). Economic efficiency. Economic efficiency at the three hospitals was measured on the basis of unit costs (see Table 3.8).88 These were computed as the sum of direct and indirect costs, using a step-down approach to allocate the latter to the various output categories. Direct costs included labor, drugs, surgical materials, and other inputs. Indirect costs included management, utilities, and various general services. Fixed asset depreciation was not included in the unit cost calculation because no accurate information on the market value 88. As can be seen from Table 3.8, production categories were broadly defined, and this could introduce some ambiguity into the interpretation of results as regards efficiency comparisons. Future comparisons of unit costs of HISMA and control hospitals should try to adjust results for case mix. 102 World Bank Working Paper Table 3.7. Technical Efficiency Performance Indicators (number of units) Performance indicators HISMA Chepo Hospital Solano Hospital Discharges per bed 59 56 52 Bed occupancy rate 48 72 65 Bed turnover rate 68 56 52 Major surgical interventions per operating room 1,154 1,110 560 Average length of stay in days 3.2 4.0 4.7 Ancillary tests per discharge 14.9 143.6 51.0 Inpatient ancillary tests per discharge 4.7 38.5 21.0 Imaging exams per discharge 2.4 11.9 1.6 Drug prescriptions per discharge 26.0 2.2 25.2 of fixed assets could be obtained from Chepo and Solano Hospitals. HISMA's unit cost of general outpatient visits was by far the highest of the three hospitals, reflecting HISMA's lower relative productivity and its lack of emphasis on general outpatient care (see above). Much of the difference in costs was ascribed to HISMA's higher spending on medicines. In contrast, the unit cost of obstetrical and gynecological (OB/GYN) visits was lower in HISMA that in Chepo Hospital, and HISMA's unit cost of pediatric visits, at US$30, was similar to Chepo's. Table 3.8. Economic Efficiency Performance Indicators (US$) Performance indicators HISMA Chepo Hospital Solano Hospital Cost per general physician outpatient visit 56 27 38 Cost per OB/GYN visit 21 32 NA Cost per pediatric visit 30 29 NA Cost per other specialty visit 30 17 37 Cost per surgical discharge 516 900 809 Cost per medical discharge 550 545 778 Cost per OB/GYN discharge 369 211 244 Cost per pediatric discharge 474 495 510 Cost per vaginal delivery 152 265 237 Cost per caesarean delivery 156 439 291 NA not available. HISMA exhibited lower unit costs of inpatient care than the two other hospitals for all services except obstetric discharges other than deliveries. For some services, these differ- ences were considerable. For example, at US$516, the average cost of a surgical discharge at HISMA was 60 percent lower than the equivalent cost of US$900 at Chepo and 36 percent lower than Solano's unit cost of US$809. Likewise, at HISMA, the unit cost of normal deliv- eries was only US$152, a figure that was between 40 and 50 percent below that Chepo and Health System Innovations in Central America 103 Solano. Even larger differences were observed for the unit cost of caesarean deliveries, where HISMA's unit cost of US$156 was about one-third the unit cost at Chepo and one- half that at Solano. HISMA's higher economic efficiency in the production of surgeries as well as normal and caesarean deliveries is attributable in part to this institution's relatively greater use of operating rooms. In the case of Chepo Hospital, frequent interruptions of the power supply occurred, which interfered with good use of resources, and at Solano Hospital, surgical suites were not suitable for deliveries. In addition to a unit cost analysis, the researchers compared the budget structure of the three hospitals.89 A budget breakdown for 1999 is shown in figure 3.3. HISMA devoted 51 per- centofitsbudgettomedicalpersonnel,aproportionthatwassignificantlyhigherthanChepo's (41 percent) and also significantly below that of Solano (58 percent). Overall, HISMA devoted a much smaller share of its budget to medical supplies other than drugs than both Chepo and Solano. At the same time, HISMA spent an about equal share of its budget on drugs as Chepo, but both spent significantly more on drugs than Solano Hospital. If the budget shares of drugs and medical supplies are combined, then HISMA allocated to these two items the same share as Solano Hospital (17 percent). Chepo Hospital, instead, exhibited a considerably higher share of 33 percent, a difference that may largely be interpreted as a combination of small dif- ferences in patient mix, clinical practices, and efficiency in the use of those inputs. Finally, it is noteworthy that HISMA devoted a greater share of its spending to administration--15 per- cent in total, compared with 11 percent in Chepo and 10 percent in Solano. HISMA's greater Figure 3.3. Cost Structures Drugs Administrative personnel 12% 14% Administrative expenses Other medical supplies 1% 5% General support services 5% Other expenses 12% Medical personnel 51% Chepo Hospital Solano Hospital Administrative personnel Administrative personnel 10% Drugs 9% Drugs Administrative expenses 3% Administrative expenses 14% 1% Other medical supplies 1% General support services 14% 5% General support services 6% Other expenses Other expenses Other medical supplies 10% 9% 19% Medical personnel 41% Medical personnel 58% 89. These figures are based on actual spending, also known as the executed budget. 104 World Bank Working Paper emphasis on administration may reflect that hospital's greater concern with quality assur- ance. This issue is taken up in the next section. Quality This section presents the main study findings concerning health care quality. Ideally, quality assessment should be an evaluation of three dimensions: quality of inputs, quality of processes, and quality of outcomes. Quality of inputs consists of a comparison between the inputs actually available in the hospital to produce the services it offers and the inputs that should be available according to nationally or internationally accepted norms to make deliv- ery of the services offered possible. Gaps between actual and normative availability signal a quality failure. Quality of processes involves contrasting the actual actions carried out by medical personnel with the normative actions set forth in norms of care. A basic requisite for adequate medical processes is the existence, appropriateness, and use of treatment protocols and other mechanisms, such medical committees and peer reviews. When such accepted norms and mechanisms are available in a hospital, the assessment of technical quality con- sists of an evaluation of the degree to which medical personnel comply with them. The lack of such norms and mechanisms signals a basic failure that most likely impedes the delivery of appropriate medical care. Quality of outcomes encompasses an analysis of both patient satisfaction and health outcomes, the latter denoting the measurement of the degree to which patients' health status changed as a result of having been in the facility. The time, resources, and data available precluded a complete analysis of quality in its three dimensions. Instead, researchers examined the existence and use of quality assur- ance mechanisms in the three hospitals by administering a detailed accreditation survey in each facility, and they carried out a survey of randomly selected patients to assess per- ceived quality of care. Quality assurance. Although the existence of quality assurance processes will not ensure that quality is good, their absence is a cause of concern--the specialized literature suggests that it is most probably a symptom of quality failures. The results of the accreditation sur- vey are shown in Table 3.9 organized by area: management, support services, and clinical services. Overall, as shown in Figure 3.4, the two control hospitals barely achieved 50 per- cent of the minimum required compliance needed for accreditation of 70 percent (equiv- alent to 3.5 points in the 0­5 point scale used here). Only HISMA exceeded this minimum, but all three hospitals exhibited accreditation problems. In the two control hospitals, mechanisms aimed at defining, evaluating, and managing expected hospital results (clini- cal, output, and financial) were generally nonexistent or not applied, and this partly explains the differences obtained in performance indicators. User satisfaction. The researchers drew a random sample of patients of several clinical services from each hospital and interviewed them to assess their satisfaction with the ser- vices received. The patient satisfaction variables evaluated were: (a) general standards of care, (b) access to and timeliness of care, (c) availability of drugs, (d) treatment by profes- sional and nonprofessional staff, (e) physical comfort and hospitality, (6) tracers of med- ical quality, and (7) overall perception of quality. This analysis yielded statistically significant differences at the 5 percent level in patient satisfaction among the three hospi- tals. Further, HISMA exhibited the highest satisfaction scores of the three hospitals and in Health System Innovations in Central America 105 Table 3.9. Summary of Accreditation Performance Indicators Chepo Solano Area of accreditation Indicators HISMA Hospital Hospital Accreditation of Hospital management 4.2 2.2 2.8 management Central quality planning and control 3.6 1.6 1.5 Human resources management 4.3 2.2 1.5 Human resources training 3.5 1.5 1.0 Personnel health and safety 2.0 2.3 2.0 Physical plant and sanitation 5.0 2.4 2.1 Installations and equipment 4.3 2.0 1.8 Average 3.8 2.0 1.8 Accreditation of Pathological anatomy 4.3 1.0 2.3 support services Blood bank 5.0 1.7 4.3 Radiology 5.0 4.5 3.5 Clinical laboratory 4.8 2.8 3.3 Emergency room 4.3 2.3 3.0 Operating room 5.0 2.0 3.8 Pharmacy 4.3 2.8 2.5 Sterilization 5.0 2.8 2.8 Catering and nutrition 5.0 3.8 3.3 Milk dietetic servicesa 3.3 2.5 2.3 Average 4.6 2.6 3.1 Accreditation of OB/GYN, perinatal and postnatal 4.5 3.0 2.8 clinical services Pediatrics 3.8 3.3 3.3 Medicine 4.5 2.8 3.8 Surgery 4.0 2.8 3.0 Average 4.2 3.0 3.2 All accreditation Average 4.2 2.5 2.7 a. Bottle feeding for newborns. all dimensions of perceived quality. As is shown in Table 3.10, more than one-half of HISMA's patients who had health insurance reported their overall satisfaction to be "very good," and nearly one-half of those uninsured offered the same ranking. Almost no patients in Chepo Hospital reported to be as highly satisfied, and only between 11 percent and 14 per- cent of Solano Hospital's did so. Further, nearly all of HISMA's patients considered quality to be either "very good" or "good," whereas the equivalent scores were significantly lower at Solano and Chepo--the latter displaying the lowest degree of favorable satisfaction. Health outcomes. Ideally, the existence of outcome indicators in all three hospitals would have made possible a comparative analysis of one key component of quality among these institutions. Unfortunately, such information was not available everywhere. Thus, it was not possible to assess widely used outcome indicators common elsewhere, such as the NIR and the frequency of medical complications from hospitalization. In fact, only HISMA kept NIR records for 1999, and none of the three hospitals kept records of readmissions 106 World Bank Working Paper Figure 3.4. Overall Accreditation Scores 100 HISMA 90 Chepo Hospital Solano Hospital 80 70 60 50 40 30 20 10 0 Management Support services Clinical services Overall services and reinterventions for that year. This omission speaks negatively of quality assurance in all three hospitals, although the mere fact that HISMA had an active NIR committee and held the corresponding records is considered to be a sign that this institution had a specific concern for supervising and improving this quality aspect in the institution. Death rates or readmissions rates were not globally available in the three hospitals; thus, a comparative assessment from this perspective was not possible. Conclusions From a number of perspectives, HISMA performed consistently better than the two control hospitals, except for population coverage and bed use. In fact, HISMA's relatively small num- ber of admissions, combined with a shorter average length of stay, left it with 50 percent of its Table 3.10. Overall Patient Satisfaction (%) HISMA Chepo Hospital Solano Hospital Ranking Insured Uninsured Insured Uninsured Insured Uninsured Very good 58 47 2 0 11 14 Good 42 51 48 62 64 57 Mediocre 0 2 46 33 25 27 Poor 0 0 3 2 0 0 Very poor 0 0 0 0 0 0 No response 0 0 2 2 0 2 Total 100 100 100 100 100 100 Health System Innovations in Central America 107 bed capacity idle. This excess capacity should not be a cause of major concern, however, given HISMA's relatively short life. Instead, HISMA should explore whether with time its bed capac- ity is appropriately suited to its assigned population. If not, it should evaluate the convenience of reducing its bed capacity or expanding its reach to other MOH and SSI populations. As regards efficiency, HISMA outperformed the control hospitals in virtually all aspects technical and economic efficiency. In the area of quality, HISMA also did better than Chepo and Solano Hospitals. In this respect, large differences among the three hospitals are worth highlighting. These pertain to the existence of an institutional mission, clinical and administrative rules and procedures, quality assurance and accreditation systems, and a client orientation. Although the HISMA reform project has not yet been implemented fully, and not all the management tools required for full autonomy have been granted to it, HISMA's board has achieved most of its commitments while efficiently delivering good quality care. Study results offer strong arguments in favor of strengthening HISMA's model and expanding it to other, traditionally managed public hospitals such as Chepo and Solano. The strong resistance seen in Panama against changing the status quo and expanding HISMA's model attests to the difficulties, faced in this country and elsewhere in Latin America, of promot- ing policies that favor beneficiaries in an environment dominated by interest groups. Addi- tionally, some aspects of the design of the HISMA model may need further consideration: First, CONSALUD is the sole purchaser of the services provided by HISMA, and HISMA, in turn, is the sole provider of health services for the local target population. This situation of a combined monopsony (a market where there is a single buyer) and monopoly power may be mitigating the potential efficiency gains of this model. Second, further refinement of the current prospective provider payment system might be needed in the future because HISMA is paid a flat fee per hospital discharge, outpatient surgery, and emergency visit. This could create incentives on the part of HISMA for "cream skimming" the healthiest patients within each category. Bibliography Bitrán, R., C. Má, P. Gómez, L. de Thomas, y G. de Rojas. 2001. "Evaluación del modelo HISMA y comparación con un modelo tradicional de funcionamiento." Bitrán y Aso- ciados para el Ministerio de Salud de Panamá y el Banco Mundial, Santiago de Chile. Eichler, R., P. Auxila, J. Pollack. 2001. "Output-Based Health Care. Paying for Performance in Haiti." Note 236, Public Policy for the Private Sector. World Bank, Washington, D.C. Govindaraj, R., and M. Chawla. 1996. "Recent Experiences with Hospital Autonomy in Developing Countries--What Can We Learn?" Data for Decision Making Project, Department of Population and International Health, Harvard School of Public Health, Cambridge, Mass. Harding, A., and A. Preker. 2003. "Private Participation in Health Services." Human Development Network; Health, Nutrition, and Population Series; World Bank, Washington, D.C. Loevinsohn, B. 2001. "Contracting for the Delivery of Primary Health Care in Cambodia: Design and Initial Experience of a Large Pilot Test." World Bank, Washington, D.C. 108 World Bank Working Paper Mills, A., and J. Bloomberg. 1998. "Experiences of Contracting: An Overview of the Liter- ature." Macroeconomics, Health and Development Series, Technical Paper 33. World Health Organization, Geneva. Mills, A., C. Hongoro, and J. Bloomberg. 1997. "Improving the Efficiency of District Hospitals: Is Contracting an Option?" Tropical Medicine and International Health 2(2):116­26. Over, M., and N. Watanabe. 2003. "Evaluating the Impact of Organizational Reforms in Hospitals." In A. Preker and A. Harding, eds., Innovations in Health Service Delivery: The Corporatization of Public Hospitals. Human Development Network; Health, Nutri- tion, and Population Series. Washington, D.C.: World Bank. Preker, A., and A. Harding. 2003. Innovations in Health Service Delivery: The Corporatiza- tion of Public Hospitals. Human Development Network; Health, Nutrition, and Pop- ulation Series. Washington, D.C.: World Bank. CHAPTER 4 Contracting Primary Health Care Services--The Case of Costa Rica James Cercone, Rodrigo Briceño, and Varun Gauri Abstract This chapter outlines the Costa Rican experience with the introduction of a contracting model and highlights some of the key results evaluating the performance of the model in comparison to the traditional public model. Using both performance evaluation data from the Caja Costarricense de Seguro Social (CCSS) contracts and a panel data set on clinic-level outputs, the chapter analyzes the development of the primary health care (PHC) contracting model in Costa Rica and compared the performance of health cooperatives in Costa Rica with traditional, publicly managed clinics in the social security system. Under both methods, the contracted providers consis- tently provide better performance than traditional CCSS public clinics. Even controlling for catchment area characteristics, time trends, complexity of clinic, and existence of performance contracts, contracted providers con- ducted significantly more general practitioner visits per capita and signifi- cantly fewer specialist visits per capita than traditional clinics. At the same time, the per capita rate of emergency visits and first-time patients seen did not differ from that in the traditional clinics. These measures combined resulted in 30 percent lower expenditures per capita in contracted providers than in the traditional clinics. The findings presented suggest that the contracting model in Costa Rica is promising. As this chapter shows, cooperation between the sectors has worked well there, where contracted providers have been in operation long enough to produce reliable results. Clear evidence is shown that the public-private partnership model offers important opportunities to increase provider performance within a sustainable framework. O ver the past 20 years, Costa Rica has achieved remarkable progress in the health of the population through a universal social security system and annual govern- ment spending in excess of 6 percent. Despite these achievements, Costa Rica 109 110 World Bank Working Paper faced considerable pressure to introduce structural changes in the health system to address rising costs (total health care spending reached nearly 10 percent of gross domestic product (GDP) in 1996), declining quality of public services, and increasing demands for improved quality and client attention by the system's users. To address these challenges, in the late 1980s, the government of Costa Rica began to formulate long-term strategies to reform the health care system. These reforms built upon earlier policy decisions to shift all responsibility for health care provision to the Caja Costarricense de Seguro Social (CCSS) and strengthen the regulatory and policy oversight role of the Ministry of Health. By 1994, the overall diagnosis of problems in the health sector and the directions for change were clear; however, little progress had been made in terms of introducing real change (Miranda 1995; CCSS 1997). Among the main issues that motivated the need for change were: a health care model based almost exclusively on hospital-level, curative attention, with limited emphasis on an integrated health care model and gaps in coverage; health care expenditures that reached nearly 10 percent of GDP and increasing evi- dence of declining returns in terms of production, quality, and satisfaction; rising levels of user dissatisfaction, due primarily to waiting lists and poor attention; a high level of centralization, leading to inefficiency in the provision of services and a lack of ownership and accountability at the local level; a rigid normative framework that emphasized ex ante controls rather than performance-based evaluations in accordance with pre-established goals for pro- duction, quality, and user satisfaction. In response to these problems, the government of Costa Rica launched an ambitious reform agenda that aimed to improve access, efficiency, and quality as well as guarantee the long-term sustainability of the social security institute. One of the key elements of the reform strategy was to increase the participation of the private sector as a provider of ser- vices to the social security institute within a regulated framework. This allowed the CCSS to leverage the investment capacity of the private sector, management experience and flex- ibility, and pay greater attention to clients' needs. The cooperation between the public and private sectors was established in the context of the outsourcing of services to non public providers whereby the CCSS maintained its role as the single payer and contracted with pri- vate providers to deliver core services. This strategic partnership between the public and private sectors produced significant improvements in coverage, efficiency, and quality of services. This chapter outlines the Costa Rican experience with the introduction of a contract- ing model and highlights some of the key results evaluating the performance of the model in comparison to the traditional public model. It is divided into four sections. The next section provides an overview of the Costa Rican health sector reforms. The second section introduces the main characteristics of the contracting model in Costa Rica. The third sec- tion highlights the comparative performances of the traditional and contracted models. The final section provides conclusions and lessons learned for other countries seeking to implement contracting models. Health System Innovations in Central America 111 The Health Sector Reforms in Costa Rica Starting in 1994, the government initiated a three-pronged approach to health reform. The main objective of the reform process was to address the weaknesses in the system through the implementation of structural reforms in the organization, financing, and delivery of services, with a focus on improving value for money in the social security system. It was decided to preserve the universal coverage and public financing role of the social security institute and focus on improving the efficiency and quality of services provided through CCSS providers. The reforms used a multidimensional approach to address several of the challenges facing the health sector. The first dimension of the reforms focused on strengthening primary health care through the implementation of an integrated health care model. The second ele- ment aimed at improving efficiency and quality by introducing new financing mechanisms to link the allocation of resources to health needs and actual production, thereby replacing the historical budget, and improve efficiency in the collection of social security contributions. The final component introduced reforms at the organizational level to promote the sepa- ration of financing, purchasing, and provision and support the changes in the health care and financing models. In 1997, the CCSS entered into performance contracts (compromisos de gestión) with seven social security hospitals and five health areas (decentralized primary health care units). The development and implementation of the performance contracts depended on two key elements. First, the planning and negotiation of the performance contracts required iden- tification of the health needs of the population, analysis of production capacity and relative performance, and design of a comprehensive evaluation framework to ensure adequate monitoring and evaluation of compliance with the performance contracts. Second, the CCSS began to act as a purchaser of health care services, clearly identifying the health needs of the population, designing strategies to improve the population's health, and establishing clear and concise objectives and indicators for each health provider. At the same time, these contracts provided the social security institute with the opportunity to provide incentives to health providers to implement the integrated health care model and enforce compliance with the standards and norms that were developed to strengthen primary health care. The contracts standardized the terms and conditions that providers were expected to fulfill and outlined a series of performance measures to guide the providers toward improved performance. The negotiation of the performance contracts changed the dynamic of the command and control system that operated under the historical budgeting and planning model. (The main clauses of the contract are outlined in Annex 4.1.) There are three key elements of the contract: First, the contract clearly establishes the population to be covered and the types of ser- vices to be delivered. For each type of service, the CCSS has established protocols and guidelines that outline the purchaser's expectations regarding the provider's role. Second, the contract outlines multiple performance indicators, grouped into four main categories related to: (a) provision, which includes coverage rates and com- pliance with protocols; (b) quality, which establishes indicators related to technical and perceived quality of services; (c) organization of the services, which covers 112 World Bank Working Paper management capacity and its role in managing patients and resources; and (d) billing, which establishes the provider's role in documenting the cost and volume of services provided. Finally, the contract establishes a system for monitoring and evaluation, which outlines not only the procedures but also the process by which the providers will be evaluated. This section also establishes what the incentives, or sanctions, the providers face will be. The initial reform efforts focused on improving the performance of public providers, and by the end of the 1990s, it was clear that further innovation would be required to achieve the desired objectives in terms of expanding coverage and improving efficiency and quality. The CCSS was limited by investment and employment constraints that limited the possi- bility of expanding coverage with traditional models in the metropolitan area. Furthermore, the CCSS had established a series of contracts with nonpublic providers that needed to be incorporated into the new regulatory, or contracting, framework established for public providers (CCSS 1997). These factors, and the pursuit of greater value for money, forced the CCSS to adopt models that emphasized the purchaser-provider split. Under this model, the CCSS, as a purchaser, contracts with the private or nonprofit sector for the delivery of primary health care services within a framework that is similar to the conditions the pub- lic sector faces but offering greater management flexibility. This solution was aimed at improving access, efficiency, quality, and satisfaction with the existing resources and within the existing constraints. The remainder of this chapter outlines how the contracting was introduced, describes its key characteristics, and evaluates the comparative performance of the model. The Emergence of PHC Contracting in Costa Rica Background A key part of these reforms was the urgent need to extend access, improve service, and lower costs by strengthening the primary health care (PHC) system. Despite the successful imple- mentation of the reformed PHC model in nearly 75 percent of the population, the CCSS faced increasing difficulties in extending the model to urban populations. The main limita- tions were important restrictions on public investment required to build capacity to attend to the nearly 300,000 people without coverage in the metropolitan area, a limitation by the Ministry of Finance on the creation of new staff positions, and hesitance of the middle-class population to use public services. After careful consideration, the CCSS decided that the best option to extend coverage and improve efficiency without increasing public invest- ment was the use of a contracting model whereby PHC services would be purchased from the private sector. This decision reflects a paradigm shift in the health sector. In the past, policy makers assumed that the state should guarantee access to basic services such as security, health, and education through direct provision of all services. However, this belief evolved, when faced with the difficulties and realities of the sector, to a situation where the state is not nec- essarily expected to produce services, but should ensure that they are provided by means Health System Innovations in Central America 113 of a third party, retaining responsibility for quality, quantity, and pricing. In the past 10 years, a number of experiences in other sectors in developing countries, and in the health sector in developed countries (England 1997; Mills 1998), have contributed to increased interest in contracting for the provision of PHC. It has been found that con- tracting for services is attractive because it has the potential to: ensure an increasing focus on the achievement of measurable results, particularly if contracts define objectively verifiable outputs and outcomes; utilize the private sector's greater flexibility and generally better management skills to improve services and responsiveness to consumers; increase managerial autonomy and decentralize decision making to managers on the ground; use competition to increase effectiveness and efficiency; allow governments to focus less on service delivery and more on other roles they are uniquely placed to carry out, such as planning, standard setting, financing, and regulation (Vining and Weimer 1990). Following the theoretical models and aiming to develop a practical solution to extend cov- erage of the PHC model to the remaining 25 percent of the population, the CCSS designed a number of contracting models for the provision of PHC services. The aim of the new CCSS model was to increase the capacity of the primary care network and bring essential services closer to the population. This involved the CCSS establishing a partnership with private entities to provide public services. These services would be offered using public facilities but under private management, with the objective of increasing coverage and improving efficiency. To achieve this, performance contracts were prepared between the CCSS and privately managed providers. The performance contracts with the private sector retained the figure of the compromiso de gestión, as in the case of public providers, to ensure parity between the public and private models and minimize the tendency to establish different conditions for public and private providers. Nearly all of the clauses within the contracts with private providers are the same as public contracts. The main difference in the contract is the use of a performance guarantee bond in the private contracts, which guarantees that at least 85 percent of the established performance targets are achieved; otherwise, the CCSS can "collect" on the bond and effectively penalize the private providers for underperforming. Public providers do not hand over a performance guarantee to the CCSS but rather "risk" 2.5 percent of the budget against low performance. Purchaser and provider separation was maintained, replacing the old system, which had operated under a budget and planning model. In addition, the new model offered important improvements in terms of organizational arrangements, incentive structures, and financial mechanisms, which potentially lead to improved performance. The approval of a law of decentralization further consolidated efforts to devolve authority and respon- sibility to public providers, allowing clinics to carry out their own administrative contracts for property and services. The law also established the compromiso de gestión as the legal instrument to be used for contracting public and private providers. The next section dis- cusses the introduction of the contracting model and its key advantages. 114 World Bank Working Paper How Does the Contracting Model Differ from the Traditional Model? Costa Rica has been a pioneer in the use of outsourcing to nonpublic providers with the introduction of a cooperative model as early as 1988. Under this model, the CCSS leased pub- lic clinics to groups of professionals organized under a new legal structure, the cooperative, with the objective of managing the clinic and providing services to the publicly insured pop- ulation. The staff of the new clinic was a combination of former CCSS staff and new staff who were recruited. Management was provided by former CCSS physicians. This model has been the cornerstone of purchaser-provider split reforms that have been recently implemented in Costa Rica. The initial model showed positive results in terms of program coverage and user satisfaction (Pezza and Barquero Bolaños 1994). Moreover, the use of private providers has not been interpreted as a privatization of services but rather as a tool to obtain public objectives of coverage, universal access, efficiency, and quality within a framework of solidarity (social cohesion and sharing of financial risks). This model appears to have important benefits over more liberal forms of privatization that have been under- taken in recent years by countries such as Chile (Homedes and Ugalde 2002). Over the years, the model has evolved to include private providers, the University of Costa Rica, and other cooperatives. In 2000, the CCSS initiated a new phase of contract- ing out when they issued a request for proposals to provide extended coverage to nearly 110,000 people in the metropolitan area. This process allowed the CCSS to stimulate com- petition among the existing providers and improve transparency in the management of the process. One of the critical differences in the Costa Rican model is that competition is generated through improvements in quality and efficiency, rather than by encouraging users to select their provider. Other countries--such as Croatia (Hebrang and others 2003), Estonia (Koppel and others 2003), and Nicaragua (La Forgia and Gutiérrez 2000)-- have had success in introducing greater competition through contracting out to private providers. The contracting model has grown to constitute a model of relationships between the public sector, as a purchaser of health services, and the nonpublic sector (including the private sector, the University of Costa Rica, and cooperatives) as providers of services. For the remainder of this chapter, the term PHC contracting will be used to refer to the model under which the CCSS contracts with nonpublic providers, describing the relationship between the CCSS and private providers or the contracts with a cooperative entity. Given the similarity of the conditions faced by public and nonpublic providers and the model's longevity, the evolution of PHC contracting in Costa Rica offers an excellent laboratory for analysis of the comparative performance of the models and highlights key features that are critical to the design and implementation of a contracting model. The main objective of the cooperative model is to generate the best possible social well- being of an assigned population with the existing resources. Within the framework of poli- cies promoting increased services and the development of an integrated care scheme,1 the cooperative performs an essential role. Four important areas--regulation, finance, orga- 1. The term integrated care scheme refers to the provision of a complete set of PHC services (includ- ing prevention, cure, rehabilitation, and social services) provided by one primary care provider. Under this model, the provider also offers all ancillary services, such as pharmacy, laboratory, and diagnostic ser- vices, required for the provision of the PHC package. Health System Innovations in Central America 115 nization, and provision--highlight the key differences between traditional public clinics and cooperatives: Regulation: Contracted providers perform their work under the general oversight of the CCSS and the Ministry of Health (MOH) but retain direct governance over their resources and business objectives. Service agreements between the provider and the CCSS establish goals for production, coverage, and other performance indicators related to quality and satisfaction. Through the use of a contract, the CCSS outlines the key regulatory aspects for the contracted provider and establishes the "rules of the game". Evaluation of the provider's performance, which ensures quality and coverage, is the direct responsibility of the CCSS. The provider must also respond indirectly to the regulations and political objectives determined by the CCSS and the MS. The governance of the contracted provider is established wholly by the provider organization. In cooperatives, a governing board is established that includes the staff of the cooperative and, in some cases, members of the community; in the private sec- tor clinics, private management boards have been established; and at the University of Costa Rica, a joint university board (composed of professors from the School of Public Health and staff from the provider level) is established for oversight. Financing: The financing function has been fully retained by the CCSS. Under Costa Rica's single-payer system, the CCSS is responsible for financing 100 percent of health care services through payroll contributions and government transfers for the uninsured. The financing of the contracting model retains public financing, and no additional fees are charged by the providers. Since the introduction of the first cooperative in 1988, three provider payment mechanisms have been used by the CCSS: per capita payment, payment in accordance with the historical budget, and per capita payment through a management agreement. The residual claimant con- dition provides an incentive to develop high levels of productivity and efficiency, and financial risk from bad management is reduced. Organization: The internal organization of the contracted providers is entirely up to their management. In the cooperatives, managers are appointed by an administrative council, which itself is selected every two years by the General Assembly. The model is guided by Law 4179 of cooperative associations and their reforms. The most impor- tant aspect of this kind of organization is the contribution by all associates to the social capital of the company. Periodically, associates receive a share of the profits gener- ated by the cooperative. Alternative management models have been established in the private companies and the university operating under the contract model; in most cases, they separate the administrative structure from the clinical structure, which provides a medical director who is subordinate to a general manager. Provision: The contracted providers are in charge of managing the provision of ser- vices without interference by the CCSS authorities. In the cooperatives, the CCSS leases the clinic for roughly US$1 per year, and the cooperative is in charge of main- tenance, improvements, and investment. Some equipment is purchased by the CCSS, but most investment is made by the contracted provider, including both fur- niture and instruments. 116 World Bank Working Paper Among the other providers, the facilities are entirely the responsibility of the contracted provider. In most cases, the private managers have rented space and made improvements to adapt the facilities to the infrastructure requirements estab- lished by the CCSS (for example, space distribution and equipment). Inputs, such as drugs and medical supplies, required by the contracted providers are bought directly from the general stores of the CCSS. Prices, determined by the accounting department of the CCSS, are calculated on the basic cost of production or acquisi- tion, plus 15 percent for administrative expenses. When stock is unavailable from the CCSS, the contracted providers purchase inputs from the private sector. The services provided by the contracted providers are identical to the package of services provided by the traditional CCSS clinics. The management contract estab- lishes the specific services and coverage rates that the contracted provider must pro- vide, and the performance evaluation process ensures that quality is maintained. Organizational Behavior: Analyzing the Incentives Facing Contracted Providers Organizational behavior is a collection of formal or informal rules and administrative pro- cedures for the efficient organization and management of resources. Theoretically, the suc- cess of an organization depends on the response of the administrators and managers to the organizational structure and incentives they face. The conduct of contracted providers can be analyzed by looking at managerial autonomy, market exposure, residual claimant sta- tus, human resources, and strategic management. Managerial autonomy. The reform process began with the decentralization of services provided by the government, which granted all providers greater autonomy to manage their resources. Unfortunately, public centers have important constraints, both political and technical, that have limited the implementation of autonomy. The providers that oper- ate under the contracted model have nearly full autonomy when it comes to making deci- sions in areas such as purchasing, human resources, and general administration. They are, however, limited to minimum standards regarding infrastructure, human resources, equipment, and materials and supplies that are established in the contract. In comparison with traditional CCSS clinics, which have only limited autonomy to make decisions in these areas, the contracted providers have considerably more flexibility in the management of their resources. Market exposure. Market exposure is the extent to which clinics face competition. The contracting process between the CCSS and the contracted providers has different levels of competition. Contracting between the CCSS and public providers implies zero competition in terms of price or quality. Populations are assigned geographically and budgets are capi- tated. In the original cooperatives, the populations were also assigned geographically and budgets were capitated. In the latest round of contracting between the CCSS and the private providers, however, competition has been introduced. Populations and geographic areas were "bid out" to the cooperatives and private providers under a competitive tender process. Providers were forced to compete for a set package of services, with the minimum standards mentioned above, and finally adjudication was made based on the lowest price per capita. The private providers and cooperatives are also required to provide a guarantee bond in Health System Innovations in Central America 117 cash, which is linked to obtaining a minimum level of performance in the contract execu- tion. If the contracted provider does not meet the performance standard of at least 85 per- cent in the performance indicators, then the CCSS, as purchaser, can execute the bond and collect a penalty of nearly 5 percent of the contract value. Increased market exposure clearly has produced improvements for the CCSS in terms of greater value for money. Although the traditional clinics have low market exposure, the process of negotiating the annual management agreements does make them accountable for performance and introduces a quasi-market situation in which the purchaser is forcing the providers to deliver better quality at a lower price. Residual claimant condition. The term residual claimant status refers to the extent to which the organization retains profits that result from improvement in management and assumes losses from bad management. In the traditional clinics, the residual claimant is the CCSS. Initially, the clinics were able to retain some of the savings, but in recent years, this has been reversed, so the traditional clinics have limited incentive to carry out improvements. The only incentive they have is the annual performance bonus that is linked to compliance with contract performance measures. This also stimulates improved per- formance in CCSS clinics. In the case of the private, for-profit, contracted providers, the residual claimant is the provider, meaning that any improvement in management will be translated into greater profits and benefits for all employees. Social functions and public property. One of the fears surrounding privatization is that the public sector could retain the production of services that are considered public goods or that complete a social function of government, and private provision will lead to higher prices and exclusion of some people. The core difference between the contract model and simple privatization is that the contracting model does not imply a sale of assets to the pri- vate sector to provide a public services, nor does it imply that the public sector is dismiss- ing its responsibility for the provision of core services--rather, it is the use of the private sector to execute a core public function, under public regulation and with public financ- ing, through the use of a private provider. The services provided by the contracted provider are delivered under the same terms as those pertaining to the public providers, which ensures that the services are available to all free of charge. Human resources. The management of human resources is an important differentiat- ing factor between the contracted providers and the public providers. In the contract model, the clinics are free to hire and fire staff as they see fit. There are no public encum- brances or labor unions to cause obstacles to the management of human resources. The contracted providers are also able to pay competitive salaries, which are greater than the salaries paid by public institutions. The contracted providers have been able to introduce performance bonuses that link the total staff remuneration to the achievements in perfor- mance indicators established in the contracts. The public institutions have not been able to introduce incentives or increase flexibility in the management of labor contracts, which are potentially important elements of overall performance. Business strategy. The success of an organization is determined by a clear mission and execution of business strategies to achieve its goals. In general, the mission and vision of 118 World Bank Working Paper the public and private providers is the same--to provide the PHC package of services to the CCSS-insured population with the highest standards of quality in a sustainable framework-- and the CCSS as purchaser establishes the terms under which services will be provided. The main difference is in terms of execution of the strategy. In this case, the contracted providers clearly have greater flexibility in managing resources and obtaining results. The next section compares the performance of the contracted providers with that of traditional clinics. The outlined combination of the advantages will translate into signifi- cant improvements in performance by contracted providers. Evaluating the Impact of Contracting Four previous comparative evaluations of cooperatives in Costa Rica have been conducted. Herrero, Villalta, and Associates (1992) compared costs between two cooperatives and five selected traditional CCSS clinics in 1991, and found that the variances in expenditures and treatments were higher in the traditional clinics. Their findings, though suggestive, were based on one year of cross-sectional data and a nonrandom sample. Durán and others (1995) compared expenditures between four cooperatives and four comparator clinics during 1990­94 and found that cooperatives had higher levels of expenditures. The nonrandom selection of the comparison clinics, as well as the use of incomplete budget data, might have biased that analysis. Picado (1999) and Rodríguez (1999) both compared 1992­98 expen- ditures between three cooperatives to those of four traditional CCSS clinics and found higher but declining relative expenditures in the cooperatives. Again, sampling issues were problematic in those studies. The performance evaluation of this study focuses on comparing the performance of pub- lic clinics with contracted providers. As indicated in the introduction, both private providers and cooperatives are considered as contracted providers. This evaluation is divided into two levels of analysis. In the first case, the overall results reported by the CCSS from the annual performance evaluations carried out by the CCSS purchasing department are examined. This evaluation is based on the performance indicators established in the annual perfor- mance contracts. The second case looks at the evolution of performance over time and establishes a multivariable framework to compare contracted providers to the CCSS type III and type IV clinics, which have relatively the same level of complexity as contracted providers.2 The use of a multivariable framework controls for several potential community-level covari- ates over time and allows more robust results than comparing mean performance on out- come indicators. Comparative Performance Based on Contracting Arrangements This section outlines the comparative performance between contracted providers and tra- ditional CCSS providers. Two levels of analysis are undertaken to test the hypothesis that 2. Type III and type IV are levels of medical complexity in Costa Rican facilities. Type III and type IV clinics include family medicine and limited specialty care, such as pediatrics and obstetrics. Two of the cooperatives were type IV and one was type III. The facilities compared to the cooperatives include all tra- ditional CCSS type III and type IV clinics. Health System Innovations in Central America 119 contracted providers offer better value for money than traditional clinics. In the first case, evaluation data from the performance contracts for 2002 are used. Here, the evaluated out- comes of nonpublic and public providers are compared. The second level of analysis looks at the comparative performance of the public versus nonpublic providers based on the use of panel data from 1990 to 1999 for output and cost indicators. Evaluating the performance of contracted providers based on contract evaluations. The data used are based on the annual evaluation of contract performance that is carried out by the CCSS purchasing department. We have selected a subset of 7 indicators from the core set of 23 indicators that are used to evaluate performance. These indicators provide a cross- set of program indicators evaluated in the contract. A combined indicator was developed to create a coefficient showing the percentage of the target coverage rate achieved by the provider. Thus, providers that have a higher rate will have obtained a better than expected outcome. The indicator is applied to both the contracted providers and the CCSS public providers from the metropolitan area. Limiting the comparison to the metropolitan area provides a more consistent framework for the evaluation of homogenous populations and clinic structures. Annex 4.2 provides a full list of the indicators evaluated in the contract. The indicators considered include: rate of Pap smears for women more than 35 years old; coverage rate for interventions established for children under 28 days; coverage rate for interventions established for children under 1 year; coverage rate for interventions targeted to the elderly; coverage rate for the diagnosis and treatment of patients with hypertension; coverage rate for the diagnosis and treatment of patients with diabetes; coverage rate for prenatal care visits. The performance indicators evaluate two dimensions. The first dimension is only the absolute coverage levels for each indicator. For each indicator, a target level is established individually for each provider; each indicator has a set of approximately five key quality indicators that are evaluated to determine if the program is being implemented according to the agreed proto- cols. The final evaluation of performance of all providers is carried out by a combination of the absolute and quality-adjusted coverage levels. Figure 4.1 shows the combined coverage and quality coefficients obtained for the providers in the selected programs.3 This evaluation is the same for contracted providers and traditional providers, although the contracted providers typically have higher target levels than the CCSS clinics. Incentive payments are awarded to both types of providers based on these evaluations. As shown in Figure 4.1, in five of the seven indicators, the contracted providers provide better performance than the comparator CCSS clinics. In the case of prenatal coverage, the difference is 20 percentage points in favor of the contracted provider; in the other areas, the difference is around 10 percent. In the case of elderly care and coverage of children under one year, the CCSS clinics outperform the contracted. However, it is important to note that this partially reflects the higher target rates negotiated for contracted clinics. 3. The Pap smear rates are not adjusted for quality because an additional indicator, percentage of Pap smears reported positive, is used to measure quality. 120 World Bank Working Paper Figure 4.1. Comparative Performance of Contracted Providers and CCSS Metropolitan Clinics, Selected Performance Indicators 89% Pap 77% Children under 28 89% days 80% 82% Children under 1 year 86% 75% Elderly care 76% Indicator 74% Hypertension 70% Prenatal care 60% 39% 55% Diabetes 42% 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% Level of achievement Traditional Contracted providers Using a multivariable environment to evaluate contracted provider performance.4 Gauri, Cercone, and Briceño (2004) used multivariable analysis to further analyze the performance of contracted providers in comparison to traditional CCSS clinics. For the analysis, data on all type III and type IV clinics in Costa Rica were available, but clinics for which data were available only for very short periods (two or three years) were excluded, as were those that were only recently set up. As a result, 23 observations were dropped, leaving the database with 24 type III and type IV clinics, 3 of which are cooperatives. The main source of information was the annual statistical bulletin of the CCSS. The data on infant and general population mortality were taken from specific studies by the Estado de la Nación, developed by the United Nations Development Programme (1990­2000), and an infant mortality study by the Insti- tuto Nacional de Estadísticas y Censos (General Department of Statistics and Census) that covered several years. The population data, including the demographic structure of the catch- ment areas, were compiled from projections made yearly since the 1984 census by the demo- graphic studies section of the Department of Actuarial and Economic Planning of the CCSS. The main objective of the empirical analysis was to determine whether health services provided in cooperative clinics in Costa Rica differed from those provided in the tradi- tional public clinics over the same period. To accomplish this, we first examined long-term 4. This section draws on a previous article by Gauri, Cercone, and Briceño (2004), under express authorization of Oxford University Press. Health System Innovations in Central America 121 trends and descriptive statistics and then modeled health service performance indicators using the general form. Yit = Coopit + Xit + Zt + eit where Y is a vector of health service performance indicators in clinic i at time t, Coop is a binary variable indicating whether or not a clinic is a cooperative, X is a vector of clinic and community characteristics, Z is a set of annual indicator variables, and e is a randomly dis- tributed error term. Panel data sets like this one typically encounter two problems when estimated using ordinary least squares (OLS): (a) autocorrelation because performance indicators in adjacent years (and therefore the error terms on adjacent observations) are correlated with each other, and (b) heteroskedasticity, or unequal variance in the errors. To address these problems, we used year dummies (the Z term above) to correct for autocorrela- tion and Huber-corrected standard errors to generate estimates robust to heteroskedasticity. It was not possible to use a fixed effect model to correct for heteroskedasticity because two of the three clinics were cooperatives for all years in the sample, and the third was a cooper- ative for 7 of the 10 years; consequently, clinic fixed effects would absorb almost all of the variation in the variable of interest, Coop. A generalized estimating equations model with- out year dummies might have been more efficient than the approach adopted, but because the correlation structure of the data was unknown, it might also have yielded substantially biased estimates. The reported results are therefore based on OLS estimates with year dum- mies and Huber-corrected standard errors. To reiterate, in the results reported below, the key parameter of interest is the coefficient on the Coop variable. Comparing the Historical Outcomes of Contracted Providers with CCSS Traditional Clinics The four CCSS clinics chosen were Clínica Carlos Durán, Clínica Jiménez Núñez, Clínica Moreno Cañas, and Clínica Clorito Picado. The contracted providers considered in this part of the study are COOPESALUD (Clínica de Pavas), COOPESAIN (Clínica de Tibás), and COOPESANA (Clínica de Santa Ana). These three cooperatives are used in this study because they were the first to be set up in Costa Rica. COOPESANA figures from 1995 onward are employed as well as annual CCSS statistics. Where information was not avail- able on the population in the area of each of the clinics, the figures of the relevant canton were used. The projected census from 1984 was used for population figures. The graphs in Figure 4.2 show trend lines over the decade for cooperatives and tradi- tional CCSS clinics. The figure shows that general medical visits per capita started at the same level in cooperatives and other clinics, then declined in cooperatives but increased in traditional clinics over the course of the decade. Specialist visits per capita also started at about the same level and declined in both cooperatives and traditional clinics. Coopera- tives gradually increased their rate of emergency visits per capita over the decade to the point that they equaled those in traditional clinics.5 Total laboratory procedures and total 5. Emergency visits include emergency services provided to CCSS members either during normal working hours or by emergency services after normal working hours. These are provided in the primary care setting of the cooperative clinic. 122 World Bank Working Paper Figure 4.2. Services Provided Per Capita in Cooperatives and Type III and IV Clinics 2.50 0.70 capita capita 0.60 per 2.00 per 0.50 1.50 visits visits 0.40 1.00 0.30 0.20 medical 0.50 medical 0.10 0.00 0.00 General 90 91 92 93 94 95 96 97 98 99 90 91 92 93 94 95 96 97 98 99 Specialty Coops Clinics type III and IV Coops Clinics type III and IV 0.50 2.00 0.40 visit 1.70 capita per per 0.30 1.40 exams visits 0.20 1.10 laboratory 0.10 0.80 Emergency Total 0.00 0.50 90 91 92 93 94 95 96 97 98 99 90 91 92 93 94 95 96 97 98 99 Coops Clinics type III and IV Coops Clinics type III and IV area visit 3.50 11,000 clinics per 3.00 the 10,000 of catchment 2.50 9,000 in colones prescribed 2.00 year 8,000 real per in 1.50 expenditure 7,000 medicines real 1.00 6,000 person 90 91 92 93 94 95 96 97 98 99 90 91 92 93 94 95 96 97 98 99 Total Total per Coops Clinics type III and IV Coops Clinics type III and IV Source: Authors' calculations based on constructed database of CCSS statistics. medications provided started at lower levels in cooperatives than in traditional clinics and remained lower at the end of the decade (though they did increase in the cooperatives). Real expenditures per capita exhibited significant annual fluctuations, but they both started and ended the decade lower in cooperatives than in traditional clinics. Table 4.1 shows that population mortality rates appeared somewhat higher in the areas around the cooperatives than in the CCSS clinics, and that infant mortality rates appeared comparable between the two groups. A t-test, however, found that the average annual infant and general mortality rates were statistically indistinguishable in the catchment areas of cooperatives and tradi- tional clinics. Similarly, the demographic structure of the populations in cooperative catchment areas was broadly comparable to that in the catchment areas of the traditional Health System Innovations in Central America 123 Table 4.1. Comparison of General and Infant Mortality Rates between Cooperatives and CCSS Traditional Type III and IV Clinics General mortality rates Infant mortality rates Traditional Traditional CCSS clinic CCSS clinic Year Coops (type III & IV) Coops (type III & IV) 1990 4.32 3.95 12.28 14.16 1991 4.33 3.94 11.48 12.32 1992 4.63 4.04 12.99 13.14 1993 4.47 4.08 11.78 12.39 1994 4.58 4.19 11.96 12.04 1995 4.75 4.44 10.71 13.77 1996 4.70 4.33 11.49 12.42 1997 4.75 3.93 17.22 14.17 1998 4.85 4.13 9.37 12.13 1999 4.85 3.85 NA NA NA not available. Source: Authors' calculations based on constructed database of CCSS statistics. clinics.6 Table 4.2 presents the same data as Figure 4.2, but in numerical form. Generally, examining time trends suggests that cooperatives provided fewer general medical visits, laboratory exams, and drugs than the traditional clinics, but systematic patterns were not evident for the other variables. Table 4.3 presents estimation results for the determinants of visits per capita, where catchment areas are used to calculate the estimated population served for each clinic, gen- eral mortality and infant mortality (both lagged one year) control for the socioeconomic level of the community served, and indicator variables for the existence of a management contract and for type III clinics control for clinic characteristics. All estimations included annual indicator variables to control for average trends at the national level. Lagged infant mortality rates were negatively related to most types of service utilization, confirming the widely observed phenomenon that services are available and utilized more in communities with higher social indicators. The variable for management contract, a dummy variable indicating the existence of a formal input- and output-based contract between the CCSS and a clinic, was significant in none of the six estimations, but the coefficient on the coop- erative indicator variable was significantly different from zero in three of them, at P<.01. 6. Using both t-tests and F-tests, we could not reject the hypotheses that cooperatives and traditional clinics had the same infant mortality rates and the same general mortality rates. The cooperatives had a slightly lower share of children under the age of 10 (20.0 percent versus 22.2 percent), a slightly higher share of women of childbearing age (23.9 percent versus 23.5 percent), and a slightly higher share people over the age of 60 (8.5 percent versus 7.5 percent) in their catchment areas than the traditional clinics. These differ- ences in demographic structure were all statistically significant; but because the cooperatives had more women of childbearing age and more older people, one would expect, absent any organizational differences, higher health care expenditures in the cooperatives. In fact, as the analysis shows, the opposite was the case. 124 World Bank Working Table 4.2. Comparison of Average Service Provision Rates per Capita between Cooperatives and CCSS Type III and IV Clinics: Selected Service Categories Paper GP visits per capita Specialists visits per capita Emergency visits per capita Lab exams per visit Medicines per visit Traditional Traditional Traditional Traditional Traditional CCSS clinic CCSS clinic CCSS clinic CCSS clinic CCSS clinic Year Coops (type III and IV) Coops (type III and IV) Coops (type III and IV) Coops (type III and IV) Coops (type III and IV) 1990 1.45 1.41 0.57 0.54 0.39 0.16 0.81 1.86 1.67 3.04 1991 1.86 1.37 0.63 0.54 0.39 0.17 0.81 1.14 1.77 3.02 1992 1.96 1.35 0.53 0.56 0.44 0.19 0.86 1.19 1.81 3.01 1993 1.89 1.26 0.55 0.52 0.44 0.18 0.92 1.23 1.91 3.06 1994 1.78 1.30 0.53 0.53 0.36 0.34 0.99 1.17 1.85 3.05 1995 1.77 1.33 0.45 0.55 0.39 0.32 0.91 1.21 2.03 3.11 1996 1.78 1.24 0.42 0.57 0.41 0.35 0.94 1.22 2.07 3.00 1997 1.82 1.28 0.44 0.54 0.41 0.40 0.99 1.33 2.11 2.84 1998 1.84 1.25 0.48 0.48 0.43 0.41 0.99 1.34 2.06 2.82 1999 1.84 1.02 0.48 0.32 0.47 0.41 1.10 1.44 2.08 3.08 Source: Author's calculations based on constructed database of CCSS statistics. Health System Innovations in Central America 125 Table 4.3. Determinants of Visits per Capita: Comparison between Cooperatives and Type III and IV Clinics, Costa Rica, 1990­99 Variable General Specialist Dental Nonmedical Emergency First-time Type III -0.277 -0.766 -0.109 -0.089 -0.170 -0.313 (2.47)* (7.22)** (5.73)** (4.65)** (3.38)** (4.54)** Cooperative 0.313 -0.360 0.109 0.001 -0.051 -0.068 (3.55)** (3.59)** (5.88)** (0.02) (1.03) (0.94) Management contract 0.115 -0.114 0.031 0.022 -0.129 -0.010 (0.57) (0.76) (0.88) (0.40) (1.00) (0.09) Lagged mortality 0.024 0.061 0.024 0.023 0.030 0.046 (0.53) (1.68) (2.86)** (2.73)** (1.14) (1.88) Lagged infant mortality -0.027 -0.017 -0.003 -0.006 0.017 -0.014 (2.78)** (2.00)* (1.87) (2.04)* (3.53)** (2.49)* Observations 198 198 198 144 181 198 R2 0.14 0.36 0.45 0.20 0.23 0.18 Note: Robust t-statistics are in parentheses. All estimations include annual indicator variables and a constant. *significant at 5%; ** significant at 1%. Source: Authors' calculations based on constructed database of CCSS statistics. The estimations showed that the cooperatives conducted an average of 9.7­33.8 percent more general visits (95% confidence interval), 27.9­56.6 percent more dental visits, and 28.9­100 percent fewer specialist visits. The number of nonmedical, emergency, and first- time visits per capita were not different from the traditional public clinics. These results sug- gest that the cooperatives substituted generalist for specialist services and offered additional dental services, but did not turn away new patients, refuse emergency cases, or substitute nurses for doctors as care providers. The findings are consistent with an interpretation in which cooperatives reduce costs by reducing specialist referrals but in which professional norms and government oversight keep them from cutting some of the most essential services. Table 4.4 presents estimation results for the determinants of the number of sick days and maternal disability days authorized per visit, the number of lab exams and medications provided per visit, and real expenditures per capita. Cooperatives authorized 30.4­60.5 per- cent fewer sick days (95 percent confidence interval), conducted 24.7­37.2 percent fewer lab exams, and gave out 26.7­38.3 percent fewer medications per visit than the traditional public clinics. (There is no charge for medications at either private or public clinics.) Real total expenditure per capita in cooperatives, reflecting the average expenditure by the clinic to treat all health care needs of an individual member, was 14.7­58.9 percent lower than in traditional clinics. The number of maternity days authorized per visit did not differ between cooperatives and the traditional clinics, although the cooperatives did authorize fewer sick days. The results suggest that cooperatives achieved cost savings by reducing the techno- logical intensity of health care services provided. The data at hand do not permit a judgment on how many of the additional medications and lab services provided in traditional clinics were unnecessary. The data do suggest, again, that residual claimant status provided effec- tive incentives for the cooperatives to rationalize care in ways that reduced costs. 126 World Bank Working Paper Table 4.4. Determinants of Services Provided per Visit and Real Per Capita Expenditure: Comparison between Cooperatives and Type III and IV Clinics, Costa Rica, 1990­99 Maternity Real per capita Dependent variables Sick days leave Lab exams Medications expenditure Type III -0.131 -0.056 -0.024 0.575 -6,820.934 (6.12)** (6.05)** (0.45) (5.87)** (7.42)** Cooperative -0.120 -0.014 -0.394 -0.945 -3,062.611 (6.03)** (1.54) (9.86)** (11.02)** (3.28)** Management contract -0.022 0.010 -0.029 -0.104 -282.469 (0.64) (0.94) (0.22) (0.73) (0.19) Lagged mortality 0.003 0.003 0.044 0.030 89.320 (0.49) (0.96) (1.83) (0.67) (0.26) Lagged infant mortality -0.007 -0.003 -0.024 -0.030 -135.774 (4.06)** (3.32)** (3.75)** (2.45)* (1.63) Observations 198 195 198 198 206 R2 0.30 0.31 0.27 0.45 0.29 Note: Robust t-statistics in parentheses. All estimations include annual indicator variables and a constant. *significant at 5%; ** significant at 1%. Source: Authors' calculations based on constructed database of CCSS statistics. Conclusions Using both performance evaluation data from the CCSS contracts and a panel data set on clinic-level outputs, this chapter has outlined the development of the PHC contracting model in Costa Rica and compared the performance of health cooperatives in Costa Rica with traditional, publicly managed clinics in the social security system. Under both meth- ods, the contracted providers consistently provide better performance than traditional CCSS public clinics. Even controlling for catchment area characteristics, time trends, com- plexity of clinic, and existence of performance contracts, contracted providers conducted significantly more general practitioner visits per capita and significantly fewer specialist visits per capita than traditional clinics. The per capita rate of emergency visits and first- time patients seen did not differ from that in the traditional clinics. Another source of effi- ciency was in lab tests and medications. The cooperatives conducted about 0.4 fewer lab tests and dispensed almost one less medication per patient attended than the traditional clinics. These measures combined resulted in lower expenditures per capita: an average of 3,000 colones (or roughly 30 percent of per capita expenditure) in nominal terms, less per capita than the traditional clinics. These results suggest that the contracted providers responded to financial incentives by shifting to a less costly mode of health care delivery but are able to meet, or exceed, the standards for quality and coverage. Furthermore, user satisfaction surveys show that CCSS members are more satisfied in the contracted model. These results provide evi- dence that the contracted providers were not dramatically reducing care, despite a cap- Health System Innovations in Central America 127 itated provider payment model, nor were they discouraging new patients, to reduce costs. The findings presented suggest that the contracting model in Costa Rica is promising. Internal evaluations and additional survey work provide additional evidence of the model's success. Sixty-two percent of users surveyed favor the introduction of the private sector in the supply of services, and 68 percent of users surveyed said that waiting times were low at the cooperatives (in contrast to 55 percent at the traditional clinics). To face the challenge of increasing coverage with existing resources, the contracting model appears to offer important opportunities for the CCSS to reduce costs and increase quality. As opposed to other experiences where the cost of care under the contracted model has increased--such as in Cambodia (Loevinsohn 2003) or in Chile, where the use of pri- vate providers has increased risk selection (Homedes and Ugalde 2002)--the use of con- tracted providers in Costa Rica has been carried out at lower costs than the public model and with improvements in quality and satisfaction. The public bidding process introduced during the last phase of contracting has succeeded in lowering the per capita price by more than 30 percent (lower than the existing cooperatives and the traditional clinics) and improved access and quality through introducing longer hours and home delivery for medications and lab results, among other improvements (Cercone and Briceño 2001). The findings presented here have several policy implications for health care provision in developing countries. First, nonprofit organizations in general, and cooperatives in par- ticular, might, with appropriate incentives and regulatory framework, be able to combine advantages of public and private approaches to health care provision. Under certain con- ditions, they might be able to maintain accessibility, a sense of mission, and solidarity and efficiency in service provision. Second, Costa Rica's experiment with cooperatives put in place service agreements, audits, and performance evaluations to study the effects of the policy innovation and ensure that the contracting model was carried out in an appropri- ate regulatory framework. In experiments like this one, these regulations need to provide useful information for enforcement of provisions but, at the same time, not be so burden- some as to compromise the autonomy of the service providers or increase transaction costs. Third, the contracting model should include performance objectives for quality, coverage, and user satisfaction, and these objectives should be set equal to or higher than the per- formance objectives established for public providers. The collaboration between the private and public sectors by way of contracting out services is a strategy used at increasing rates by developing countries to deal with challenges facing the health sector. These include a rise in coverage, improvements in quality and effectiveness of services, and a higher level of efficiency. Three critical aspects are tradi- tionally identified in contracting models. The first is identification of clear objectives and targets, the second is the design of adequate incentives, and the third is the establishment of monitoring and evaluation systems. To check whether a contracted provider is sustain- able, its value must be useful to society as well as to participants. As this chapter shows, coop- eration between both sectors has worked well in Costa Rica, where contracted providers have been running for long enough to produce reliable results. Clear evidence is shown that the public-private partnership model offers important opportunities to increase provider performance within a sustainable framework. 128 World Bank Working Paper Annex 4.1 Outline of CCSS Performance Contracts General considerations Chapter 1: Objective Package of services Production targets Chapter 2: Management of services Organization Service delivery conditions Information systems Continuous quality improvement Chapter 3: Control and evaluation Evaluation procedures Monitoring Annual evaluation Chapter 4: Financing General conditions Budget limit Capitation rate Performance indicators Objectives, indicators, and methodology for evaluation: Delivery of services Quality Organization Equity in resource allocation Management capacity Relationship with specialized care Billing of services Signatures and protocol Annex 4.2 Key Indicators Included in Performance Contracts Provision of Health Services Comprehensive care: To provide comprehensive health services to the entire popula- tion based on the "Basic Supply of Primary Health Services and Comprehensive Care Standards." 1. Distribution of the population. 2. Coverage of first time visits within the year. 3. Distribution of population by age and gender. 4. Existence of the tri-annual Strategic Health Plan of Action, with social participa- tion, and based on the Situation Analysis. Health System Innovations in Central America 129 5. Coverage of Growth and Development of children under one year old, with qual- ity criteria (1%); Iron Prescription (1%); Early Capture (1%). 6. Coverage of Growth and Development of children of 1 to 6 years old, with quality criteria (0.5%). 7. Vaccination coverage with DPT, polio, hepatitis and H. influenza (0.25%). 8. Vaccination coverage with MMR (0.25%). 9. Care coverage for school-aged children, with quality criteria (0.5%). 10. Percentage of reference of problems of visual acuity in school-aged children (0.5%). 11. Plan for the promotion of healthy lifestyles for adolescents (2.5%). 12. Individual comprehensive care coverage for adolescents (2.5%). 13. Coverage of cervical cancer detection in women from 15 to 34 years old (1%). Under 20 years old From 20 to 35 years old Coverage of cervical cancer detection in women 35 years and older (1%) From 35 to 59 years old 60 years and older 14. Percentage of irregular PAP smears according to quality criteria (0.5%). 15. Prenatal care coverage, with quality criteria (1%). 16. Application of a Diagnostic Tool on the access to contraception and prevention methods for STDs and AIDS (0.5%). 17. Coverage of people with diabetes, cared for the first time in the year, including quality criteria (2.5%). 18. Coverage of people with hypertension, cared for the first time in the year, includ- ing quality criteria (2.5%). 19. Detection screening for hypertension, in the population aged 20 years and older (1%). 20. Coverage of screening for people 60 years and older, including quality criteria (1.5%). 21. Plan of promotion of healthy lifestyles in people 60 years and older (2.5%). 22. Coverage of preventive dental care for children from 2 to 9 years old (1%). 23. Coverage of preventive dental care for adolescents from 10 to 19 years old (0.5%). 24. Coverage of preventive dental care for pregnant women (1%). 25. Evaluation of the dental status of the population (0.5%). 26. Establishment of a School for Parents (1.5%). 27. Evidence of the establishment of a course on the psychophysical preparation for giving birth (1.5%). 28. Percentages of emergencies (0.66%). 29. Percentages of emergency references (0.66%). 30. Emergency per inhabitant (0.68%). 31. Laboratory tests by outpatient consultation (0.4%). 32. Drug prescriptions by outpatient consultation (0.4%). 33. Drug prescriptions per inhabitant (0.4%). 34. Waiting time for drug delivery (0.4%). 35. Radiological studies requested by outpatient consultation (0.4%). 36. Evidence of coordination between levels (1.5%). 37. Referral levels (1.5%). 130 World Bank Working Paper Quality of Services To promote and encourage the delivery of quality services, placing special emphasis on timely access by the population to health services, technical quality, and user and provider satisfaction, using the continuous improvement approach. 38. Health records with incorporated forms and complete information (2.5%). 39. Current family files (2.5%). 40. External user satisfaction. 41. Comprehensive analysis of a critical area derived from the user satisfaction survey. 42. Study of a tracer entity. 43. Analysis of infant mortality. 44. Compliance with the terms established for sending the biomedical information, with the defined quality level (2%). 45. Compliance with the terms established for sending the expense reports, with the defined quality level (2%). 46. Establishment of an Office of the Controller of Health Services (1%). 47. Progressively incorporate pertinent elements to the quality assurance of the processes of care, as established in the ISO­9000 standards. Organization Managerial Function To develop and strengthen the local managerial function that allows the adequate use of existing resources, meets goals, and obtains a positive impact on the health of the pop- ulation. To make the administrative systems operational that permit greater self-sufficiency in management. Second Level of Care % of referrals sent to the tertiary level by specialty. Total of referrals at the tertiary level by specialty times 100/total of consultations of the area by specialty. Analyses of the 5 most frequent causes of referrals at the tertiary level, by specialty. % of referrals served by specialty (1%) Total of referrals served by specialty times 100/total of consultations of the spe- cialty (1%). Analysis of the referred cases at the primary and secondary level by specialty (1%). Review of the referrals (at the primary and secondary level) prior to being sent to the tertiary level. Percentage of counter-referrals from the secondary to the primary level (1%). Educational activities at the primary level. Nº of educational activities carried out by a specialist based on the analysis of the references and inter-consultations and/or the most frequent problems cared for in the clinic (1%). Active participation of the specialists in the selected tracer and in the Infant Mor- tality analysis (0.5%). Health System Innovations in Central America 131 Percentage of counter-referrals from the secondary to the primary level (1%). Total of counter-referrals times 100/total of references. Analysis of the 5 reference causes from the primary level to the secondary level (1%). Analysis of the waiting period by specialty. Waiting period for care for cases of cytologies with moderate dysplasia and beyond. Analysis of the compliance by specialty, of including in the patient's file: the clin- ical history, diagnosis, identification of risk and plan of intervention for first time cases by specialty. Analysis of the compliance with protocols or standards in addressing the two most frequent pathologies by specialty. Bibliography Bennett, Sara, McPake, Barbara, Mills, Anne London. 1997. Private Health Providers in Developing Countries: Serving the Public Interest? London: Zed Books. Buglioli, Marisa. 2000. "El gasto no deja de crecer." Available at www.escenario2.org.uy/ numero2/buglioli (accessed March 7, 2005). Caja Costarricense de Seguro Social (CCSS). 1997. Hacia un nuevo sistema de asignación de recursos. San José, Costa Rica. Cercone, James, and Rodrigo Briceño. 2001. La cooperación pública y privada en acción: el caso de las cooperativas de servicios de salud en Costa Rica. 1988­2000. San José, Costa Rica: Sanigest Internacional. Chee, Grace. 1998. "Working with Private Providers to Improve the Delivery of Priority Services." Partnership for Health Reform. Available at www.phrproject.com (accessed March 7, 2005). Departamento de Estadística, CCSS. Several years. Anuario Estadístico. Dirección Actuar- ial y de Planificación Económica, Presidencia Ejecutiva. San José, Costa Rica. Durán, Valverde, Fabio Muñoz, Daniel Corea, and Elizabeth Rodríguez Jiménez. 1995. "Una evaluación económica y financiera de las modalidades de gestión de servicios ambulatorios de salud mediante Cooperativas." Departamento de Estudios Económi- cos, Dirección Actuarial y de Planificación Económica, Presidencia Ejecutiva, Caja Costarricense de Seguro Social, San José, Costa Rica. Eichler, Rena. 2002. "Adquisición estratégica en Haití para mejorar los sistemas sanitarios: ¿Por qué pagar por inversiones cuando se puede pagar por resultados?" Paper presented at Foro de Europa y las Américas sobre Reforma del Sector Salud, Málaga, May 24­27. Eichler, Rena, Paul Auxila, and John Pollock. 2001. Performance-Based Payment to Improve the Impact of Health Services: Evidence from Haiti. Washington, D.C.: World Bank Institute Online Journal. England, Roger. 1997. Contracting in the Health Sector: A Guide to the Use of Contracting in Developing Countries. London: Institute for Health Sector Development. Gauri, V., J. Cercone, and R. Briceño. 2004 "Separating Financing from Provision: Evidence from 10 years of Partnership with Health Cooperatives in Costa Rica." Health Policy and Planning 19(5):292­301. 132 World Bank Working Paper Giustu, D., B. Criel, and X. De Bethune. 1997. "Viewpoint: Public versus Private Health Care Delivery: Beyond the Slogans." Health Policy and Planning 2(3):192­98. Hebrang, A., N. Henigsberg, V. Erdeljic, S. Foro, V. Vidjak, A. Grga, and T. Macek. 2003. "Privatization in the Health Care System of Croatia: Effects on General Practice Acces- sibility." Health Policy and Planning 18(4):421­28. Herrero, Villalta y Asociados. 1992. Diseño de la estructura de costos y estructura tarifaria para los servicios contratados con las cooperativas autogestionarias de salud.San José, Costa Rica: Comisión de Enlace, Gerencia Financiera y Gerencia Médica, Caja Costarricense de Salud. Homedes, N., and A. Ugalde. 2002. "Health Services Privatization: The Experiences of Chile and Costa Rica." Gaceta Sanitaria 16(1):54­62. Ibern, Pere. 1999. Incentivos y contratos en los servicios de salud. Barcelona: Springer. Instituto Nacional de Estadísticas y Censos (General Department of Statistics and Census). Various years. Mortalidad infantil y evolución reciente. Boletín Anual. Area de Estadís- ticas Continuas. Unidad de Estadísticas Sociodemográficas. San José, Costa Rica. Koppel, A., K. Meiesaar, H. Valtonen, A. Metsa, and M. Lember. 2003. "Evaluation of Pri- mary Health Care Reform in Estonia." Social Science and Medicine 56(12):2461­66. La Forgia, Gerard, and Roberto Gutiérrez. 2000. "Las dificultades del cambio: el modelo de seguridad social nicaragüense en transición." Paper presented at Foro de Europa y las Américas sobre Reforma del Sector Salud, San José, Costa Rica, May 24­26. Lòpez-Casasnovas, Guillem. 2003. El interfaz público privado en sanidad. Barcelona: Masson. Loevinsohn, Benjamin. 2003. "Contracting for the Delivery of the Primary Health Care in Cambodia: Design and Initial Experience of a Large Pilot Test." Working Paper. World Bank, Washington, D.C. Mills, Anne. 1998. "To Contract or Not to Contract? Issues for Low and Middle Income Countries." Health Policy and Planning 13(1):32­40. Ministry of Health, Information Center. 2000. "Evolución de la mortalidad infantil en Costa Rica. 1988­1998." San José, Costa Rica. Mintz, Patricia, Gerard La Forgia, and William Savedoff. 2001. "Contracting Health Services: Getting from Here to There." World Bank, Washington, D.C. Miranda, Guido. 1995. "La salud en Costa Rica." En Estado de la Nación en Desarrollo Humano Sostenible. San José, Costa Rica: Programa de Naciones Unidas para el Desarrollo. Musgrove, Phillip. 1996. "Public and Private Roles in Health: Theory and Financing Patterns." Health, Nutrition, and Population Discussion Paper. World Bank, Washington, D.C. Newbrander, William, Carlos J. Cuella, and Barbara K. Timmons Barbara K. "The PROS- ALUD Model for Expanding Access to Health Services." Management Sciences for Health, Inc., Boston, Mass. Nieves, Isabel, and Gerard La Forgia. 2000. "Contratación gubernamental de ONG a gran escala para extender los servicios básicos de salud a las poblaciones pobres en Guatemala." Paper presented at Foro de Europa y las Américas sobre Reforma del Sector Salud, San José, Costa Rica, May 24­26. Ovretveit, John. 1995. Purchasing for Health. Buckingham: Open University Press. Picado, Gustavo. 1999. "Gestión de las cooperativas de salud: Una evaluación con base en algunos indicadores financieros y de producción." Documento de Trabajo de la Direc- ción Actuarial y de Planificación Económica. Departamento de Estudios Económicos, Caja Costarricense de Seguro Social, San José, Costa Rica. Health System Innovations in Central America 133 Palmer, N., A. Mills, H. Wadee, L. Wilson, and H. Schneider. 2003. "A New Face for Private Providers in Developing Countries: What Implications for Public Health?" Bulletin of the World Health Organization 81(4):292­97. Pezza, P. E., and J. F. Barquero Bolaños. 1994. "Clinica Tibas: a public-private partnership for health care delivery in Costa Rica." American Journal of Public Health 84(2):315­16. Precker, Alexander, and April Harding. 2000. La economía de los roles público y privado en salud. Washington, D.C.: Health, Nutrition and Population and World Bank. Puig-Junoy, Jaume. 1999. "Manejando los incentivos de selección de riesgo en las refor- mas a los sectores de salud." International Journal of Health Planning Management 14(4):287­311. Reich, Michael. 2002. Public-Private Partnerships for Public Health. Harvard Series on Population and International Health. Cambridge, Mass.: Harvard University Press. Rodríguez, Elizabeth. 1999. "Costo financiero de la modalidad de atención de servicios de salud mediante cooperativas: un análisis de su evolución reciente." Documento de Trabajo de la Dirección Actuarial y de Planificación Económica. Departamento de Estudios Económicos, Caja Costarricense de Seguro Social, San José, Costa Rica. Slack, Catherine, and William Savedoff. 2001. "Public Purchaser-Private Provider Contract- ing for Health Services: Examples from Latin America and the Caribbean." Sustainable Development Department Technical Papers Series. Inter-American Development Bank, Washington, D.C. Taylor, Rob, and Simon Blair. 2002. "Public Hospitals. Options for Reform through Pub- lic-Private Partnerships. Public Policy for the Private Sector." Note Number 241. World Bank Group, Private Sector and Infrastructure Network. Available at http:// rru.worldbank.org/Documents/PublicPolicyJournal/241Taylo-010802.pdf (accessed March 7, 2005). United Nations Development Programme. 1990­2000. Estado de la Nacion. San José, Costa Rica. Vining, Aidan, and Steven Globerman. 1999. "Contracting-Out Health Care Services: A Conceptual Framework." Health Policy 46(2):77­96. Vining, A. R., and D. L. Weimer. 1990. "Government Supply and Government Production Failure: A Framework Based on Contestability," Journal of Public Policy 10:1­22. Widdus, Roy. 2001. "Public-Private Partnerships for Health: Their Main Targets, Their Diversity, and Their Future Directions." Bulletin of the World Health Organization 79(8):713­20. World Bank. 1995. Bureaucrats in Business: The Economics and Politics of Government Ownership. New York: Oxford University Press. Wouters, Anne Marie. 1998. Métodos alternativos de pago a proveedores: incentivos para mejorar la prestación de los servicios de salud. Bethesda, Md.: Partnerships for Health Reform. CHAPTER 5 The Nicaraguan Social Security Health Insurance Scheme-- A Promising Work in Progress John L. Fiedler and Gerard M. La Forgia Abstract In 1993, the Nicaraguan Social Security Institute (Instituto Nicaragüense de Seguridad Social [INSS]) introduced a new health care financing and services delivery scheme, the Previsional Health Model (Modelo de Salud Previsional [MSP]). The MSP is a competitive, managed-care model that pays a fixed, monthly capitation fee to certified public and private sector firms to provide a defined, basic package of health services to social security affiliates and their eligible dependents. It constitutes a new approach that marks a decisive break from the INSS's traditional mode of operation and the way in which most social security health systems in Latin America function. The coverage of the MSP has expanded slowly and is now approaching a half million persons, 8 percent of the national population. It provides more than 40,000 hospital- izations and more than 2 million outpatient visits. Provisional health firms (empresas médicas previsionales [EMPs]) provide affiliates an average of more than five medical consultations annually, and the affiliates' hospitalization rate is 98 per 1,000. In comparison to the Ministry of Health, the outpatient consultation rate of the EMPs is 10 times higher, their hospitalization rate is 30 percent higher, and their hospitalization rate exclusive of births is 60 per- cent higher. Satisfaction surveys testify that the EMPs have achieved generally high levels of satisfaction among both patients and physicians. This chapter provides a case study of the structure, operations, performance, evolution, and impact of the MSP. The goal is to provide an understanding of the model and how it was implemented to serve as a model for other countries to con- sider adopting. 135 136 World Bank Working Paper I n 1993, the Nicaraguan Social Security Institute (Instituto Nicaragüense de Seguridad Social [INSS]) introduced a new health care financing and services delivery scheme, the MSP. The MSP is a competitive, managed-care model that provides a fixed, monthly capitation fee to certified public and private sector firms to provide an explicitly defined, basic package of health services to social security affiliates and their eligible dependents. This chapter provides a case study of the structure, operations, performance, evolution, and some measures of impact of the MSP. The MSP represented a radical shift in how health care is financed, organized, and pro- vided in Nicaragua. The introduction and evolution of the model reflect both contextual fac- tors as well as trends in health system reform evident in the 1990s. We turn first to context. The MSP became a reality because it was not inserted into, nor did it replace, an exist- ing system. In the early 1990s, the INSS did not possess a direct delivery system. The San- dinista regime (1978­92) had created a unified system in which all financing and inputs were unified. All public funding (including payroll taxes) for health care was channeled through a centralized system in which medical facilities were owned and operated by a single state institution, the Ministry of Health (Ministerio de Salud [MINSA]). MINSA directly purchased all inputs, including (salaried) personnel and supplies, through bud- getary allocations. With the end of the Sandinista regime in 1992, the new government sought to explore alternative models for social insurance-financed health care more in line with its focus on developing a market economy. After regaining institutional control over the use of funds derived from the health portion of the payroll tax, the INSS explored strategies, organizational structures, and institutional arrangements with the intent of establishing an alternative system. What emerged was a compromise model that repre- sented profound institutional and structural change while retaining direct links to MINSA. This chapter describes the major features of the model, how they changed over time, and their impact on performance. The MSP is a resource allocation and purchasing arrangement that features key health system functions related to financing, risk protection, purchasing, provision, and over- sight. It contains reform elements common to reforms implemented elsewhere in publicly financed and social insurance systems pertaining to the introduction of market mecha- nisms such as competition, contracting, consumer freedom of choice, and partnership with the private sector (Organisation for Economic Co-operation and Development 1994; Salt- man and Figueras 1997; Slack and Savedoff 2001; Schieber 1997). It is another example of an emerging managed-care model developed in Latin America (such as Colombia's Health Promotion Entities, Uruguay's Collective Institutions of Medical Assistance, and Argentina's Social Works) in which medical care organizations are contracted by payroll tax-financed social insurance institutions and receive a capitated payment to provide a defined benefit package to specific populations (Medici and others 1997). These organizations generally focus on service provision but also assume an insurance function because they are at finan- cial risk for the cost of providing the benefit package. This chapter first describes the analytic framework used to assess the MSP. The second section provides an overview of Nicaragua's health and the health care delivery system. It then describes the MSP's design, structure, and performance, and assesses its impact on MINSA and the entire health sector. The chapter concludes with a discussion of lessons learned from this innovative and still evolving program, and considers its potential replic- ability in other countries. Health System Innovations in Central America 137 Analytic Framework This chapter draws on health system frameworks used to analyze insurance systems, resource allocation and purchasing arrangements, and managed care practices (Figueras, Robinson, and Jakubowski 2005; WHO 2000; Preker and others 2001; La Forgia, Griffin, and Bovbjerg 1993; Enthoven 1993). This section presents the major elements of the frame- work with a brief discussion of potential issues relevant to the Nicaraguan and MSP con- texts that are identified in the broader literature on health service purchasing. Financing How to raise (and collect) funds effectively and fairly? Similarly to social insurance systems throughout the world, the MSP is funded through a mandatory payroll tax. Issues can involve: (a) complementary financing or subsidies originating from general government revenues and directed to the social insurance system, contributing to broader system inequity; (b) the tapping of pension reserves to finance deficits in the health regimen, which suggests inefficiencies in the organization and delivery of care; (c) employer evasion, which threatens the financial stability of the entire social insurance system; and (d) degree of cost sharing with patients to address moral hazard without leaving them unprotected (Jönsson and Musgrove 1997; Mesa-Lago 1992; Baeza 2005). Benefit Package Definition What to buy? Ideally, the benefit package must be within the financial means of the over- all system, aligned with broad national health priorities and policies to maximize health status while addressing cost-effectiveness, equity, and consumer preference concerns. This is no easy task, in part because of theoretical and methodological obstacles and the absence of robust data in most countries. In practice, benefit definition is often based on an implicit process and does not follow a single method or model (Hauck, Smith, and Goddard 2005). Another benefit design issue relates to moral hazard. Because the insured tend to demand more services that they would without such coverage, insurance systems may establish exclusions for pre-existing conditions or limit the scope and coverage of services. At the same time, the package should be sufficiently comprehensive to avoid inequities resulting from "service selection" and dumping by providers. In other words, in systems with sepa- rate social insurance and government-financed delivery systems, a truncated benefit pack- age in the former can result in the selective use of the public system for more expensive care. This amounts to a hidden cross-subsidy from the public system (Puig-Junoy 1999; Lobato 2002). Membership and Enrollment For whom to buy? In developing countries, social insurance schemes generally cover com- paratively well-off segments of society while health ministries cover the poor. Issues concern the inefficiencies and inequities related to multiple-tier or segmented systems' inadequate coverage of dependents, as well as competition for scarce inputs, particularly doctors and nurses (Londoño and Frenk 1997; Gwatkin 2005; Inter-American Development Bank 1996; 138 World Bank Working Paper Mesa-Lago 1992). In systems in which potential members select a medical care organiza- tion or provider, they need unbiased information on benefits, providers, co-payments, and so forth to make a meaningful selection. Provider- or employer-based enrollment may lead to distorted information and possible provider risk selection (La Forgia and others 2004; Baeza 2005; Gold, Sparer and Chu 1998; Barry and Connors 1997). Risk-pooling and Payment Mechanisms How to distribute the financial burden fairly? How to pay? Risk pooling involves collecting revenues, grouping high-cost (high-risk) and low-cost (low-risk) individuals, and transfer- ring both revenues and individuals fairly among medical care organizations and providers. Social insurance systems generally form large, compulsory insurance pools consisting of for- mally employed workers. This reduces the probability of adverse selection. However, if the risk pool is fragmented into smaller risk pools--through permitting competition for mem- bers among medical care organizations (that combine insurance and provision functions) or providers, which in turn are also paid on a capitated basis--these organizations have an incentive to adopt risk selection and avoidance strategies, such as enrolling low-risk indi- viduals (Puig-Junoy 1999; Chollet and Lewis 1997; Smith and Witter 2005; Langenbrun- ner and Liu 2005). Although removing the incentive for overprovision, flat-rate capitation payments create incentives for providers to reduce the volume and quality of care. Calcu- lation of the capitation payment is a data-intensive exercise that is made more difficult by the weak information environments of many health systems in developing countries. Contracting Arrangements and Service Provision From whom to buy? How to contract? Government purchasing of private health services is becoming increasingly common throughout the world. At the same time, however, this practice remains controversial. Although no ideal contracting strategy can be applied to all situations, contracting works best when the purchaser has the institutional and technical capacity to select providers, specify services and outputs, monitor contract compliance, evaluate performance, and maximize competition. A favorable regulatory environment (such as few barriers to market entry) and a sufficient supply of capable providers to stimu- late competition are also considered enabling factors (Taylor 2003; Mills 1998; Montenegro Torres and Baeza 2005; Slack and Savedoff 2001). But the absence of purchasing capacity and enabling factors should not be viewed as a deterrent to contracting (see La Forgia, Mintz, and Cerezo in this volume). Contracts in which full or partial risk is transferred to providers present another set of issues. These are often known as risk and performance-based contracts and contain mech- anisms (such as withholds, bonuses, carve-outs, and so on) that hold providers account- able for costs, volume, or quality (Kongstvedt 2001a, 2001b). Finally, Palmer and Mills (2003) assert that formalistic competitive contracting may not be appropriate in some low- capacity and resource-poor settings evident in many developing countries. Palmer and Mills suggest that in these settings, relational contracting may work best because main- taining the contract is paramount to both parties and thus stimulates "the need to harmo- nize conflict and preserve the relation . . ." (p. 1007). Health System Innovations in Central America 139 Management and Oversight How to ensure results that are reliable and acceptably managed? Purchasing agencies act as an organizer of providers, guide to enrollees, and monitor of performance to ensure the delivery of high-quality care. They hold providers accountable, but need to know they are getting what they paid for. Important oversight functions include regu- lation (for example, compliance with standards and requirements); consumer advo- cacy, monitoring, and evaluation; and information collection and dissemination. However, these functions are generally weakly developed in health ministries and social insurance institutions (Slack and Savedoff 2001). Also, although utilization control, cost con- tainment, and quality management are essential functions of providers and medical care organizations, weak managerial capacity in these areas appears endemic in both the private and public sectors and thwarts the performance of reform efforts (Medici and others 1997; Kutzin and others 2001; Mills 1998; Lember 2002). Poor oversight and weak managerial capacity threatens the long-term viability of most resource allocations and purchasing arrangements, but they are particularly detrimental to ones with a market orientation (Preker and others 2001). Overview: Health and the Nicaraguan Health Care System With a per capita gross domestic product (GDP) of about US$500 in 2000, Nicaragua is the second-poorest country in the Western Hemisphere. Nearly half of the population of 5.2 million lives in poverty. Nicaragua is in the early stages of an epidemiological transi- tion, with elevated levels of infant and maternal mortality (40 per 1,000 live births and 139 per 100,000 live births, respectively), a high prevalence of infectious and parasitic diseases, and pervasive malnutrition. Although infant mortality has declined from 83 per 1,000 in 1985 to 40, it remains 33 percent above the Central American average. Infant and mater- nal mortality continue to account for the majority of all premature deaths and are due to infectious diseases, malnutrition, and birth-related mortality. Although Nicaragua devotes a higher share of GDP to health than other countries in the region, it has one of the lowest absolute levels of per capita spending. There have been three distinct eras in the recent history of health care services in Nicaragua. The first era began in 1957 with the founding of the INSS and lasted until 1979, when the Sandinista government ascended to power. During this era, the INSS owned a network of health facilities, had its own staff of health care providers, and was a direct provider of medical care--much like many other social security systems in the region. Dur- ing the second era, the Sandinista government established a single, national health system, Sistema Nacional Unico de Salud (Universal National Health System [SNUS]), and elim- inated the INSS health care system by transferring INSS facilities and personnel to the MINSA. With the establishment of SNUS, INSS beneficiaries experienced a substantial reduction in access to care and in the quality of care they received. Many INSS beneficia- ries and their employers began to decry that their social security contributions had been transformed into a nonspecific, indirect tax, and enrollment in the program fell as social security tax evasion grew. Dissatisfaction with MINSA services by the INSS insured con- tributed to widespread evasion of the payroll tax by employers. 140 World Bank Working Paper The third era began in November 1993 and was marked by the introduction of the MSP. The new system was structured so as to separate the functions of financing and health care provision. It reintroduced health care services transparently financed by the INSS, but rather than resurrecting its previous role as a direct provider of care, the INSS now con- tracted certified public and private entities--referred to as previsional medical firms (empresas médicas previsionales [EMPs])--to provide a defined set of health care services to INSS-insured workers and their dependents in exchange for a monthly fixed fee per insured worker. Separation of the financing and care functions within the new scheme was regarded as essential to ensuring that the structure of the system would provide better incentives to contain costs and provide quality care. Since its founding, the MSP has grown at a faster pace than INSS officials anticipated. Between 1994, the program's first full year in operation, and 2001: The number of INSS affiliates enrolled in an EMP grew by more than fivefold. The number of new EMPs has tripled. The number of outpatient visits provided by EMPs increased 16-fold. The number of EMP-provided hospitalizations increased 28-fold. In 2002, 16 percent of the labor force participated in the MSP, and 396,000 MSP benefi- ciaries--8 percent of the national population--were enrolled in 48 EMPs and received a total of 2 million outpatient visits and roughly 39,000 hospitalizations. The MSP has become an important and dynamic component of the health sector of Nicaragua. The Design and Structure of the MSP Enrollment Eligibility and Beneficiaries INSS programs vary by type of benefit, type of worker, and whether the worker's affiliation is mandatory or voluntary. The basic package of INSS benefits consists of the Incapacity, Old Age and Death Insurance (Invalidez, Vejez y Muerte [IVM]) and the Professional Risks Insurance (Riesgos Profesionales [RP]). More than 99 percent of INSS affiliates participate in both of these programs, the combination of which is referred to as the IVM-RP. The INSS also has a Sickness-Maternity Insurance (Seguro Enfermedad-Maternidad [SEM]). This is the program that provides medical care services through the MSP. Not all INSS affil- iates participate in the SEM, thus not all INSS affiliates are eligible to participate in the MSP.1 In 2000, 76 percent of INSS affiliates who were enrolled in the IVM-RP also con- tributed to the SEM. Workers who contribute to both the IVM-RP and the SEM are in what is referred to as the integrated regime (régimen integral [RI]), and only they are eligible to participate in the MSP by enrolling in an EMP. 1. Historically (in the pre-1979 era) only those workers who lived in a department where the INSS owned and operated a health facility, and thus had access to INSS care, were required to contribute to the SEM. Under the MSP, this criterion has been reintroduced. Because the location of EMPs extends to new departments, so too does the mandate to participate in and contribute to the SEM. The proportion of all INSS affiliates who are in the RI has grown from 63 percent in 1993 to 84 percent in 2002. Health System Innovations in Central America 141 The MSP provides coverage to the contributing worker, as well as limited health care coverage for the worker's family. If the worker is male and married, his wife is entitled to receive prenatal, childbirth, and postpartum care. Spouses of female workers are not en- titled to any health care services. Originally, the contributing worker's children were enti- tled to care only up until their third birthday. In 1997, the coverage of children was extended through five years of age. Although there are no precise data on their numbers, dependents are estimated to constitute 40 percent of total MSP beneficiaries. As is common in payroll tax-based social insurance schemes, enrollees represent a relatively well-off segment of the Nicaraguan population. Figure 5.1. Growth in EMPs' Annual Service Provision 2,500 45,000 38,232 38,919 37,075 40,000 2,000 32,543 35,000 2,048 1,903 30,000 visits 1,500 26,010 1,762 1,633 25,000 1,258 1,362 admissions 20,000 1,000 20,811 Outpatient 12,593 15,000 474 Hospital 10,000 500 605 118 6,089 5,000 1,392 0 - 1994 1995 1996 1997 1998 1999 2000 2001 2002 Outpatient visits Hospital admissions The Benefits Package The original MSP benefit package was designed by MINSA, and it primarily covered treat- ment of common illnesses and maternity care. MINSA used two principal criteria in defining the package: (1) the services had to be available from MINSA, and (2) the provision of the entire package had to be affordable by the INSS. The MSP identifies the specific pathologies included in the package by using the Tenth Version of the International Classification of Dis- eases codes and the surgical procedures, diagnostic tests, and medicines (including dosages) that are supposed to be used to treat them. High-cost procedures and expensive diagnos- tic tests generally require preapproval by the INSS. There is a lengthy list of exclusions, including dental and vision care; some specific, expensive procedures (for example, spinal and vascular surgery, dialysis, chemotherapy, rehabilitation therapy); some chronic dis- eases (for example, cancer, renal disease, certain skin diseases); and some catastrophic con- ditions (for example, severe burns). Initially, the benefit package did not include preventive services, with the exceptions of immunizations and prenatal and postpartum care. This has changed over time, as will be discussed below. Dental and vision services are still not covered. 142 World Bank Working Paper In addition to the medical care benefits, the MSP pays monetary subsidies for maternity and short-term sickness. MSP Financing and the EMP Contracting and Payment Mechanism The SEM branch of the INSS is financed by an 8.5 percent payroll tax, known as the tech- nical quota (cuota técnica [CT]). The CT is the source of revenues used to pay for both the medical care and the monetary subsidies received by MSP affiliates who are granted mater- nity or sick leave. The CT is one-half of the total INSS payroll tax. Health care agencies interested in participating in the MSP must be certified to become an EMP. To become certified, a would-be EMP must: demonstrate its capacity to deliver the benefit package--primarily by meeting INSS- established, minimal qualitative and quantitative human resource and physical infrastructure requirements;2 provide evidence that it is incorporated; demonstrate its financial viability by (a) performing a market analysis that projects revenues based on assumptions concerning the number of enrollees the EMP is expected to garner and (b) meeting a minimum capital requirement, which con- sists of presenting a bank guarantee or providing evidence that it has assets that are equivalent in value to the estimated gross revenues derived from the MSP during one month; agree to allow the INSS to retain a reserve fund of at least 5 percent of the EMP's monthly total capitation payment in a separate account to protect the EMP against higher than expected expenses and bankruptcy; agree to provide the basic package of services to all of its enrollees and their eligi- ble dependents; agree to pay the economic subsidy that the INSS pays sick and temporarily inca- pacitated, insured workers; agree to receive a specified per capita allotment as payment in full for all of the item- ized basic services demanded by the insured and their eligible dependents; sign a contract agreeing to these terms. Most of these EMP requirements are minimally structural, legal, and financial in nature. There are a few process indicators--some service delivery statistics must be reported to the INSS--but no quality or outcome indicators. The contract between the INSS and an EMP is a "take it or leave it" proposition. There are no reinsurance requirements, and no EMP uses reinsurance in managing its financial risks. The INSS uses an open contracting strategy in certifying EMPs: any organization that fulfills the requirements of certification is offered a contract. This approach supports INSS goals of expanding coverage and improving access to care, as well as encouraging the orga- nization of the private provider market. 2. The most important of these requirements is that they must have a specified minimum number of physicians and hospital beds per enrollee. Health System Innovations in Central America 143 Supervision and Monitoring of the EMPs INSS responsibilities in implementing the model include certifying would-be EMPs, mak- ing monthly payments to each insured worker's EMP, and supervising compliance with the agreement. The INSS employs 20 physician supervisors, roughly half of whom are per- manently stationed in one of the largest EMPs to monitor all MPS-related matters. The other half makes on-site visits to EMPs, generally in response to an enrollee complaint related to being poorly treated or the EMP's provision of only a portion of a covered ben- efit or refusal to provide treatment. Since 1994, the INSS has made more than 1,200 super- visory visits to EMPs. During these visits, INSS supervisors inspect the physical plant and equipment of the EMP, physicians and patients are interviewed, and medical records are reviewed. EMPs found not to be in compliance with the terms of their contract are subject to sanctions. There are three categories of sanctions--minor, serious, and severe. Minor infractions result in either a verbal reprimand or a fine of up to 15 percent of the EMP's total monthly capitation revenues. A serious infraction includes a fine plus suspension of the contract for one or two months. A severe infraction results in a fine and contract termi- nation. Since the inception of the MSP, the INSS has penalized 25 EMPs for minor infrac- tions. The 25 EMPs that have been involved in minor infractions have received a total of 31 verbal reprimands and 22 economic sanctions. Ten EMPs have been charged with seri- ous infractions, and their participation in the MSP was temporarily suspended. Seven EMP contracts have been terminated for severe violations. In 1996, the INSS conducted a consumer/patient satisfaction survey to obtain more detailed information about the operations of the EMPs and more proactively monitor the MSP. The survey was expanded and made a permanent, biannual feature of the MSP in 1998. Surveys were conducted in 1996, 1998, and 2000. (Results of the survey are discussed below.) Participating in the MSP: Selecting an EMP INSS affiliates who are in the RI are free to enroll in the EMP of their choice. Enrollees are allowed to switch EMPs once a year. How many actually exercise this right is unknown. The INSS provides brochures that explain the benefits provided by the MSP, but enroll- ment is the responsibility of the individual affiliate. The EMPs recruit and directly enroll clients and later inform the INSS. Many EMPs run marketing campaigns, advertising through billboards, radio and TV spots, and directly in workplaces. According to INSS sur- veys, the EMPs are enrollees' main source of information on the MSP, including the com- position of benefits. The MSP's Performance Growth in the Coverage of the MSP There has been significant growth in the number of EMPs, MSP beneficiaries, and EMP- provided services. The growth in the number of beneficiaries has been the result of several factors: the restoration of economic growth, growth in the proportion of all INSS affiliates who were in the RI (and who thus were obligated to contribute to the MSP and therefore were eligible to participate in it), and growth in the proportion of MSP-eligible workers who chose 144 World Bank Working Paper to participate in the program (by enrolling in an EMP).3 (See Annex Figures 5A.1-5A.4 for details.) Even though there has been substantial growth in the absolute number of MSP enrollees and MSP beneficiaries, their relative numbers--that is, the proportion of the total economically active population (EAP [the labor force]) who were INSS affiliates or MSP enrollees and the total number of MSP beneficiaries as a percentage of the population-- has remained relatively modest. As shown in figure 5.2, in 2002, all INSS programs cov- ered about 6 percent of the national population and 17 percent of the workforce, but the RI covered only 13 percent of the workforce, and only 12 percent of the workforce partic- ipated in the MSP. Figure 5.2. Relative Measures of the Growth in Coverage of the MSP 18% 17% 16% 17% 16% 16% 15% 15% 14% 14% 14% 14% 12% 12% 12% 11% 10% 9% 8% 8% 8% 7% 7% 6% 6% 6% 5% 4% 4% 4% 4% 3% 2% 2% <1% <1% 0% 1993 1994 1995 1996 1997 1998 1999 2000 2001 INSS affiliates as % EAP MSP enrolled as % EAP Beneficiaries as % population In 1994, only 7 of Nicaragua's 17 departments had one or more EMPs. Since then, the number of EMPs has more than tripled, and the number of departments in which they are located has more than doubled. Since 2000, only one department, Madriz, did not have at least one EMP. Financial Performance of the MSP Figure 5.3 presents the annual revenues and expenditures of the INSS unit that manages the MSP, the SEM branch. (See Annex Figure 5A.1 for a more detailed account of revenues 3. In 2002, 99 percent of all INSS affiliates participated in the MSP. About three-quarters of those who were not enrolled in the MSP were ineligible because, as municipal workers who are not mandated to par- ticipate and whose employers opted not to contribute, they did not pay into the program. The remainder of INSS affiliates who did not participate in the MSP contribute to the scheme but do not receive any ben- efits because they had not enrolled in an EMP. Health System Innovations in Central America 145 and expenditures.) Each year since the inception of the MSP, the revenues of the SEM have exceeded its expenditures. It is generally accepted that the relatively low outlays of the MSP are the main driver of the observed surplus of the SEM. Over the entire nine-year period, expenditures constituted 64 percent of revenues. The remaining 36 percent is a surplus. From 1994 until 2001, the cumulative total surplus of the MSP--in its entirety, C$1,293 million was transferred to the underfunded pension branch to subsidize chronic deficits. Since the start of 2002, the practice of commingling the funds of these two branches of the INSS has stopped, and the MSP no longer cross-subsidizes the pension branch. Figure 5.3. Revenues, Expenditures, and Surplus of the INSS Sickness and Maternity Insurance Branch 900 800 s 700 245 294 600 córdoba 258 500 144 nominal 400 186 of 300 97 571 79 471 70 200 361 386 Millions 105 279 224 100 171 197 75 0 1994 1995 1996 1997 1998 1999 2000 2001 2002 Expenditures Surplus Total payments made to EMPs have annually averaged only 46 percent of total expen- ditures since the program began. The SEM's other expenditures have been split among administrative costs, monetary benefits, and other services and have averaged 11, 6, and 1 percent, respectively. In comparison to spending in the pre-Sandinista direct delivery model, the MSP appears to be a bargain. Comparing two five-year periods, 1974­78 and 1994­98, real per enrollee spending was 39 percent less in the latter era (US$167 compared to US$274). As Table 5.1 shows, there have been three increases in the capitation rate. None of the three adjustments that have been made in the per capita payment were based on cost or actuarial analyses, and the increases have not kept pace with inflation. In real terms, the current nom- inal capitation rate of C$169.93 is equal to 83 percent of the original, 1994 rate of C$97.00. The Service Delivery Record As noted, the total number of outpatient visits and hospitalizations provided by the EMPs has grown steadily since the start of this new program. As Figure 5.4 shows, even when the 146 World Bank Working Paper Table 5.1. MSP Monthly Per Capita Payment to EMPs per Enrolled Worker In nominal In real (1994) Real 1994 In nominal Year córdobas córdobas level (%) US$ 1994 97.00 97.00 100 14.43 1995 97.00 87.44 90 12.88 1996 113.00 91.25 94 13.40 1997 113.00 83.55 86 11.96 1998 113.00 73.91 76 10.68 1999 130.00 76.46 79 11.01 2000 130.00 68.54 71 10.25 2001 169.93 83.46 86 11.33 2002 169.93 80.79 83 12.19 Source: Authors' calculations; Nicaraguan Central Bank data from the World Bank country data Web site. growing number of beneficiaries is taken into account (by calculating the average num- ber of services provided per beneficiary), the service provision performance of the EMPs is impressive. From 1994 through 1997, the average number of visits and the average number of hospital discharges per beneficiary increased each year. Since then, these utilization rates have stabilized. In 2002, the average MSP beneficiary had 5.2 out- patient visits and 10 percent of all EMP beneficiaries had hospitalizations. These are high rates of use. According to the World Health Organization (WHO), the minimum acceptable con- sultation rate for a country is 2.0--that is, on average, each person should have at least two Figure 5.4. Evolution of Annual Average EMP Use Per Beneficiary (Affiliates and Eligible Dependents) 7 120 6.5 108.8 104.2 105.5 98.2 98.4 6 109.0 100 per 4.7 5 5.8 5.6 5.4 5.2 80 3.8 75.4 4.9 visits 4 60 3 47.5 admissions/1,000 beneficiary 1.5 40 2 beneficiaries/year Outpatient 1 20 17.8 Hospital 0 0 1994 1995 1996 1997 1998 1999 2000 2001 2002 Outpatient visits Hospital admissions Health System Innovations in Central America 147 consultations annually. Table 5.2 provides some points of comparison between the care provided by the EMPs and that provided by MINSA. As may be seen in the top portion of the table, MINSA provides roughly the WHO-recommended minimum annual average number of consultations per person. There is a marked and growing difference between the performance of the MSP and MINSA. In 1995­96, the average number of consultations per beneficiary of the MSP was roughly twice that of MINSA. Since then, the difference has increased, to a factor of three or more.4 The hospitalization rate of MSP enrollees, too, is higher than MINSA's. After growing rapidly from 1994 to 1997, it has stabilized at about 40 percent above the MINSA level. Two other differences in the patterns of outpatient care provided by the MSP and MINSA are evident in Table 5.2. The proportion of all visits that are first visits for the treat- ment of a particular illness episode (a crude indicator of the quality of care) is consistently about twice as high in the MSP compared to MINSA, suggesting that the EMPs do a bet- ter job of resolving health problems in a single visit. The other major difference is in the average number of prescriptions per visit: the EMPs provide twice as many filled prescrip- tions per outpatient visit. In assessing these service delivery levels, it is important to consider the nature of the financial incentives of the MSP and MINSA. In the MSP, the financing and care provision functions reside in two distinct entities, and the providers--the EMPs--are paid a fixed, capitated rate for providing all covered services to their enrollees. Thus the EMPs have incentives both to husband their resources, and (to the extent that their enrollees are aware of qualitative differences in the care provided by different EMPs, and resign from those they perceive as providing poorer-quality care to enroll in those they regard as good care providers) EMP providers are motivated to provide high-quality care. In contrast, MINSA providers are less financially motivated to: (a) be concerned about whether they are providing too much or too little care or (b) be concerned (beyond their professional commitment) with pleasing their patients and providing high-quality care. MINSA providers are assured of being paid their same salary regardless of how well or how poorly they perform their jobs. The different incentive structure of the MSP, which encourages a more patient-focused approach to providing care, motivates caregivers to provide care more efficiently, and provides for a market-based (as opposed to a more personalized and bureaucratic-based) type of accountability. The introduction of an incentive structure that would fundamentally change health care providers' motivations and encourage them to be more cost conscious was a key consideration in the design of the MSP. The only notable difference in the two institutions' provision of inpatient care is that the MSP's average length of stay is about half of MINSA's. This is consistent with the incentives 4. Experience in other countries suggests that new enrollees in a capitated system commonly have higher than average utilization rates. Generally, it is thought that such persons have "pent-up" demand for health care that they satisfy once they have joined the capitated provider. The temporary, abnormally high utilization rate of new enrollees slowly decreases, and after two or three years, it reaches a lower, longer-term utilization level. It is also likely that the track of annual utilization rates of the MSP enrollees reflects "moral hazard"--that is, the tendency of persons with insurance, and those with more compre- hensive insurance, to use more health care services. Table 5.2. Comparisons of Care Provided by the MSP and MINSA 148 System Indicator 1994 1995 1996 1997 1998 1999 2000 2001 2002 World Outpatient and emergency care Total visits per beneficiary per year 1.5 3.8 4.7 6.5 5.8 5.6 5.4 4.9 5.2 Bank Proportion of all visits that are first visits NA NA 43 51 47 45 42 40 40 Working Lab tests per visit 0.6 0.6 0.5 0.4 0.5 0.5 0.5 0.6 0.6 Prescriptions per visit 1.1 1.3 1.8 1.7 2.1 1.9 2.1 2.2 2.2 Emergency visits per beneficiary per year 0.0 0.1 0.2 0.4 0.4 0.5 0.5 0.5 0.5 Paper Percent of all visits that are emergency visits 3% 3% 4% 6% 3% 8% 9% 10% 9% MINSA Total visits per beneficiary per year 1.7 2.0 2.0 2.0 1.9 1.7 1.8 1.7 Proportion of all visits that are first visits 14.1 11.5 21.3 NA NA NA NA NA NA Lab tests per visit 1.1 0.6 0.6 0.6 0.6 0.8 0.8 0.9 Prescriptions per visit 1.1 0.9 0.9 0.9 1.0 1.3 1.2 1.2 Emergency visits per beneficiary per year 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 Percent of all visits that are emergency visits 12% 10% 9% 9% 8% 12% 13% 12% Inpatient care MSP Admissions per 1,000 beneficiaries per year 18 48 75 109 109 104 105 98 98 Admissions (not including deliveries) per 1,000 beneficiaries 11 31 49 79 82 79 80 75 74 Average length of stay (days) 1.4 1.6 2.5 2.5 2.4 2.5 2.3 2.3 2.3 Proportion of all admissions that are for delivery 37% 36% 35% 28% 24% 24% 24% 24% 25% Proportion of all deliveries that are by cesarean section -- -- 42% 46% 44% 46% 47% 47% 48% MINSA Admissions per 1,000 beneficiaries per year 60 70 60 82 66 73 74 70 Admissions (not including deliveries) per 1,000 beneficiaries 40 40 40 59 47 52 52 48 Average length of stay (days) 4.5 4.5 4.6 4.6 4.5 4.4 4.2 4.3 Proportion of all admissions that are for delivery 33% 43% 33% 28% 29% 29% 29% 32% Proportion of all deliveries that are by cesarean section 18% 18% 19% 18% 16% 19% 20% 19% NA not available. Note: MINSA beneficiaries are defined as 80 percent of the national population. Sources: Quintanilla 2001; INSS data; Ubilla, Espinosa, and Bitán 2000; authors' own calculations. Health System Innovations in Central America 149 provided by capitation to reduce costs. Roughly three-quarters of EMPs contract with hos- pitals to provide their enrollees with hospital care, and all EMPs require their members to obtain an authorization before being admitted. EMP Patient Satisfaction No information is available about the general level of satisfaction of MSP enrollees, but there are two sources of data on MSP patient satisfaction--the INSS-administered surveys conducted in 1996, 1998, and 2000 under the auspices of the MSP and a one-time survey undertaken in 1999 of 320 ambulatory patients of 11 EMPs.5 All four surveys found that the MSP enrollees were generally satisfied with their EMP-provided care. Patients' overall impression was favorable, with 73 percent reporting their level of satisfaction being excel- lent or good. The level of satisfaction with other aspects of care ranged from a high of 90 percent being satisfied (good or excellent) with the service to the low of 44 percent being dissatisfied (fair or poor) with their drug benefit. Since 1996, when the INSS conducted its first patient satisfaction survey, the survey instrument has evolved, with additional questions being added in each round. The 2000 version of the questionnaire consisted of 23 questions. For most questions, interviewees are asked to select from possible response categories of "satisfied," "fair/more-or-less sat- isfied," and "dissatisfied." A summary of results of all of the questions in each of the three rounds of the survey is presented in Table 5.3. Although the two different types of surveys are not directly comparable, it is note- worthy that the independent 1999 survey posted higher levels of satisfaction than the INSS- conducted survey. As Table 5.3 shows, the results of the 2000 survey show that the EMPs received high marks (79 percent or higher) on a number of other measures capturing patients' views of EMP care, including how well the physicians treat the patients (item 7 in Table 5.3), the proportion of respondents who reported being treated well in maternity (item 15) or pediatric services (item 18), the proportion of children's cases that were sat- isfactorily resolved (item 19), and the proportion of patients who reported being treated well in admission (item 4). In contrast, only 59 percent of respondents reported being satisfied with the EMP care they personally received (item 1). The apparent incongruity between the various services for which EMPs get high marks and the relatively low level of satisfaction seems to reflect beneficiaries' judgment that their EMP had been unable to resolve the respondents' own personal health problem. Other shortcomings of the EMPs appear to be their relatively low levels of coverage for children (45 percent) and the relatively high and stable proportion (17 percent) of respondents who said they had to visit another health facility to have their 5. The sample selection methodology of the two surveys differed, and neither sample was representa- tive of all MSP enrollees. The biannual survey has tracked the same 18 EMPs in each of its three waves, and its sample of interviewees is selected from a sample of 131 worksites. Potential interviewees had to have made at least one visit to their EMP in the preceding six months. The 1999 survey developed a pur- poseful sample of 11 EMPs based on criteria that would ensure highly different EMP characteristics. The 11 EMPs were visited, and 30 patients who visited the facility on the day of the survey were interviewed. Neither sample is representative of the total EMP member population, nor of the total MSP-enrolled pop- ulation. Both the way in which the sample was selected and the site of the interview probably result in an upward bias in the level of satisfaction reported in both surveys, especially in the one-time 1999 survey. 150 Table 5.3. Patient Satisfaction with EMP-Provided Care Evolution of INSS-Administered Patient Satisfaction Survey Indicators: 1996, 1998, 2000 World Patient satisfaction indicator 1996 1998 2000 1998­2000 change 1 Reported to be satisfied with EMP care -- 57% 59% + 2% Bank 2 Reported that the EMP's care resolved their health problem 70% 62% 61% -- 3 Had to obtain consultation from other than their EMP -- 17% 17% -- Working 4 Reported that they were treated "well" on admission 80% 82% 83% -- 5 Did not encounter problems being admitted -- 9% 11% + 2% Paper 6 Waiting time prior to being attended 65 min 60 min 53 min - 7 min 7 Reported they were poorly treated by the physician 7% 4% 4% -- 8 Received their complete prescription of medicines -- 20% 21% -- 9 Had to purchase medicines 50% 16% 16% -- 10 Paid for laboratory examinations -- 2% 3% -- 11 Prenatal care coverage of insureds -- 21% 20% -- 12 Prenatal care of insured was provided by a obstetrician-gynecologist -- 87% 91% + 4% 13 Female beneficiaries received a Pap smear -- 57% 63% + 6% 14 Female beneficiaries aware of their right to have a Pap smear -- 72% 80% + 8% 15 Beneficiaries reported they were treated "well" in maternity -- 90% 86% - 4% 16 Coverage of eligible children -- 60% 45% - 15% 17 Children who were attended by a pediatrician -- 70% 83% + 13% 18 Insured whose children were attended "well" 71% 78% 79% -- 19 Children's health problems were resolved -- 85% 85% -- 20 Did not have problems with subsidies -- 7% 5% - 2% 21 Satisfaction with knowledge imparted by EMP physician -- 9% 38% + 29% 22 Satisfaction with hospital care -- -- 80% -- 23 Obtaining authorization for hospitalization was "easy" -- -- 87% -- -- no change or no significant change. Source: Gerencia General de Salud Provisional­INSS, UCRESEP, CIET International (2000). Health System Innovations in Central America 151 problem adequately treated. It may be that some EMPs are discouraging at least some of their patients from seeking care. This is one of the negative consequences of the incentives provided by capitation, and it is likely to be particularly common in a capitated system in which there is little consumer-based policing or competition for clientele and in which no risk adjustments are made in the capitation rate. In assessing patients' satisfaction with the MSP, it is important to take into consider- ation the beneficiaries' relevant benchmarks. First, because the Sandinistas actively dis- couraged private practice between 1979 and 1992, MINSA was very nearly the only source of care available when the MSP was introduced in late 1993. MINSA, therefore, is likely the reference point to which EMPs are being compared. Second, the INSS's RI affiliates had received nothing for their earmarked payroll health taxes for more than a decade. It is likely that their expectations were low, although these may increase as the program matures. Household health interview surveys provide another source of data and additional insights into the nature of the comparison point, MINSA-provided care, that are probably reflected in the level of satisfaction of EMP patients. Although 16 percent of EMP outpatients bought medicines, in 1995, 40 percent of MINSApatientsdidnotreceiveanyofthemedicinestheywereprescribedandanother 15 percent received only a portion of them (David and others 1996, Table 7.14). By 1998, 50 percent of MINSA patients did not receive any of the medicines they were prescribed and one-third of those who obtained them paid an average of C$46 for them (authors' calculations based on the 1998 Demographic and Health Survey data). The mean waiting time in MINSA health centers is 78 minutes, compared to EMPs' 53 minutes (Fiedler 1999, Table 6). Seventeen percent of EMP patients reported they had to obtain additional care from other than their EMP to resolve a health problem, and 7 percent of persons who reported they sought care at a MINSA health center were refused care alto- gether (David and others 1996, Table 7.9). It is not known what proportion of MINSA patients who were refused care sought care elsewhere. EMP Providers' Satisfaction Focus group discussions with physicians have found that 63 percent are "satisfied" or "very satisfied" overall with the MSP (Ubilla, Espinosa, and Bitán 2000, table 13). In part, this is because it has increased their overall remuneration and, on average, is the source of two- thirds of the total income of physicians participating in the program (see Table 5.4). The Impact of the MSP MSP Effects on the Organization and Structure of the Health Sector One result of the introduction of the MSP has been the increased organization and changes in the structure of the private health care sector. When the MSP was first introduced, the sector overwhelmingly comprised individual private physician's practices complemented by a hand- ful of small, generally religious-affiliated hospitals. The MSP has added 48 multiphysician 152 World Bank Working Paper Table 5.4. Level, Relative Importance, and Growth of Physicians' Real EMP Incomea EMP income All other sources Total income EMP income as Year US$ Percent US$ Percent US$ Percent % of 1994 level 1994 856 65 451 35 1,307 100 100 1995 1,222 71 489 29 1,711 100 143 1996 1,569 71 632 29 2,201 100 183 1997 1,738 67 862 33 2,600 100 203 1998 2,381 63 1,374 37 3,755 100 278 1999 3,517 69 1,596 31 5,113 100 411 a. In 1999 U.S. dollars. Source: Ubilla, Espinosa, and Bitán (2000). EMPs to this mix, and it has encouraged a growing number of small (25-bed) private hospitals. Three types of EMPs can be distinguished based on the relationship between the owners of the EMP and the EMP's provider network. An EMP may be: a vertically integrated entity (that is, a group of physicians who own and operate the EMP and provide all of the services--outpatient and inpatient--in the benefit package); exclusively an administrative entity that contracts out all service provision; an organization that combines in-house and contracted care provision. About three-quarters of EMPs are the third variety, the "mixed" model. These mixed mod- els most commonly consist of a group of private sector physicians who have come together, formed an outpatient clinic, and subcontracted with a hospital to provide inpatient ser- vices. Especially in the first years of the MSP, when the size of the private sector was so small, most of the subcontracted hospitals were perforce MINSA facilities. The trend since then, however, has been toward the development of vertically integrated service delivery with newly developed, small, private hospitals providing most of the EMPs' care. There are very few EMPs that exclusively contract out both outpatient and inpatient care. There is no information on how closely EMPs monitor the provision of care by their subcontractors. Hence there is no information on how patient dissatisfaction with inpatient care may result in resignation from a given EMP, nor has there been an analysis of affiliates' resignation. Of the 48 EMPs operating in 2002, 20 were in the capital city, Managua, where roughly three-quarters of all INSS-insured enrollees were located. In nine of the departments in which there is at least one EMP, the insured have a choice among two or more EMPs (see table 5.3). The degree of market concentration among EMPs in or outside Managua (that is, the share of total members enrolled in any single EMP) has continually declined since 1994. Impacts of the MSP on MINSA: Some Good, Some Not During the first full year of the MSP, 1994, 2 MINSA facilities participated as EMPs and constituted 21 percent of all EMPs. A third MINSA facility entered the EMP ranks in 1996. Health System Innovations in Central America 153 Thereafter the number of MINSA EMPs remained unchanged until 1999, when it jumped to 8. The following year, 2000, it increased to 10, and the proportion of all EMPs that were MINSA facilities reached its zenith, 26 percent. That year, MINSA EMPs' monthly average number of affiliates reached 38,812 (10 percent of all affiliates). MINSA facilities have benefited financially from participating in the MSP, both as EMPs and as subcontractors providing hospital services to EMPs. In 2000, the MSP paid the eight MINSA facility-based EMPs nearly C$57 million (US$4.5 million). Although the MINSA EMPs constituted only an annual average of 11 percent of all EMPs, they annually earned an average of 23 percent of total capitation payments from 1997 through 2000. This is due to the disproportionately large number of affiliates of the two MINSA EMPs that were in operation throughout this period. MINSA facilities have also benefited financially from the MSP by being subcontracted by EMPs to provide inpatient services. Because only a small number of private hospital beds survived the Sandinistas' SNUS, MINSA facilities have always made up the overwhelming share of subcontractors in mixed EMP systems. Despite slow but steady growth in the num- ber of small, private hospital EMPs, MINSA facilities still constitute 77 of 91 (85 percent) of all EMP subcontractors (as of February 2002). The MSP maintains no information about the level of use of EMP subcontractors, the amounts they are paid, the mechanisms by which they are paid, or patient satisfaction with their services. The possibility that a MINSA hospital could become an EMP or that it could be sub- contracted by public or private EMPs to provide MSP care created new incentives for MINSA hospital managers. To be contracted as an EMP or subcontracted by an EMP, they must provide differentiated services (that is, offer a higher quality of service than that pro- vided to the general public). MSP-derived revenues constitute the first significant financial incentives that most MINSA hospital directors have ever had, and their response to these incentives has been highly varied. Most MINSA hospitals participating in the MSP earn the equivalent of roughly 5­15 percent of the budget allocations they receive from MINSA, but the most entrepreneurial MINSA hospital has regularly earned the equivalent of 40­50 per- cent of its MINSA-provided budget. As a result of the new incentive structure the MSP has introduced into MINSA hospitals, some of the hospitals have improved their managerial systems, skills, and the efficiency of their operations, which has probably also contributed to their using their MINSA resources more efficiently. Other MINSA hospital directors, however, have made few or no changes in their management style or systems. They have responded to the MSP-derived revenues as if they were a windfall, and have not tried to more proactively manage this potentially lucrative portion of their hospitals' operations. The impact of the MSP on MINSA, however, is not entirely positive. The financial gains from participating in the MSP are not widespread among MINSA facilities or employees. In fact, only a small proportion of facilities and employees enjoy this supple- mental income. MINSA has not adopted any rules or regulations to govern the use of these monies, and the disposition of these monies is reported to vary substantially by facility. Generally, however, it is reported that from one- to two-thirds of the revenues MINSA facilities garner from their MSP-related activities are used to augment salaries of MINSA physicians working in MINSA hospitals. (The hospitals are the only type of MINSA facil- ity participating in the MSP.) Inequality in access to this opportunity gives rise to discord and jealousies and undermines the morale and productivity of those MINSA staff who are excluded from participating in the MSP--the vast majority of them. 154 World Bank Working Paper Another adverse impact of the MSP on MINSA stems from the fungibility of MINSA resources coupled with the enticement of increased facility or personal remuneration or both. The MSP provides both the opportunity and the motivation for MINSA employees to devote disproportionate amounts of their resources--including those provided by MINSA--to the treatment of MSP patients in order to attract and maintain MSP enrollees and EMP subcontracts. To the extent that MINSA personnel do so, it results in MINSA effectively subsidizing the treatment of MSP enrollees. Even if the MSP-derived monies are not used to augment MINSA salaries, the fact that MINSA facilities are paid in cash for participating in the MSP has also probably encouraged MINSA facility directors participating in the MSP to devote disproportionate amounts of the ministry's resources to ensuring that they remain participants. MSP revenues are likely to be of disproportionately greater importance to MINSA facility directors and administra- tors (relative to MINSA-derived resources) because these revenues constitute most of the fungible resources available to them. For the most part, MINSA resources are distributed in kind, leaving relatively little room for the exercise of discretion and the timely purchase of the specific types and quantities of resources that a given facility might need. In contrast, MSP revenues allow MINSA staff to purchase exactly what they feel they need, as opposed to, at best, being able to choose from what is available from MINSA and having to accept the uncertainties of when and if it will actually be delivered. Moreover, MSP earnings are more readily available to meet unanticipated needs as they arise. These important differ- ences are likely to encourage inefficiencies within the MINSA system, because administra- tors are willing to use substantial amounts of existing MINSA resources (for example, personnel time) to obtain a much smaller value of liquid, fungible, MSP-related revenues. Doing this may be economically irrational from the MINSA system's perspective, but from the individual hospital manager's perspective, it may be entirely rational. There is yet another piece to the utilization puzzle. When EMP enrollees require ser- vices that are not included in the MSP benefit package, they must either obtain them from their EMP (or another private provider) and pay out of pocket for them, or they must obtain them from MINSA facilities, where fees are low or nonexistent. The INSS estimates that MINSA annually attends 1,500 MSP enrollees who have "complex" cases that are excluded from the MSP package. In informal discussions with MINSA staff, one often hears that these patients--who are likely to be especially expensive to treat--are "dumped" on MINSA and that this is evidence that the MSP is not paying its "full way"--that it is being subsidized by MINSA. This is an accurate portrayal. INSS staff, however, have a very different interpretation--one that is generally couched in terms of history and the evolution of the MSP. They concede that the MSP is not a completely independent financial intermediary, but go on to maintain that it has con- tributed mightily to MINSA by allowing MINSA to "off-load" a substantial number of patients, saving the ministry a significant amount of money by doing so. This too is an accurate portrayal of the MSP-MINSA nexus. According to the INSS point of view, for the most part, EMP-provided care is a substitute for MINSA-provided services, reducing the amount of care that the ministry would otherwise have to produce. If one does a simple counterfactual and assumes that the population-based, MINSA utilization rates would also characterize MSP enrollees if there were no MSP, a rough approximation of the additional services MINSA would otherwise have had to provide in 2002 is 540,000 outpatient con- sultations and 25,000 hospitalizations (Fiedler 1999). There is no information about the Health System Innovations in Central America 155 cost of MINSA services, thus it is not possible to estimate the ministry's cost savings attrib- utable to the MSP's activities--but no doubt, they are considerable.6 The INSS is sensitive to the criticism that it is being subsidized by MINSA and is attempt- ing to address this issue. In 2001, the MSP expanded the basic services package. In 2002, it began implementing a phasing-in of the extension of coverage to INSS pensioners, and in 2003, it began a feasibility analysis of incorporating informal sector workers. Thus, although the differences between these two perspectives have narrowed as the MSP's coverage has grown, whether the differences will eventually disappear remains to be seen. The Impact of the MSP on Managerial Practices and the Efficiency of the Health Sector Mention has already been made of the financial risk providers (EMPs) incur in a capitated system and of the need to control costs as a means of managing that risk. To the extent that they do so, EMPs and the MSP contribute to the improved efficiency of the health sector. Although there is no systematic information on these aspects of their contractual arrange- ments, it is generally thought that most, if not all, of the EMPs pay their subcontractors on a fee-for-service basis. This heightens the financial risk of the EMP itself. If the EMPs instead paid their subcontractors a capitation fee or a fixed amount per case, they would give the sub- contractors greater incentive to contain their costs. The EMPs reliance on fee-for-service subcontracts manifests their as yet relatively unsophisticated approach to management. EMPs' contracts with their physicians do not encourage their providers to contain costs either. Nearly all EMP physicians are paid through fixed salaries without retainers, bonuses, or other schemes that stimulate cost containment. The only systematic informa- tion that has been collected on EMPs' physician payment mechanisms comes from a 1999 survey of 46 physicians (Ubilla, Espinosa, and Bitán 2000). That survey found only one physician who was not exclusively paid a fixed salary. That physician was on a fixed salary augmented by incentives. The fact that very few EMPs have yet to introduce variable physi- cian remuneration, or use payment mechanisms that put their hospital subcontractors at risk, suggests that physicians and hospitals can be encouraged to be more cost conscious than they have been to date, and it suggests that the EMPs (and thus the MSP and the Nicaraguan health sector) still have potential efficiency gains to realize. To the extent that EMPs encourage their physicians to be cost conscious, they apparently do so by promoting utilization controls through the use of verbal exhortation and peer pressure.7 6. This is the minimum amount of care MINSA avoided having to provide, based on national popu- lation-based estimates. In contrast to the general population, MSP enrollees are more likely to (a) live in locales where they have greater access to care, (b) have higher incomes, and (c) have higher education lev- els, all of which are associated with higher utilization rates in Nicaragua (Fiedler 1999, table 2). If it were assumed that MSP enrollees would demand the same amount of care from MINSA that they obtained from their EMPs in 2000, these figures would be 3.5 times higher for outpatient visits (1.7 million) and 7.3 times higher for hospitalizations (167,000). These two sets of extreme values establish the range within which the actual figures probably lie. 7. The real value of the capitation rate has fallen by more than 20 percent since 1993, which suggests that there is growing pressure on EMPs to contain costs, yet there is little evidence that they are attempting to do so systematically. The only common utilization control mechanism used by EMPs is the requirement that would-be inpatients obtain preadmission authorization. This has been a requirement of nearly all of the EMPs since the start of the program, however, and has not been a response to their growing financial con- straints. Given that EMP physicians' real remuneration has increased (see Table 5.2) while the real capi- tation rate has fallen, it may be inferred that average EMP physician productivity has increased. 156 World Bank Working Paper There are some indications, however, that some EMPs have responded to the incentives inherent in capitation to provide care more efficiently. For instance, compared to MINSA, EMPs have lower average lengths of stay and economize on the use of laboratory tests (see items 9 and 3, respectively, in Table 5.3). The sum of the evidence, however, is ambiguous, because other indicators suggest that the EMPs provide more care or more expensive types of care than MINSA facilities. For instance, EMPs' cesarean section rates, use of emergency care, and number of prescriptions dispensed are higher than those of MINSA (see items 11, 5, and 4, respectively in Table 5.3). Other EMPs--a minority--have attempted to reduce costs and avoid risk by reducing the range of eligible services they provide to enrollees (which has prompted the MSP to take disciplinary action against a handful of them). In some capitated systems, the capitation rate is actuarially adjusted to take into account the variation in expected rates of use and costs due to variation in the mix of enrollees. In systems where capitation rates are not risk adjusted--such as in the MSP-- another approach to managing risk is to attract and retain enrollees who are likely to use below-average amounts of health care--that is, the healthy instead of the ill, men instead of women, and the young instead of the old. The MSP's policy of charging the EMP with the responsibility of providing care for not only the affiliated worker, but also his or her eligible dependents, while still paying the same fixed capitation rate per insured worker-- independent of the number of eligible dependents--also creates an incentive for EMPs to enroll affiliated workers with smaller families. Despite the absence of risk adjustment in the MSP, there is little evidence of risk selec- tion by EMPs and only anecdotal evidence of adverse selection by EMP enrollees. Analy- sis of a small sample of patients from 11 EMPs revealed that the demographic profile of enrollees (age and gender, as well as the number of dependents) varies little among EMPs and appears unaffected by EMP size, years in business, or location or whether it is public or privately owned (Ubilla, Espinosa, and Bitán 2000, p. 16). However, the small and non- random nature of the sample means that, at present, there can be no unequivocal, general statement about EMPs' risk selection. Moreover, although it would be necessary to per- form more detailed analysis than data presently available permit, a casual review of EMPs' average utilization rates of outpatient and hospital services suggests that the variation in these rates are neither minor nor random. This is an important issue that the MSP needs to investigate and continually monitor so as not to put EMPs at undue financial risk. Thus there is mixed evidence about the impact that the MSP has had on the manage- ment practices and efficiency of health care provision in Nicaragua. Moreover, there is evi- dence that the MSP is currently characterized by conditions identified as encouraging or allowing the incentive structure of capitation to adversely affect the quantity and quality of health care provided, which puts it (and its beneficiaries) at risk of being taken advan- tage of by its member EMPs. The INSS needs to more fully develop some of the MSP's structural features and must take a more proactive approach to protecting consumers and policing EMPs if the MSP is to fully deliver on its promise. Modifying the MSP to Better Ensure Quality Care and Spawn More Competition There are four means by which the negative influences of financial incentives may be kept in check: Health System Innovations in Central America 157 aggressively monitoring the quality of service provision to ensure that covered ser- vices and needed services are not withheld or the quality of care debased; monitoring patient satisfaction; providing consumers with the basic knowledge they need about their rights and responsibilities in a capitated system; particularly their ability to resign from an EMP that they feel is providing poorer-quality services and enroll in one they believe will provide better service; providing consumers with the information they need to evaluate and compare the performances of EMPs so they can make informed choices. Consumers have a potentially important role to play in monitoring EMPs and fostering competition among them. Their role and their capability to date, however, have not been adequately cultivated. For instance, what recourse do dissatisfied patients currently have? They can complain to the MSP, which might prompt a visit by the MSP and could result in some type of sanction, or they can choose to enroll in a different EMP, which would send an important market signal to the EMP. A recent survey conducted by INSS found, how- ever, that a large proportion (one-third) of enrollees are uninformed or incorrectly informed about MSP membership rules, including their right to switch EMPs, and only 44 percent could correctly identify their prescription benefit (personal communication, Dr. Jaime Fuentes, MSP Director). The MSP does not maintain information on the fre- quency with which affiliates change their EMP. It is thought, however, not to be a common occurrence. This is a fundamental structural weakness of the program and one that must be addressed if the MSP is to live up to its full potential as a consumer-focused, competi- tion-based system promoting the efficient provision of quality care. The MSP has only half-heartedly embraced its role as the monitor and supervisor of the EMP care it purchases. Although it collects some service provision data, it does not col- lect all that are needed to fulfill this function. The MSP does require EMPs to report about the health services they provide, and it has conducted three patient satisfaction surveys. However, it has yet to develop a means for monitoring the quality of care, it does not make adequate use of the data it does collect, and it does not make its monitoring data readily available to consumers. The data submitted to the MSP are published, but they have not been made widely available. The information the MSP currently collects could be used by the INSS to police EMPs--that is, to launch special investigations that might result in an EMP being sanctioned--but to date it has not been used this way. Similarly, the INSS has instituted consumer satisfaction surveys, but it has not realized the full potential of these tools to educate, inform, and empower consumers and promote competition among EMPs. Although the intent was to conduct these surveys every two years, financial constraints reportedly led to the cancellation of the 2002 round. These sur- veys are not expensive, and they should be carried out annually. Consumers need infor- mation about the quality of care and the satisfaction of EMP users, and they need it in a form in which they can easily incorporate it into their health decision-making processes and thereby force EMPs to be more consumer oriented. As long as the MSP's consumer advocacy capability remains underdeveloped, the potential for making EMPs more consumer- focused--so that they feel greater need to, and are better able to, compete for clientele-- will remain underdeveloped. This is an important issue that needs more MSP atten- tion. The MSP needs to develop structural approaches to collecting and disseminating this 158 World Bank Working Paper information, and it is well past due to begin educating enrollees about their rights and how they can get the most out of the MSP system. Another, potentially complementary, means by which to increase EMP competition would be to change the way EMPs are contracted. At present, any EMP that fulfills MSP certification requirements can potentially become a provider of care. Whether or not they do depends on whether RI affiliates choose to enroll in them. An alternative would be to introduce competitive bidding or selective contracting to determine which EMPs would be allowed to provide care, or to introduce a more stringent certification process. Some health sector analysts maintain that open or noncompetitive contracting provides little incentive to EMPs to exceed the minimal requirements, and it may be encouraging the overdevelopment of the market. This, they argue, may be fragmenting the enrollment base among an inordinately large number of EMPs, which may compromise their financial via- bility. Alleviating this situation would encourage EMPs to compete more on the basis of the quality of care, the type of insured services offered (going beyond the required mini- mum package), and the convenience of access to care they provide than they presently do. INSS officials maintain that the EMP market is still too immature for this to be a viable approach. Indeed, at present, only 3 of Nicaragua's 17 departments have more than one or two EMPs, suggesting that the introduction of such a system would be premature--there simply are not enough EMPs to provide meaningful competition, at least outside Managua. Greater encouragement of the development of EMPs is needed, especially outside Managua. Eventually, selective contracting is likely to be an attractive option, but not yet. When and if it is eventually pursued, it would probably be best to initially confine it to Managua, where the EMP market is more mature and EMPs are more numerous, larger, and perhaps better organized and managed. Before selective contracting is introduced on a national scale, how- ever, the MSP needs to get its own house in order. It needs to improve its EMP monitoring and public and consumer education functions, which will also improve competition and encourage EMPs to offer more than the basic minimum service package.8 There is no reason to leave the development of EMPs to the market. These are vehi- cles for implementing the MSP. Why not assist them in overcoming their doubts, uncer- tainties, and inexperience? There are several actions that the MSP could undertake to better ensure the continued viability of existing EMPs and encourage the more rapid develop- ment of new EMPs, thereby accelerating the expansion and coverage of the MSP: Provide financial and technical assistance for publicity campaigns to promote understanding of the MSP and catalyze development of new EMPs, especially outside Managua; inform workers enrolled in the MSP of their rights; provide consumers with the information necessary for them to police their EMP and thereby encourage EMPs to provide higher-quality care and foster EMP competition; inform the general public of its eligibility to voluntarily affiliate with an EMP for the sole purpose of purchasing health insurance. 8. Most EMPs in Managua are already expanding their service package to improve their competitive position by offering, for instance, free dental checkups or providing free care to children, aged 6 to 10 or 6 to 15, who are not covered by the INSS package. Health System Innovations in Central America 159 Aid EMPs and prospective EMPs to overcome administrative bottlenecks by pro- viding them with a set of specially designed software packages to aid them in set- ting up management systems and fulfilling their MSP reporting requirements. Develop a reinsurance mechanism or cultivate a reinsurance market for EMPs. Conduct actuarial analyses of the adequacy of the capitation rate. Conduct feasibility analyses to examine the possibility of introducing varying cap- itation rates or changes in the composition of the benefit package to rural and other areas that will otherwise not be served by EMPs. Building on Experience: Starting Simply and Learning by Doing Over the course of the nine years the MSP has been in existence, the structure and opera- tions of the program have changed substantially. The initial design of the system was sim- pler and more restrictive--especially for the consumer. As the program and EMPs have matured, developed greater administrative capability, and learned by doing, several key original features of the program have been modified to provide more coverage, more choice, and more flexibility. The most important of these have been: The coverage of the MSP was expanded in 1997 by increasing the age limitation for eligible children from less than three years of age to less than six. This resulted in an estimated increase of 16 percent in the number of beneficiaries. Initially, the choice of an EMP was tied to the workplace. Once each year, all RI- insured workers at a workplace voted for the EMP of their choice, and everyone had to enroll in the EMP that won the election. This approach was motivated by the desire to minimize the administrative requirements that a system based on indi- vidual workers' choice would entail for both the MSP and the EMPs. Effective Jan- uary 1997, RI-insured workers have been allowed to individually enroll in the EMP of their choice and to change EMPs once a year. Originally, RI-insured central government employees were not allowed to choose their EMP, but were mandated to enroll in MINSA's Policlínica Oriental in Managua. As a result, in the second semester of 1994, this EMP had 24,000 enrollees, represent- ing 38 percent of all MSP affiliates. This was a conservative approach that would min- imize the extent to which the private sector would be allowed to participate, while ensuring MINSA some of the benefits of the new system, and it represented a com- promise between the INSS and MINSA's labor unions. Subsequently, when the "experiment" came to be better understood and opposition to it had waned, cen- tral government workers were extended the right to choose their EMP (effective January 1997).9 9. When MSP officials were queried as to why most central government employees who work in Man- agua remain enrolled in the Policlínica Oriental, they indicated that most of these workers probably do not know that they now can enroll in the EMP of their choice. This is another example of the low prior- ity that the program has accorded enrollee education and the paucity of communication it has with its beneficiaries. 160 World Bank Working Paper The first patient satisfaction survey was conducted in 1996. It was also conducted in 1998 and 2000, and it is expected that the survey will be routinely conducted approximately once every two years. Although the intent was to conduct these sur- veys every two years, financial constraints reportedly led to the cancellation of the 2002 round. The phasing-in of a more restricted benefit package of care to 27,000 INSS-pensioned elderly nationwide began in 2002. In 2001, the benefit package was expanded to include additional illnesses, medi- cines, and diagnostic examinations. The introduction of more demanding certification requirements for EMPs was initiated in 2002 and completed and evaluated in 2004.10 The litany of modifications in the MSP is not likely to end here. INSS officials report that the MSP is currently investigating the possibilities of: further expanding the medicines and the diagnostic and special examinations included in the benefit package, integrating additional severe illnesses into the benefit package (with the specific intent of reducing the subsidy provided by MINSA), extending coverage to portions of the informal sector. Moreover, there is increasing discussion of the potential for expanding and fundamentally altering the role of the MSP in the future. One scenario consists of the MSP eventually becoming the administrator of a demand-side subsidy from the government of Nicaragua that would enable groups not covered by the INSS, including the poor, to "purchase" care from an EMP of their choice. The MSP could then become a sister institution of MINSA, focusing on personal health care, especially curative care, while the ministry becomes more specialized in public health, disease prevention, and health promotion. Conclusions and Lessons Learned The introduction of the MSP marked a radical departure from earlier health care delivery systems in Nicaragua. Contracting out health care provision to third-party entities and paying them a fixed capitation fee introduced incentives to contain costs and keep con- sumers (patients) satisfied. The outpatient consultation rate of the EMPs considered as a group is 10 times higher than MINSA's; their hospitalization rate is 30 percent higher; and their hospitalization rate exclusive of births is roughly 60 percent higher. EMPs also appear to provide higher-quality care, as indicated by their ability to resolve a medical problem. Sixty percent of all EMP outpatient visits are first visits, compared to less than 20 percent for MINSA. The EMPs have achieved generally high levels of satisfaction among both patients and their physicians. The system now provides coverage to more than 450,000 per- sons residing in 16 of Nicaragua's 17 departments. As figure 5.5 shows, the private sector's 10. The new system resulted in the closure of four EMPs that could not comply with the certification requirements. Health System Innovations in Central America 161 Figure 5.5. Evolution of MINSA's and the Private Sector's Curative Health Care Market Shares, 1993­98 90% 83% 77% 80% 67% 70% 60% 50% 33% 40% 23% 30% 17% 20% 10% 0% 1993 1996 1998 MINSA Private, commercial share of the total health care market has grown rapidly and significantly over the course of the 1990s.11 The MSP has contributed significantly to this trend. The MSP also appears to be effectively containing costs. Given the amount of care it pro- vides, the MSP looks to be a "good deal" for Nicaragua--especially for those Nicaraguans who are enrolled in the MSP. The cost per beneficiary of the INSS's pre-Sandinista direct delivery system (in real terms) was 54 percent greater during 1974­78 than during 1994­98 (Ubilla, Espinosa, and Bitán 2000).12 In part, this "good deal" reflects the fact that the EMPs responded to the capitated system's financial incentives by requiring that hospital admis- 11. In Figure 5.5, private is defined as the "private, commercial" sector, which includes only private physicians, private hospitals, and INSS polyclinics. (Sources are for 1993, Fiedler [1994]; for 1996, David and others [1996]; and for 1998, authors' calculations based on the 1998 Living Standards Measurement Survey database.) 12. Analyzing the changes observed between the eras before and after the introduction of the MSP is fraught with pitfalls because of changes in the INSS insurance package, technology, and the covered population's epidemiological profile. These confounding factors are responsible for at least a part of the observed changes as well. The magnitude of the relative contributions of these factors cannot be ascer- tained with available data. 162 World Bank Working Paper sions be preauthorized and reducing the average length of stay (which is about one-half that of MINSA facilities). Despite its considerable accomplishments, there remains considerable room and a compelling need for the MSP to change so as to better fulfill its role and realize its poten- tial. Now more than 10 years old, the program still covers only 8 percent of the population and 17 percent of the labor force. There is no oversight of the quality of care provided. The level of the capitation payment has been modified four times since the start of the program, but it has never fully recovered its initial real level. The relatively low per capita, when cou- pled with inadequate monitoring of the quality of care and the substantial proportion of enrollees who are unaware of their rights and benefits in the program, provides fertile ground for EMPs to abuse the system. Moreover, there is currently no adequate mechanism by which to determine if EMPs are complying with the terms of their contracts with the INSS. In 2000, a substantial proportion (17 percent) of EMP enrollees report they had to obtain care elsewhere to resolve their health problem. Furthermore, much of the satisfaction with the system probably reflects comparisons that physicians and patients alike make with the performance of MINSA during the Sandinista era. That era was characterized by a nearly nonexistent private sector, few viable options for would-be patients, and poorly paid health professionals. Given this reference point, is 59 percent of enrollees reporting that they are "satisfied" with their EMP good enough? Finally, outside Managua, there are only two departments where consumers have a choice between more than one or two EMPs in which to enroll. As this would suggest, there is still relatively little competition in the market, and EMPs have not devised payment methods to encourage either their physician staffs or their subcontracted hospital partners to be cost conscious. As a result, the impact of the MSP on the efficiency of the health care system has not been as great as might have been expected. To date, the INSS has played too passive a role in the development of this new market, and, as a result, has put consumers unnecessarily at risk of being provided less care and lower quality of care than might otherwise be the case. The INSS has been content to set up basic guiding rules and parameters to encourage growth of this new market. It has not, however, played an adequately proactive role to ensure that the negative impulses of a cap- itated system have been anticipated and are stemmed. It needs to do so before this nascent system accumulates too many negative experiences and its reputation is tarnished. Already there are elements of Nicaraguan society calling for re-establishment of SNUS. There is no reason why Nicaraguans should have to accept a system that is more equitable, but less pro- ductive, less efficient, and less promising. Replicability Considerations The MSP presents an important test case of a purchasing model-based reform. The model was developed (and continues to evolve) in the market, political, and institutional contexts of Nicaragua. As this case study has demonstrated, the effective implementation of such a reform entails a long, steep, and bumpy learning curve. It is noteworthy that the intro- duction of a purchasing model is a viable reform option that merits consideration--even in small, poor countries with weak institutional capacity and little previous experience with public purchasing of health services from the private sector--and should not be regarded as simply an infeasible strategic option in such settings. Important determinants of the probability of success and ultimately of the degree of effectiveness of the program, how- Health System Innovations in Central America 163 ever, include a combination of persistence, the development of technical know-how and the capacity to negotiate and make incremental changes without departing from the more general, underlying strategic approach. Different health care market structures provide different sets of obstacles, constraints, and opportunities. The options available to a particular country for introducing a contracting- based, competitive model through which to purchase their social security institute's or ministry of health's service delivery are dependent upon a host of political and technical factors. The most pertinent of the technical considerations are various aspects of the health care market structure, most notably the size, composition, degree of organization, and capacity utilization of the private sector. The politics, institutions, and history of a coun- try are important parameters that will shape what can and will be considered necessary, feasible, and desirable and what is an appropriate pace of reform. Still, there are some com- monalities in terms of the structure of the reform, motivating factors, and market condi- tions, as well as in terms of strategic approaches. Structure and Sequencing The basic elements of an MSP-like reform package include: 1. a member identification, enrollment, and tracking system; 2. a defined benefit package; 3. a capitation payment, with a rate (or risk-adjusted rates) based on actuarial analysis and a mechanism for regularly, periodically updating the rate(s); 4. a system for certifying health care-providing entities or contractors thereof; 5. a health services reporting system with specific, identified reporting requirements; 6. a consumer advocacy office with a consumer education and marketing department (in-house, contracted out, or some combination thereof); 7. a monitoring and review system with clearly identified and communicated sanctions for infractions; 8. a medical audit capability (in-house, contracted out, or some combination thereof). There is a logical sequence to setting up most of these components which coincides with their ordering above. The sequencing of items 5 through 7 is not likely to be as strictly adhered to, because work on these items will probably be undertaken concurrently and their complete development is likely to be a lengthy process. This is not to suggest that these items are of lesser importance. They are critical, and, as the Nicaraguan experience testifies, they can be easily overlooked or slighted, but not without adverse consequences eventually developing. Also, political or technical considerations may not allow or enable adhering to this specific sequence, or they may not allow fully addressing a particular component of the package, as may eventually prove necessary or desirable. For instance, establishing a medical audit capability--which may involve developing treatment protocols, possibly with the assistance of the national medical association--is likely to be a lengthy process. In addition, some of the activities will change substantively over the course of the roll-out of the project. For example, in the initial stages of the reform, it will be essential to have a major public information campaign run by the consumer advocacy office. Then, after the project has been established for a year or two, the emphasis may shift to this office perform- ing more of an ombudsman-like function, serving as the key point of contact between the 164 World Bank Working Paper institute and its insured. In this capacity, it will likely maintain a database of performance- related information about EMPs that it will make available to consumers. Elements of Strategy In most low- and middle-income countries (and particularly in Latin America and Africa), there is a dearth of knowledge about the private health sector. In most countries, the sector is generally relatively unorganized and atomized--that is, single physician practices are the rule, and multiphysician practices the exception. This makes the development and administration of an MSP-type competitive scheme--one that contracts out to third-party providers--a dif- ficult and slow undertaking, and one that is likely to incur high start-up costs and high fixed costs of administration. This suggests that it makes sense to start small (for example, only in the capital city), to have a relatively long planning horizon (5 to 10 years or more), to think of meansbywhichtolessenadministrativerequirements(whichwillstillconstituteabottleneck), and, in the event that EMP-like structures are to be relied upon, to actively assist them early in the effort in forming, organizing, and becoming capable of participating in the program. As has been demonstrated, much of the history of the MSP is a tale of the introduc- tion of a series of incremental changes that together reflect a general strategic approach that the INSS has followed in implementing the program. Start simply. See how things work. Keep your eye on the big picture--on where you want to get--and continually assess how changing circumstances and capabilities allow you to introduce new features. Slowly phase in modifications. This approach enables nurturing the development of greater adminis- trative capacity and being able to develop a more informed understanding of participants' behaviors and their responses to change (both past and proposed), which enables further fine-tuning of the approach. This is a good, general, strategic approach that others inter- ested in reforming their health care systems would be wise to follow. In large part because of its very different health care market structure, the case of Nicaragua is unique. The minuscule private sector circumscribed and made the establish- ment of the social security institute's purchase of care a more protracted process than is likely to be the case elsewhere. The INSS was forced by default to work with MINSA to implement the reform simply because the ministry had such an overwhelming presence in the sector, while the private sector remained small and unorganized. This is not the case in most other Latin American countries. Other countries, therefore, will have more latitude in deciding if they want their ministry of health to be an active participant in the reform. Although MINSA's participation in the MSP has created some difficulties because of the limited nature and unevenness of the participation of its staff and facilities, and has com- plicated the short-term politics of the effort, it would be advisable to encourage the min- istry of health's participation in other countries as well. This is especially true if keeping the door open to eventually extending the reform to incorporate the ministry as well (and thereby creating a single-payer system) is desirable. Enabling Factors The case of Nicaragua is unique from another perspective as well. From 1979 until the introduction of the MSP in late 1993, INSS RI affiliates received nothing for their health- earmarked their payroll taxes. The INSS was able to start at ground zero, and did not have its choices of a service delivery mechanism constrained by having to fight an entrenched Health System Innovations in Central America 165 set of service providers who, for example, with the introduction of the contracting of ser- vices would either have to be put at risk of, or would be directly put out of business by, the newly contracted service providers. In other countries, it will be much more important to perform an analysis of stakeholders to identify the other key actors that will be involved in or affected by the reform, and to understand what the impact of the reform on them will be and how they are likely to react to it. This information must be used to shape the reform so as to obviate or at least minimize opposition. Every effort should be made to design the reform in such a way as to make participation in it, at least early on, voluntary for providers and the insured. This will make it more widely acceptable and will lessen the likelihood that political opposition will coalesce early on and derail the effort. Few other countries are likely to be able to start at ground zero. Instead, most will con- front the difficult task of having to take into account the current system's vested interests. Many also have a different set of enabling conditions that provide a different set of oppor- tunities--some of which bode well for the possibility of being able to implement reform. Primary among these beneficial enabling conditions are excess capacity in the private sec- tor (often stemming from an excess supply of private physicians) and widespread dissatis- faction with a wide-ranging and significant number of the key structural aspects of existing systems.13 The existence of excess capacity means that there is opportunity for keeping the costs of the reform relatively low--saving the social security institute and perhaps the gov- ernment and general public money, while providing private physicians and health care organizations with more patients and income. Those are attractive possibilities that will engender substantial political support. In many Latin American countries in particular, great dissatisfaction with the current system has resulted in substantial proportions of social security insured and their employers paying out of pocket for private care even though they must continue to pay their health-earmarked social security tax. This level of dissatisfaction means a significant portion of the general public in these countries is likely to be keenly receptive to the idea of developing an alternative approach. In many countries (particularly the Andean countries of Latin America and the Cen- tral American republics), the private sector has been relatively slow to organize and con- solidate because it has historically been crowded out by the combination of free or low-priced ministry of health services and mandatory enrollment in social security systems with their own health care delivery systems. The recent transformation of these historically small, fragmented, and inefficient markets has created a large gap in service potential that a competitive scheme like Nicaragua's MSP can well take advantage of. Indeed, domestic and foreign, multinational, managed care organizations have recently started to recognize this and have begun making inroads in the Latin America market (Freudenheim and Krauss 1999). Private investors recognize this opportunity, now social security institutes, ministries of health, and governments need to do so, as well. 13. Even in many African countries, where there are relatively small numbers of private physicians, demand constraints result in unused private sector capacity (Fiedler and Wight 2003). A 1995 World Bank survey in El Salvador, for instance, found that (a) private physicians average only 1.2 consultations per hour; (b) 20 percent of physicians reported they could increase the number of private consultations they provide by between 50 and 100 percent, and another 20 percent stated they could increase it by more than 100 per- cent; and (c) private hospitals had an average occupancy rate of only 54 percent (Iunes 1995, pp. 12­13). Similarly, a United States Agency for International Development study in Quito and Guayaquil, Ecuador, found private, for-profit physicians to be operating at only about 2 percent of their service provision capac- ity, and less than one-tenth that of the health sector considered as a whole (Fiedler and Nelson 1996, p. 5). Annex 166 World Figur Bank es Annex Figure 5A.1. Evolution of Real GDP and the Number of INSS Affiliates, 1987­2000 Working 3 0 , 0 0 0 3 5 0 , 0 0 0 Paper 3 0 0 , 0 0 0 2 5 , 0 0 0 2 5 0 , 0 0 0 rdobas) 2 0 , 0 0 0 programs) ó c (all 2 0 0 , 0 0 0 1980 of 1 5 , 0 0 0 3 . 0 % 1 5 0 , 0 0 0 affiliates 7 . 4 % (millions - 0 . 7 % 5 . 1 % 1 0 , 0 0 0 INSS - 1 . 7 % - 0 . 2 % - 0 . 4 0 % GDP 4 . 3 % 5 . 5 % 1 0 0 , 0 0 0 of 4 . 1 % - 1 2 . 4 Real No. - 0 . 1 % 0 . 4 % 3 . 3 % 4 . 8 % 5 , 0 0 0 5 0 , 0 0 0 0 0 1 9 8 7 1 9 8 8 1 9 8 9 1 9 9 0 1 9 9 1 1 9 9 2 1 9 9 3 1 9 9 4 1 9 9 5 1 9 9 6 1 9 9 7 1 9 9 8 1 9 9 9 2 0 0 0 2 0 0 1 2 0 0 2 R e a l G D P ( l a b e l s = a n n u a l g r o w t h r a t e ) A l l I N S S a f f i l i a t e s Health System Innovations in Central America 167 Annex Figure 5A.2. The Proportion of All INSS Affiliates Eligible to Participate in the MSP and the Proportion that Participates 90% 84% 80% 80% 76% 77% 75% 72% 73% 75% 69% 67% 68% 70% 67% 64% 63% 56% 60% 53% 49% 50% 43% 40% 30% 27% 20% 10% 1% 0% 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 MSP-eligible as % of all INSS affiliates Participating in MSP as % of all INSS affiliates Annex Figure 5A.3. Growth in the Number of MSP Beneficiaries, Including EMP-Enrolled INSS Affiliates and Their Eligible Dependents 450,000 427,139 417,453 400,000 351,470 350,000 312,190 300,000 238,997 250,000 190,871 200,000 167,105 150,000 128,128 100,000 78,349 50,000 2,024 0 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 168 World Bank Working Paper Annex Figure 5A.4. Growth in the Number of EMPs, 1994­2000 50 45 40 35 30 EMPs of 25 48 48 46 40 20 38 Number 30 15 24 20 10 5 8 0 1994 1995 1996 1997 1998 1999 2000 2001 2002 Bibliography Baeza, Cristian. 2005. "Taking Stock of Health Sector Reform in Latin America: Trends and Challenges of Health Sector Reform." In Beyond Survival: Protecting Households against the Impoverishing Effects of Health Shocks. Washington, D.C.: World Bank. Barry, Patricia, and Sharon Connors. 1997. "Medicaid Managed Care Contracting Guide." Center for Health Care Strategies, Princeton, N.J. Chollet, Deborah J., and Maureen Lewis. 1997. "Private Insurance: Principles and Practices." In George J. Schieber, ed., Innovations in Health Care Financing. Washington, D.C.: World Bank. (Originally published as World Bank Discussion Paper 365.) David, V., M. Largaespada, M. J. Fredersdorff, and I. M. Fajardo. 1996. "Estudio de demanda de servicios de salud en Nicaragua." Inter-American Development Bank Project con- sultancy report. Enthoven, Alain C. 1993. "History and Principles of Managed Competition." Health Affairs 12(Suppl):24­48. Fiedler, John L. 1994. "LSMS/Nicaragua: Self-Reported Health Status and the Use of Health Care Services." World Bank consultancy report. Washington, D.C. ------. 1996a. "A Simulation Model of the Impact of the Social Security Institute's Pro- grama de Salud Previsional on the Ministry of Health's Level and Mix of Services Delivery and Its Financing Requirements." Management Sciences for Health consul- tancy report prepared for the Inter-American Development Bank. Boston. ------. 1996b. "The Efficiency, Financing and Role of the Ministry of Health of Nicaragua: A Retrospective and Prospective Analysis." United States Agency for International Development/Nicaragua Project NEPAI consultancy report. ------. 1996c. "The Privatization of Health Care Systems in Three Latin American Social Security Systems." Health Policy and Planning 11(4):406­17. Health System Innovations in Central America 169 ------. 1999. "Demand, Market Structure and the Formalizing of Ministry of Health Fees: A Nicaraguan Case Study." World Development 27(7):1173­89. Fiedler, J. L., and D. Nelson. 1996. "Private Sector Healthcare Services in Ecuador: A National Survey of Private Health Facilities" Consultancy report to the United States Agency for International Development's Initiatives Project, John Snow Incorporated, Arlington, VA. Fiedler, John L., and Jonathan B. Wight. 2003. "Privatization and the Allure of Franchis- ing: A Zambian Feasibility Study." International Journal of Health Planning and Man- agement 18(3). pattihereFigueras, J., R. Robinson, and E. Jakubowski, eds. 2005. Purchasing to Improve Health System Performance. Berkshire: Open University Press. Freudenheim, Milt, and Clifford Krauss., 1999. "Dancing to a New Health Care Beat; Latin America Becomes Ripe for U.S. Companies' Picking." New York Times, June 16. Gerencia General de Salud Previsional, Instituto Nicaragüense de Seguridad Social. 1999. Directorio nacional de empresas medicas previsionales. Managua. Gold, Marsha R., Michael Sparer, and Karyen Chu. 1998. "Medicaid Managed Care: Lessons from Five States." In Marsha R. Gold, ed., Contemporary Managed Care: Read- ings in Structure, Operations and Public Policy. Chicago: Health Administration Press. Gwatkin, Davidson R. 2005. "For Whom to Buy? Are Free Government Health Services the Best Way to Reach the Poor?" In Alexander S. Preker and John C. Langenbruner, eds., Spending Wisely: Buying Health Services for the Poor. Washington, D.C.: World Bank. Hauck, Katharina, Peter C. Smith, and Maria Goddard. 2005. "What to Buy? Revisiting Pri- ority Setting in Health Care." In Alexander S. Preker and John C. Langenbruner, eds., Spending Wisely: Buying Health Services for the Poor. Washington, D.C.: World Bank. Inter-American Development Bank. 1996. Economic and Social Progress in Latin America: Making Social Services Work. Washington, D.C. Iunes, Roberto. 1995. "A Survey-Based Analysis of the Private, For-Profit Health Sector of El Salvador." World Bank consultancy report, San Salvador. Jönsson, Bengt, and Philip Musgrove. 1997. "Government Financing of Health Care." In George J. Schieber, ed., Innovations in Health Care Financing. Washington, D.C.: World Bank (Originally published as World Bank Discussion Paper 365.) Kongstvedt, Peter R. 2001a. "Compensation of Primary Care Physicians in Managed Health Care." In Peter R. Kongstvedt, ed., The Managed Care Health Handbook. Gaithersburg, Md.: Aspen. ------. 2001b. "Negotiating and Contracting with Hospitals and Institutions." In Peter R. Kongstvedt, ed., The Managed Care Health Handbook. Gaithersburg, Md.: Aspen. Kutzin, Joseph, Ainura Ibrainova, Ninell Kadyrova, Gulaim Isabekova, Yevgeny Samyshkin, and Zainagual Kataganova. 2001. "Innovations in Resource Allocation, Pooling and Purchasing in the Kyrgyz Health Care System." Paper presented at the Department for International Development-World Bank Conference, London. La Forgia, Gerard M., Charles C. Griffin, and Randall R. Bovbjerg. 1993. "Extending Cov- erage and Benefits of Social Financing Systems in Developing Countries." Health Financing and Sustainability Project (funded by the United States Agency for Inter- national Development), Major Applied Research Paper 3. ABT Associates and the Urban Institute, Bethesda, MD. 170 World Bank Working Paper La Forgia, Gerard M., and Roberto Gutierrez. 2000. "Growing Pains: The Nicaraguan Social Insurance Health Model in Transition." In Magdalena Rosenmöller, ed., Challenges of Health Reform. Barcelona and Washington, D.C.: IESE Business School, Universidad de Navarra. La Forgia, Gerard M., Ruth E. Levine, R. Diaz, and Madalena Rathe. 2004. "Fend for Your- self: Systemic failure in the Dominican Health Market." Health Policy 67(2):173­186. Langenbrunner, John C., and Xingshu Liu. 2005. "How to Pay? Understanding and Using Payment Incentives." In Alexander S. Preker and John C. Langenbruner, eds., Spending Wisely: Buying Health Services for the Poor. Washington, D.C.: World Bank. Lember, Margus. 2002. "A Policy of Introducing a New Contract and Funding System of General Practice in Estonia." International Journal of Planning and Management 17:41­53.1997; Lobato, Ana. 2002. "Aspectos distributivos do gasto público na saúde." Preliminary study, Ministry of Health. Brasília. Processed. Londoño, Juan Luis, and Julio J. Frenke. 1997. "Structured Pluralism: Towards an Innov- ative Model for Health System Reform in Latin America." Health Policy 41(1):1­36. Medici, André, Juan Luis Londoño, Oswaldo Coelho, and Helen Saxenian. 1997. "Man- aged Care and Managed Competition in Latin America and the Caribbean." In George J. Schieber, ed., Innovations in Health Care Financing. Washington, D.C.: World Bank. (Originally published as World Bank Discussion Paper 365.) Mesa-Lago, Carmelo. 1992. Health Care for the Poor in Latin America and the Caribbean. Washington, D.C.: Pan American Health Organization and the Inter-American Foundation. Mills, Anne. 1998. "To Contract or Not to Contract? Issues for Low and Middle Income Countries." Health Policy and Planning 13(1):32­40. Ministry of Health, Nicaragua. 1997. "Producción de servicios: 1991­1996." Managua. Montenegro Torres, Fernando, and Cristian Baeza. 2005. "From Whom to Buy? Selecting Providers." In Alexander S. Preker and John C. Langenbruner, eds., Spending Wisely: Buying Health Services for the Poor. Washington, D.C.: World Bank. Organisation for Economic Co-operation and Development. 1994. "The Reform of Health Care Systems: A Review of Seventeen OECD Countries." Organisation for Economic Co-operation and Development Health Policy Studies 5. Paris. Palmer, Natasha, and Anne Mills. 2003. "Classical versus Relational Approaches to Under- standing Controls on a Contract with Independent GPs in South Africa." Health Eco- nomics 12:1005­20. Preker, Alexander S., Cristian Baeza, Melitta Jakab, and Jack Langenbrunner. 2001. "Resource Allocation and Purchasing Arrangements that Benefit the Poor and Excluded Groups." World Bank, Washington, D.C. Processed. Puig-Junoy, Jaume. 1999 "Managing Risk Selection Incentives in Health Sector Reforms." International Journal of Health Planning 14:287­311. Quintanilla, Pedro J. 2001. "Evaluación de la gestión técnica-administrativa e impactos del Seguro de Enfermedad-Maternidad 1997­2001." Gerencia General de Salud Previ- sional, Instituto Nicaragüense de Seguridad Social, Managua. Saltman, Richard B., and Josep Figueras. 1997. European Health Care Reform: Analysis of Cur- rent Strategies. Copenhagen: World Health Organization Regional Office for Europe. Health System Innovations in Central America 171 Schieber, George J., ed. 1997. Innovations in Health Care Financing. Washington, D.C.: World Bank. (Originally published as World Bank Discussion Paper 365). Slack, Katherine, and William D. Savedoff. 2001. Public Purchaser-Private Provider Contract- ing for Health Services: Examples from Latin America and the Caribbean.Washington,D.C.: Inter-American Development Bank. Smith, Peter C., and Sophie N. Witter. 2005. "Risk Pooling and Purchasing." In Alexander S. Preker and John C. Langenbruner, eds., Spending Wisely: Buying Health Services for the Poor. Washington, D.C.: World Bank. Taylor, Robert F. 2003. "Contracting for Health Services." In Private Participation in Health Services. Washington, D.C.: World Bank. Ubilla, G., C. Espinosa, and R. Bitán. 2000. "The Use of Capitation Payment by the Social Security Institute and Previsional Medical Enterprises in Nicaragua." United States Agency for International Development's Partnerships for Health Reform Project, Major Applied Research Working Paper 3, Bethesda, MD. World Health Organization (WHO). 2000. World Health Report 2000--Health Systems: Improving Performance. Geneva. CHAPTER 6 A New Dimension for Health Reform--The Integrated Community Child Health Program in Honduras Marcia Griffiths and Judith S. McGuire Abstract This chapter describes a national, community-based, public health and nutri- tion program in Honduras called AIN-C (Atención Integral a la Niñez en la Comunidad [Integrated Community Child Health Program]). The program was launched in 1990 because, although infant mortality had dropped dramati- cally in Honduras, malnutrition in young children remained virtually unchanged at a very high level. In its seven year evolution from concept to consolidated pro- gram, AIN-C moved from being a facility-based growth monitoring and inte- grated disease management program to a community-based preventive health care and nutrition program built on a foundation of monthly growth promotion, on-demand management and referral of disease in children under age five, and care of the newborn. Innovative characteristics of the program include its focus on preventing malnutrition in all children under two through tailored and nego- tiated counseling on behavior change, development of counseling messages in the community through Trials of Improved Practices (TIPS), the use of a team of volunteers in each community, and a simple but elegant information system used for supervision, community mobilization, health coverage statistics, moni- toring, and evaluation. The program currently covers almost half of the health areas in Honduras. An impact evaluation carried out in 2002 showed that AIN-C significantly improved both height and weight of participants compared to nonparticipants and that AIN-C had significantly improved key household behaviors in health and nutrition. A careful cost study, also carried out in 2002, showed that AIN-C cost US$2.73 in annual recurrent costs per child under five in participating communities or US$1.60 in incremental annual recurrent costs. Many other countries--both in the Latin America region and in Africa and Asia--have studied AIN-C and adapted the model to their own circumstances. 173 174 World Bank Working Paper T his chapter is about an innovative, community-based public health and nutrition program in Honduras called AIN-C (Atención Integral a la Niñez en la Comunidad [Integrated Community Child Health Program]). After first reviewing the justifi- cation for a program like AIN-C in health reform, the chapter reviews the history and evo- lution of the program. It then examines the program's conceptual framework, impact, and basic processes. Finally, the chapter draws attention to design innovations considered cen- tral to AIN-C's effectiveness as a key component of health sector reform. Background Two recent documents that outline goals and steps to overcome health disparities clearly show the role that a program like AIN-C can play in filling critical gaps. According to the World Bank's most recent World Development Report, public services are failing the poor in four ways: public expenditures are not spent on services needed by the poor, the money does not always reach the frontline service provider, the service providers have little incentive to provide services to the poor, and the poor may not take advantage of the services available because of lack of information, sociocultural factors, or accessibility (World Bank 2003). The Bank also points out that health services often fail to reach the poor because clients do not play an active role in tailoring services to their needs and monitoring providers. The Honduran program described in this chapter addresses all of these issues. AIN-C is a community-based service program, funded and implemented by the government (or by nongovernmental organizations [NGOs] in specific locales), that addresses the most critical health problems of poor children. The program connects the government's public health system with families through a child growth promotion pro- gram run by monitoras (community volunteers). AIN-C monitoras empower families to raise healthy children by providing monthly, tailored counseling to parents on child feed- ing, home care of common illnesses, and appropriate use of health services. Community participation is a central tenet of AIN-C. The community has responsi- bility for selecting the AIN-C team, reviewing children's progress, and solving detected problems that impede child health. AIN-C's simple but elegant information system pro- vides an objective means to measure the performance of families, community volunteers, and local health services, and it generates the quantitative evidence a community needs to identify and resolve common problems and demand resources from local authorities. The program addresses the market failures in health service delivery by overcoming informa- tional asymmetries: parents and communities now have the information they need to assess their own children's health. Two of the eight Millennium Development Goals (MDGs) will not be met unless child malnutrition is improved. The first MDG, eradication of extreme poverty and hunger, is measured in terms of underweight in children under the age of five. The fourth MDG, reduc- tion by two-thirds of under-five mortality, is also directly related to child nutrition. It has been well documented that undernutrition is a direct or underlying cause of more than half of under-five child (Pelletier 1994). By preventing healthy newborns from faltering in growth over the first 18­24 months of life, malnutrition in children younger than five and under-five mortality can be dramatically reduced. Prevention of growth faltering requires proactive Health System Innovations in Central America 175 attention to child growth, feeding practices, immunization, health and hygiene behaviors, and care of the sick child. AIN-C addresses all of these causes of growth faltering in a low- cost way at the community level, bringing to life the causal framework (see Figure 6.1) put forward by the United Nations Children's Fund (UNICEF) in 1992 for malnutrition. Figure 6.1. UNICEF Framework Malnutrition Manifestation Inadequate Disease dietary intake Immediate causes Insufficient Inadequate Insufficient health Under- household food maternal and services and unhealthy lying security child care environment causes Inadequate education Resources & control Basic causes Political and ideological superstructure Economic structure Potential resources Rationale for AIN-C-type Programs in Health Reform Active community participation, ownership, and stewardship are essential to successful democratic development. Sustainable and equitable economic development, including promoting a healthy population, must include broad-based participation in the process. Program and project experience has shown that the organized community is essential to success in water and sanitation, education, health, reforestation, watershed management, and public works. In many cultures (Indonesia, for example), there is a long history of community cooperation and action. The community can be a source of redistribution and relief (in some communities in Africa, for instance, the community maintains an emer- gency food store). And, in most developing countries, the community is looked to as the labor pool for program implementation and a source of political support and financing for 176 World Bank Working Paper many public programs. One measure of success is that the resources reach the poor. As the World Development Report 2004 points out, one way to ensure that resources reach the poor is to transfer responsibility to the clients and the community (World Bank 2003). The ideal of broad-based community participation and stewardship is rarely achieved, however. Too often, development programs are designed in the capital city or donor head- quarters and begun in communities without any local consultation. They are entirely sup- ply driven, providing goods, services, or both, yet simultaneously expecting labor, goods, and utilization in exchange. Often, these programs that reach the community level are cap- tured by local elites, seldom reaching those in need. In other cases, where community members may be more involved, they still may not participate in program governance. Community members' work is often taken for granted, with recognition given only dur- ing elections or catastrophes. The health sector needs community action at least as much as other sectors. Basic preventive health measures and appropriate use of health services by the sick require that community leaders and community-based health workers reach those who are hard to reach with education and information and mobilize and govern community action in a responsible fashion. To take action, citizen-clients need to know about immunization days, new services available at the local health post, eligibility criteria for new programs, trash disposal, sanitation, maintenance of community water sources, and emergency assistance. The community also should provide feedback to the health system. Effective epidemiological surveillance requires information from the community level rather than from facilities alone (see Table 6.1). Table 6.1. Why Community Participation Matters to Health Reform Public provision Community participation required Cold chain, vaccines, and Community organized for immunization days vaccinators Preventive health care services, Education and follow-up on basic preventive health care such as nutrition, diarrhea behaviors for the family, such as home or self-care, control, prenatal care, family appropriate use of basic supplies, and health-seeking planning, bed nets, and so on behavior Acute care facilities and staff Detection, referral, and transport of acutely ill people to acute care facilities; follow-up and compliance with drug and treatment regimens once patients return home Water and sanitation Hygiene education and maintenance of water and sanitation infrastructure systems; payment systems Environmental sanitation Organization of the community to clean up air, water, and regulations and standards; land; social pressure to dispose of waste properly garbage collection in cities Service statistics Provision of information for epidemiological surveillance, birth and death registries, and civil registry Because health sector reform touches on the broad array of systems that influence cost- effective health care provision, it must design the system around community participation in health. The community needs to be drawn in to ensure good health outcomes. Community participation facilitates health reform by getting services to needy clients, organizing demand, Health System Innovations in Central America 177 informing supply, overseeing personnel, identifying unmet needs, and educating the con- sumer. Broad-based community participation makes it more likely that health services will be accessible to the poor and needy, a key goal of health reform. However, the community and the household have not been given adequate attention in most health reform programs. At the same time, community participation depends on health reform for long-term sustainability. The community needs to be able to count on back-up from the health sys- tem, whether the support comes in the form of supplies, training, or supervision for com- munity workers or timely, high-quality health care at the facility. Although successful in several ways, an evaluation of the Honduras Rural Penetration Program in the 1980s found that the program was not as effective as it could have been because of the failure of the health system to back it up (Bossert and others 1988). Most health reforms are carried out at the central government or ministry level, and focus on changing policies and norms with an institutional perspective. Even decentral- ization and performance-based contracts generally stop at the health district level and don't reach down to the most peripheral service level. Since the early 1980s, Honduras has been reforming its health sector (albeit in fits and starts, with changes of administration and per- sonnel) and strengthening community participation as part of an integrated strategy. The country went beyond simple decentralization and has constructed more facilities closer to the population. It improved access, equity, and outcomes by developing several commu- nity programs (particularly for immunization and for detection and treatment of dehy- dration and pneumonia). The final step has been to place the program in the community with governance by the community and only support from the health services. History Honduras has made remarkable progress in both improving health outcomes and providing equitable access to health services, especially given its low per capita income. (Gross national product per capita in 2002 was US$920 [World Bank 2003].) Although the economy overall is quite inequitable, health care is equitably distributed. For instance, in the mid-1990s, the poorest quintile of the population earned only 3 percent of the income but used 21 percent of preventive health care, 17 percent of ambulatory care, and 15 percent of hospitalizations (World Bank 1997). The differences between the wealthiest and poorest quintiles in overall health care use are small, but the poor use predominantly public health services and the rich use private health services (World Bank 1997). Infant mortality in Honduras dropped from 110 per 1,000 in 1970 to 39 per 1,000 in 1990, and life expectancy increased apace. This rate of decrease was fast even for Latin America, and notable because Honduras is one of the poorest countries in the region. Between 1990 and 1995, maternal mortality fell from 182 per 10,000 to 108 per 10,000 (Walker 2003), but the decline in infant mortality has slowed since 1990. In 2000, infant mortality was 35 per 1,000. The early, sharp decline in infant mortality was attributed not so much to the amount of money invested in the health system (about US$20 per capita in 1995; World Bank 1997) but to the effectiveness of the programs (Walker 2003). Particularly effective were vacci- nation programs using community volunteers that addressed diarrhea and respiratory ill- ness (World Bank 2001b). Part of this drop in mortality was also due to declines in fertility rates from 7.5 in 1970 to 4.4 in 2001 (Walker 2003). 178 World Bank Working Paper Health, however, is more than the absence of death. Although Honduras did a good job preventing death and controlling infectious disease agents, it did not do so well on non- infectious causes of disease--for example, malnutrition. While Honduras was saving lives through immunization and timely treatment of acute infections, it was doing much less about health promotion. Malnutrition was persistently high in the 1970s and 1980s, in spite of massive food aid and a conditional cash transfer program (Programa de Asignación Familiar [PRAF]). Over the course of the 1990s, fortunately, nutrition has improved. In 1987, stunting stood at 44 percent, but it had fallen to 38 percent by 1996 and to 33 per- cent by 2001. Overall malnutrition, as measured by weight-for-age, fell from 24 percent to 18 percent in the same period (see Figure 6.2). Figure 6.2. Honduras Mortality and Malnutrition 120 Stunting 100 Infant Mortality Rate Under 5 Mortality Rate 80 centre 60 P 40 20 0 1975 1980 1985 1990 1995 2000 2005 Year The reason for the persistence of malnutrition until the mid-1990s and the recent drop is not well understood. Neither economic growth nor poverty reduction can explain the changes in the malnutrition numbers. Malnutrition is concentrated among the poor, yet increased household income does not immediately result in improved child nutrition (Haddad and others 2002). Annual economic growth between 1981 and 1990 was 2.5 per- cent, and it increased to 3.1 percent in the 1990s (World Bank 2001b). This rate of growth was dampened, however, by population growth rates, which averaged 2.9 percent annu- ally between 1990 and 2000 (World Bank 2002). The Evolution of AIN-C In 1990, the Secretariat of Health realized that malnutrition was persistent and suspected that it was the reason why mortality rates were stagnating. This commonsense suspicion was but- tressed by the international literature, which showed that more than half of under-five mor- tality globally was due to malnutrition, particularly to mild and moderate malnutrition (Pelletier 1994). Research in neighboring Guatemala showed that the effects of malnutrition in early childhood were not reversed in later life (Martorell 1990; Martorell, Khan, and Health System Innovations in Central America 179 Schroeder 1994). A close look at the growth curves of young children in Honduras showed what was happening. For the majority of children, malnutrition began at about 3 or 4 months of age, with its severity increasing progressively until the child reached about 18 months of age, by which point the child's growth was significantly retarded (see Figure 6.3). Figure 6.3. Declining Nutritional Status of Honduran Children in Standard Deviations below the Norm, Encuesta Nacional de Epidemiología y Salud Familiar 1991­92 0 3 6 9 12 15 18 21 24 ­0.2 ­0.4 ­0.6 cores) ­0.8 -sz( ­1 Weight-for-age ­1.2 ­1.4 ­1.6 Age (months) Honduran health authorities realized that children seen periodically at health centers were becoming malnourished, but the health system was not detecting it. Child growth over time reflects everything that is going on in the child's home and community environment that affects health: disease, nutrition, psychosocial stimulation, and child care. Malnutrition does not come upon a child in a matter of days, like an infectious disease. It is insidious, start- ing with a child showing signs of anorexia and gradually being reflected in slower growth. The common perception of malnutrition--bloated belly, coppery hair, and sticklike limbs-- is not typical of 90 percent of the malnutrition in the world. Contrary to common practice, nutritional status (weight-for-age at any one point in time) is not a very useful programmatic tool to detect malnutrition. Malnutrition is a dynamic process, the result of daily stresses of inadequate feeding, frequent bouts of disease (often minor diseases), and inadequate care. One cannot diagnose malnutrition in its early stage without keeping track of a child's growth (weight gain) from one month to the next. Assessing adequate weight gain is like watching the onset of an illness, from general malaise to high fever or pox. Realizing that inadequate weight gain is the first sign of a problem, Honduran health officials revised health center norms to require the detection of growth faltering in children and use that as an additional way to triage children for services. Any child who failed to gain adequate weight was subject to careful review of his or her health and home care situa- tion. Inadequate growth was key to exploring multiple causes and integrating care proto- cols as needed. A special form, a checklist of causes of poor growth, was used to guide the exploration. Using growth faltering (the failure to gain the expected minimum amount of weight) as a prompt was successful. There was a 25 percent decline in malnutrition rates in program clinics after three years.1 1. Dr. Gustavo Corrales reported this history at a World Bank Nutrition Training Course October 1998. Dr. Corrales was Director General of Health at the inauguration of the clinic-based AIN program and a consultant on the design of AIN-C. 180 World Bank Working Paper By 1994, as the secretariat was reviewing the clinic-based strategy, they recognized that it would not result in a positive impact on public health unless they could put it in the hands of the community and reach all children. The community AIN approach (AIN-C), with growth promotion at its heart, was piloted in two health areas in about a dozen communi- ties. The piloting was done in an ad hoc way, left up to local managers who combined it dif- ferently with the existing community diarrhea and acute respiratory infection (ARI) control programs. This resulted in a 1995 report on experiences but inconclusive results. A national meeting on AIN-C called for systematizing the program process and changing the techni- cal content to include just the most efficacious essentials at the community level, refining the assessment of adequate weight gain (how to quantify an adequate growth trajectory), and placing more emphasis on how to counsel and on improving feeding of young children. Simultaneous with the Honduran Secretariat of Health's interest in systematizing their community program for national expansion, the World Bank had just completed a review of worldwide experience with growth promotion programs (Griffiths, Dickin, and Favin 1996). The best practices highlighted in that review, plus the lessons learned from communities that piloted the strategy in Honduras, directed AIN-C program development. At the point of pro- gram development, the Secretariat of Health requested technical assistance from USAID's child survival project, BASICS. Over approximately an 18-month period (mid-1996 to late 1997), program procedures were worked out with a few communities and with authorities at different levels in the Secretariat of Health. Research was conducted to better understand household child feeding practices, and manuals, training guides, and educational materi- als, including a detailed set of counseling cards, were written, tested, and published. In early 1998, the program was relaunched by the Secretariat of Health in 9 of the most populated of Honduras's 47 health areas. Since 1998, the AIN-C program has expanded through sec- retariat facilities to cover communities under every health center in 24 health areas. At least eight NGOs implement AIN-C, and, in additional communities, the program is operated by a special family assistance program (PRAF) from the president's office in collaboration with the Secretariat of Health. Expansion of AIN-C has continued each year, as have technical and managerial refine- ments and additions (see box 6.1). Critical to the adjustments was revamping the child health actions to reflect more closely the World Health Organization's and UNICEF's rec- ommendations on integrated management of childhood illnesses (IMCI). In 2002, the pro- gram strengthened attention to newborns. In sum AIN-C has emerged as an integrated, basic package of care for the young child in the community. Does AIN-C Work? What Does it Cost? In 1998, the government of Honduras carried out a baseline survey for the AIN-C program. A midterm evaluation was carried out in 2000 (Van Roekel and others 2002), and a final impact survey was carried out in 2002 (Plowman and others 2004). The midterm evalua- tion showed that although AIN-C communities were more disadvantaged than control vil- lages, they had made enormous strides in knowledge, attitudes, and practice concerning common diseases of childhood, child feeding, and administration of micronutrient sup- plements and vaccines. AIN-C communities met or exceeded control communities on vir- tually all indicators (see Tables 6.2 and 6.3). Health System Innovations in Central America 181 Box 6.1. Continually Seeking Improvement In an effort to ensure that AIN-C program activities were making a difference, the U.S. Agency for International Development (USAID)­supported Basic Support for Institutionalizing Child Survival Project (BASICS) program worked with the Honduran Ministry of Health to conduct a comprehen- sive impact evaluation of the program. The baseline survey was done in July 1998, and the midterm survey was conducted in July and August 2000. One of the assumptions of the AIN-C program design is that rates for fully immunized children by 12 months would be significantly improved in AIN-C communities because of its emphasis on universal enrollment, focus on early enrollment (the first months of life), and on frequent (monthly) contact. Midterm results showed that even in a country with high immunization rates, coverage could still be improved when a community-based program such as AIN-C is functioning. In AIN-C communi- ties at midterm, vaccinations for diphtheria-pertussis-tetanus, polio, and measles were 88.1 per- cent, 87.7 percent, and 86.0 percent, respectively, compared to control communities where the rates were 83.7 percent, 82.5 percent, and 77.6 percent. The one exception was Calmette-Guérin bacillus (BCG) immunization. Coverage rates in AIN-C communities were lower than control com- munities (85 percent versus 94 percent). The explanation, that fewer women in AIN-C communities were delivering their babies in a facility, meant that AIN-C needed to get babies born at home to health care services. In response, an activity focusing on newborns was added to AIN-C. Now the monitora is responsible for visiting newborns and enrolling them in AIN-C within 48 hours of a home birth or of the mother returning home from an institutional birth. Among other activities, check- ing that the infant received BCG is fundamental. If the child is unimmunized, the monitora ensures receipt of the vaccine through active referral or notifying the nurse for her next community visit. In this way, the "hard to reach" are covered by this vital service. Table 6.2. Caregivers' Exposure and Participation and Use of Key Health Services AIN-C Control 1998 2000 1998 2000 Variable (baseline, %) (%) (baseline, %) (%) Caregivers know about the CBGP program in their 27 96 7 15 community Caregivers participate in the CBGP program in their 30 92 21 23 community Caregivers attend weighing session 3 or more times in 30 70 38 44 the past 3 months Caregivers have their children fully immunized by 62 76 65 66 the age of 12 months Caregivers have their children 4 months of age or 2 47 4 4 older take iron supplements 182 World Bank Working Paper Table 6.3. Caregivers' Key Health Promoting Practices AIN-C Control 1998 2000 1998 2000 Variable (baseline, %) (%) (baseline, %) (%) Exclusive breastfeeding of 21 39 15 13 children under 6 months of age Offering complementary 70 76 70 63 foods at an appropriate time Giving oral rehydration 32 50 30 32 solution to children with diarrhea Giving oral rehydration 37 57 36 42 therapy to children with diarrhea Giving children fluids and 21 33 17 16 continued feeding during a bout of diarrhea The impact evaluation of 2002 looked at most of the same indicators of program performance and impact as the midterm, but it also measured nutritional status. By 2002, 24 percent of the children in the control communities were participating in growth promo- tion, either in the community or at their local health facility. This confounded the evaluation design. Therefore, results of the 2002 evaluation were analyzed based on participation or nonparticipation in growth promotion programs (community-based or facility-based), regardless of original community designation of control or program. Participants in community-based growth promotion (CBGP) and AIN-C continued to be disadvan- taged on virtually every indicator of wealth and welfare (except for water source), but the behavior differences between AIN-C and control at midterm (2000) continued to be significant in 2002, when children who were not in growth promotion programs were com- pared with participants. Overall, the advantages of participation seen at midterm were maintained but did not increase. For example, the rate of exclusive breastfeeding of children under six months of age was significantly greater among participants in AIN-C than among nonparticipants (57 percent versus 42 percent), and overall child feeding was significantly better among AIN-C participants than nonparticipants: 72 percent of the former fed their children appropriately versus 60 percent of the latter. The important new information from the 2002 evaluation is in two areas--seeking care for ARI and nutritional status. The advent of the illness management module made a difference in care seeking for participants in growth promotion. Participant families sought care for an episode of ARI about one full day earlier than they did prior to the illness management activities. Furthermore, home-based treatment of diarrhea (giving increased fluids, including breast milk) was significantly higher among AIN-C participants compared to nonparticipants--51 percent versus 38 percent. In addition, appropriate home treatment of diarrhea (62 percent versus 38 percent gave oral rehydration solution or any fluids) and care seeking for diarrhea (45 percent versus 33 per- cent sought care) were significantly better among participants in AIN-C compared to non- Health System Innovations in Central America 183 participants. In addition, significantly more AIN-C participants knew the danger signs of ARI compared to nonparticipants. Far more AIN-C participants had received iron and vitamin A supplements compared to nonparticipants. Participation in growth promotion appears to result in significantly better-nourished children when controlling for socioeconomic and caregiver variables. Full participation in AIN-C was associated with significantly greater weight-for-age (P<.01) and height-for-age (P<.05). This translates into a 0.395 z-score incre- ment in weight-for-age in participants over nonparticipants. Children in facility-based growth promotion programs showed no difference from nonparticipants when controlling for caregiver and socioeconomic factors. A detailed cost analysis was also car- ried out in 2002 (Fiedler 2003). The cost Table 6.4. Comparative Costs of Nutrition study was one of the most comprehen- Programs for Children under 5 sive empirical costing studies done to date on a community health program. It Annual unit cost Program per participanta found that AIN-C cost US$1.60 per AIN-C CBGPb US$1.60 (incremental participating child under age five in cost) or US$6.80 total long-term annual incremental recurrent cost (2003) costs or US$0.44 per capita of popu- TINP (India) CBGP plus US$9.00 (1985) lation covered. The community-based selective feedingc program costs one-ninth of what the Iringa (Tanzania) CBGPc US$10.00 (1991) facility-based growth-monitoring pro- BINP (Bangladesh) CBGP US$18.00 (2001) gram costs and provided more than plus feedingd three times the coverage of child- Indonesia CBGPe US$2.00 (1985) ren under age two. Compared to pub- Food supplementationf US$46.00 (1985) lic health expenditures of US$34 per School feedinge US$12.00 (1985) capita in 2000, AIN-C's cost must be a. Dollar figures are the equivalent of the cost in considered marginal. AIN-C's cost is local currency in the year noted in parentheses. low relative to similar programs in other b. Fiedler (2003). countries and relative to food distribu- c. Gillespie, Mason, and Martorell (1996, table 5.2). d. Mason et al. (2001, table 4.3). tion programs (see Table 6.4). e. Ho (1985). f. Kennedy and Alderman (1985). How AIN-C Works Source: BINP Bangladesh Integrated Nutrition Pro- gram; TINP Tamil Nadu Integrated AIN-C works by focusing on many of Nutrition Program. the same principles applied in health reform but at the household and community level. The program uses objective data on growth to determine need. It targets services by prioritizing children under age two, particularly those beginning to falter in growth, and focusing on adequate nutrition, care, and health-seeking practices instead of expensive treatment and supplementary feeding regimes. It brings equity by insisting on 100 percent coverage of under twos, and it decentralizes decisionmaking to the community and household, recognizing that the root of many problems is local. The central focus of AIN-C is frequent contact with and precise advice to the parents of those most at risk--all children from birth to 24 months--because that is the age of extremely rapid growth and high susceptibility to environmental insults. AIN-C's premise is that most infants are born healthy but sustain most of their lifetime "injury" during the first 24 months of life. Therefore AIN-C's goal is to maintain newborn health (or recuper- 184 World Bank Working Paper Figure 6.4. Conceptual Results and Operational Framework for Community-Based Growth Promotion Programs Community-Based Growth Promotion framework Community Frequent contact Caregiver adoption Improved adopts and early of health-promoting Improved nutritional CBGP detection of practices and child status and problems timely care seeking growth and reduced Results health mortality Community identifies and Community takes collective action to address supports a team of volunteers problems affecting child growth. and encourages all families to have under 2s weighed each Volunteers weigh children to month and all sick under 5s detect faltering, counsel mothers, Volunteers inform taken to volunteers for care. treat or refer (or both) sick community about children, and make home visits progress and support health system. (participation, child framework programs. growth). Trains, supervises and provides material support for volunteers Provides health services in the Monitors program community (for example, progress (participation, immunization, micronutrient Operational Mobilizes immunization supplements) community coverage, children failing to grow) Health center nurse or NGO Health sector support Municipal or local government Source: Manoff Group (2004). ate low birthweight) by ensuring adequate monthly growth (weight gain) until the age of 24 months and after that to see those children (until they are five years old) only when they are sick, to be sure that they receive timely and appropriate attention. All children under two in the community are eligible, and their parents are encouraged to participate in the monthly growth promotion session, which includes weighing and coun- seling on their growth performance.2 Adequacy of growth is determined by comparing actual weight with the expected weight based on the previous month's weight, using regional norms for weight gain.3 The adequacy of the child's growth is a trigger for a dialogue with the caregiver. From the dialogue, the monitora gains insights into the child's health, nutri- tion, and care during the past month that might affect his or her growth. The monitora then tailors her advice to the mother of the specific child with the aid of a set of counseling cards, differentiated by child's age, adequacy or inadequacy of weight gain, illness status, and breastfeeding status. Each card has pictorial cues for the mother. Using dialogue and the pictures, the monitora determines precisely what the mother is doing (or not doing) now and negotiates with the mother to determine what one practice she might be willing to alter in the 2. See Annex 6.1 for a schematic of monthly weighing. 3. The minimum expected weight gain table was made using two studies by Martell and others (1981) at Uruguay's Latin American Perinatology Center (Centro Latino-Americano de Perinatologia [CLAP]), which looked at the growth of 112 children of 0­23 months of age. Health System Innovations in Central America 185 next month to improve the situation. There may be several areas for improvement suggested by the counseling card, but the monitora selects only the one (or two) that the mother is will- ing to follow in the ensuing month. The improved practice could be as simple as nursing from both breasts at each feeding or giving half of a tortilla to the child at two meals during each day. The next month, the mother gets feedback (in the form of the child's weight gain) that shows whether the behavior change made a difference. If the mother was not able to com- ply with the agreed changes, then the monitora tries to find out why and what can be done to further assist the mother. When a child is seriously ill or has persistent or acute growth failure, the monitora refers the caregiver to the health clinic (on occasion accompanying the mother) and, when the nurse has seen the child, a counter-referral slip is sent back to the monitora to tell her what follow-up is needed at the community level. The nurse auxiliary from the health post nearby often attends the monthly growth pro- motion session, where she updates immunizations; distributes micronutrient supple- ments; provides medicines to children with pneumonia, fever, or worms; arranges for clinic appointments; and discusses family planning with new mothers. Other important AIN-C activities flow from the monthly growth session. The first of these is the home visit, prioritized for those children with problems or who do not attend the weighing session. The home visit is a time for more dialogue to reinforce messages and more direct observation. Home visits are usually done for between one-third and one-half of children under two each month. These visits provide more frequent contact with the family, reinforce messages, prevent difficult situations from getting worse, and help build confidence in the mother and the family that they can properly care for their child. The second critical follow-up activity is reporting of the child growth information to the community (see Box 6.2). Community meetings are held every quarter to permit the Box 6.2. A Monitora Seeks Support from a Men's Group It was the end of the month, and Gladys was completing her report for Laura, the health center nurse, about the status of her community's young children in the AIN-C program. It was easy to see on the bar graph that almost all the children had been weighed that month, but she had to seek out more children at home for weighing than usual. She was annoyed by this, but what alarmed her was that more children were failing to gain weight adequately this month. The previous month had also had a number of children with poor growth, and this month she had 6 children out of 18 children under two years of age who had failed to gain adequate weight for two consecutive months. Gladys knew each case and felt there was more each family could do if motivated. She colored in the bar chart with the nurse, and they discussed it with the mothers of the young children during the nurse's visit. Gladys felt that the point of the importance of each child being seen and the alarm that should be felt when a child fails to grow for two months had not been understood by the community. About a week after the meeting with the mothers, a group of men was repairing a small footbridge over a storm drainage ditch not far from her house. As Gladys watched them work, she realized that many of the men were the fathers of AIN-C children. During a rest period, she showed her graph of the children's growth to the men and explained it to them. The men took a lot of interest in what she said--they wanted to know about their children and what they could do. Many topics were covered, including the use of family money to purchase liquor and venting cooking fire smoke from the house. The men obviously discussed what they learned with their wives, because Gladys was able to report nearly perfect par- ticipation and improvement in the number of children gaining adequate weight the next month. The mothers also told her that they appreciated her calling the men's attention to childcare. Source: Laura Molina, Region II AIN-C supervisor, personal communication, 2003. 186 World Bank Working Paper community at-large to make decisions and work collectively for the betterment of the com- munity's children. Collective community action is key because many problems causing poor child growth go beyond the power or authority of a family to correct. Contaminated water sources, garbage disposal, childcare, and poor health center outreach are all situa- tions that families working together can help resolve. Another important action taken by the AIN-C program is the detection, assessment, and treatment of common childhood illnesses, especially diarrhea and pneumonia, in chil- dren under five. These cases come to the monitora's attention either during counseling or by spontaneous demand. The monitora either treats the disease or, for severe cases, she refers the mother immediately to the health center. Once the monitoras have mastered the core AIN-C program for under twos, focused primarily on home-based preventive actions, they are trained in the illness and newborn modules, which focus more on identifying dan- ger signs, expedited referral, and some community-initiated treatment. Then, to diagnose pneumonia, the monitoras are given timers (to help them make the diagnosis by counting the number of breaths per minute) and an antibiotic to treat it. Just as the monthly session is the core activity, the determination of adequate weight gain is the pivotal indicator around which the information system is structured. The mon- itora maintains a register of all under twos that is constantly updated with births or new arrivals in the community. Each month, each child's weight is recorded with a note on whether or not it was an adequate increase from the previous month. The adequacy of the weight gain is also recorded on the child's health record for the mother. At the end of the month, the monitoras compile the individual growth information from all of the children and display it in bar graphs that use five simple indicators: number of children under two in the community, number weighed that month, number gaining adequate weight, num- ber with inadequate weight gain, and number gaining inadequate weight for two or more months (see Figure 6.5). These five indicators are used for targeting home visits (persistent growth failure and illness get the most attention), to target and focus supervision, for com- munity mobilization, and for health system reporting. These same five indicators are being imported into the management information system and used from the top to the bottom of the health system. AIN-C's system is visual and requires no advanced math skills (such as long division or ability to calculate percentages). The bars make the achievement of the goal clear: monitoras want the first three bars to be of equal height, or if not, the fourth bar should be as low as possible, and the fifth bar should be at zero or the column should be empty. Communication related to improved health and nutrition practices at the community and household levels goes hand in hand with the determination of adequate growth. After all, measuring growth will not have a positive impact on health outcomes if nothing is done when growth faltering is detected. Again, AIN-C has tried to make a relatively complex task (providing tailored advice) as straightforward as possible. The counseling card set described previously in this chapter is the source of the program's health and nutrition technical con- tent. These cards for counseling enable the monitora to tailor general advice to specific sit- uations. In addition, using the audience research done for the development of the cards, other communication materials have been produced for other purposes and for the media. Small flipcharts are available for use by nurses in the health centers, a radio program pop- ularizing key behaviors is broadcast to AIN-C and non-AIN-C communities, and these same generic concepts are disseminated in churches, women's groups, and literacy classes under the national program, Programa de Comunicación en Salud Infantil (COMSAIN). Health System Innovations in Central America 187 Figure 6.5. Monthly Program Indicators 18 17 16 15 14 13 12 11 10 9 8 7 6 5 4 3 2 1 1 2 3 4 5 Indicators: 1. Number of children younger than 2 years listed in the register 2. Number of children younger than 2 years who attended the weighing session this month 3. Number of children younger than 2 years with adequate growth this month 4. Number of children younger than 2 years with inadequate growth this month 5. Number of children younger than 2 years with inadequate growth this month and last month AIN-C devoted considerable care and time to developing its training model and mate- rials using state-of-the-art principles of adult education. Health area staff and nurses from the sector level are trained first to make sure that they understand the program and sup- port it. The area nurse is a critical actor in AIN-C and is responsible for overall supervision of the program in her area. The five-day training program for health area personnel is simultaneously an education for the health area staff and a training of trainers. Even when those trainers go on to train health facility staff and community volunteers, they are ini- tially accompanied by the master trainers to make sure that the training is done correctly. Community volunteers and the nurse auxiliaries who will supervise them are trained together, once again for five days, by the health area and sector staff. In addition, there are monthly meetings at the health center that provide in-service training to volunteers. When new modules are added to AIN-C (for example, the illness module), the same cascade-type training is done, but for two days at the health area level and three days at the health cen- 188 World Bank Working Paper Box 6.3. Municipalization and AIN-C As Honduras's government systems decentralize, municipalities are taking on more responsibility and authority in health policy and health services provision. Municipalities must now make decisions about how to respond to community health needs. The information generated from the AIN-C pro- gram on child growth is extremely helpful. They can see from comparisons of the bar graphs across the months (see Figure 6.6) when situations such as a poor harvest, disruption of work, or pneumo- nia season cause children to fail to grow. The municipal response in terms of mobilizing health care or instituting new policies can be more timely. An example of this comes from a municipality in the western mountainous region of Honduras. Two young children died over a two-month period, each of pneumonia. When municipal health officials looked into the deaths of these children, they realized that although the children were from communities with the AIN-C program, the children had not been taken to the monitora for treatment of their illness. Instead, parents waited and then transported their children several hours to the health service, where it was too late to help the very young children. With these two examples, municipal and health center authorities are meeting with AIN-C com- munities and monitoras to discuss the mothers' seeking care from the monitora at the first sign of a problem and the subsequent referral to the health center. To reinforce the first referral to the monitora, this municipality is testing a new policy: health center personnel will see sick children only if they have been seen first by their community's monitora. To avoid problems, a lot of com- munity education about this new policy is taking place. A large sign is posted at the health center, and parents of sick children are gently reminded of this policy at the health center. Source: Victoria de Alvarado, Basics II country representative, personal communication, 2003. ter and community levels. Training constitutes 54 percent of the first year's costs (Fiedler 2003, table 10). This falls to 41 percent in the second year. Supervision by the health system personnel is designed to be supportive, regular, and methodical. Although the weakest part of the program, supervision is stronger than usual in health systems but is an ever-increasing challenge as the program expands to cover all com- munities and adds new modules, such as the illness management module. The health system in Honduras is divided into health areas at the highest level, health sectors below them, and health centers (CESARs and CESAMOs) below health areas. In each health area, there are about 300 communities across five health sectors, each of which includes about six health centers covering and average of 10 communities each. (Each year, a health center's staff intro- duces AIN-C in two communities. On average, about six years are needed for complete cov- erage of all communities in a health center's catchment area.) Initially, when the program is introduced to a community or a health facility, the supervision is more intense than it is later, when volunteers and auxiliaries have mastered the basic program skills. The supervision load is 52.5 person-days per year at the health area level, 78.0 days per year at the health sector level, and 30.0 days per year at the health center level (Fiedler 2003). A full 39 percent of costs are for supervision in the first year for the first cohort of communities (usually two) in a health area and 33 percent for subsequent cohorts (Fiedler 2003, table 10). In subsequent years, supervision costs rise proportionally (because training costs diminish) and become vir- tually the entire program cost by the sixth year (Fiedler 2003, Table 12). This relatively high proportional cost for supervision is due to the very low cost of service delivery. The high cost makes it clear that Honduras understands that community-based programs need frequent and skilled supervision. Health System Innovations in Central America 189 Figure 6.6. Five-Bar Graph Presented to the Community to Stimulate Discussion of Changes in Child Growth over Time April May June July 25 2 25 25 5 24 2 24 24 4 23 2 23 23 3 22 2 22 22 2 21 2 21 21 1 20 2 20 20 0 19 1 19 19 9 18 1 18 18 8 17 1 17 17 7 16 1 16 16 6 15 1 15 15 5 14 1 14 14 4 13 1 13 13 3 12 1 12 12 2 11 1 11 11 1 10 1 10 10 0 9 9 9 9 8 8 8 8 7 7 7 7 6 6 6 6 5 5 5 5 4 4 4 4 3 3 3 3 2 2 2 2 1 1 1 1 1 2 3 4 5 1 2 3 4 5 1 2 3 4 5 1 2 3 4 5 Indicator Indicator Indicator Indicator Indicators: 1. Number of children younger than 2 years listed in the register 2. Number of children younger than 2 years who attended the weighing session this month 3. Number of children younger than 2 years with adequate growth this month 4. Number of children younger than 2 years with inadequate growth this month 5. Number of children younger than 2 years with inadequate growth this month and last month For the first six months of program implementation, either the health center nurse or the health sector nurse attends each community weighing session. All volunteers working in the catchment area for each center meet at the health center monthly, for the life of the program, to discuss problems and get new supplies. This is a form of supervision and train- ing at the same time. What are Other Design Features? The design of the program does not tell the full story about AIN-C and why it is an unusual program. There are several particular characteristics of the program that differentiate it from programs that might appear to be similar, and this makes it an obvious choice for extending the principles of health reform to the community. Evidence-based decisionmaking. Perhaps the most important attribute of AIN-C is that it provides a mechanism for identifying community needs directly linked to improving health status. The objective indicator, adequacy of child growth, offers the community a 190 World Bank Working Paper structured means to analyze its problems (growth of young children is a good proxy for many social, economic, and biological problems in the community), measure progress, and present their situation, ideas, or improvement needs to outside entities (for example, municipalities, NGOs, donors). Data on child growth status are sensitive over fairly short time periods to changes in the community and are sensitive to a wide range of stresses even beyond health problems. Thus the use of child growth as an organizing principle for the community opens the door to discussing and tackling a number of social problems, including lack of adequate child care, HIV/AIDS, poor water and sanitation, alcoholism, domestic abuse, illiteracy and edu- cational failure, agricultural productivity, intrahousehold allocation of resources, gender relations, and more. AIN-C is structured in such a way as to uncover local needs and make them, and local ideas for their resolution, accessible to all. Equity. AIN-C is targeted to all children under two in the community, and their identi- fication and enrollment is an important indicator of program performance. Including all children sends a clear message of concern that everyone's children should be growing ade- quately, not just the poor children, the malnourished, or those at highest risk. There is no stigma attached to being a participant. At the same time, so far, AIN-C has been geographi- cally targeted to the worst-off areas and the worst-off communities within those areas. The communities selected for first entry into the program are those with the poorest access to health care or in the worst socioeconomic status quintiles in the health center's catchment area. This targeting helps a health system bring greater equity to health services provision. Empowerment. AIN-C starts with the assumption that the family has the responsibil- ity and the right to keep its children healthy and well fed. The program provides guidance on small changes in behavior that require minor changes in allocation of resources within the household (perhaps 300 calories all together or a half-hour of time). AIN-C's design- ers felt that this was the best place to start: give the family the opportunity and confidence to nurture children within its own resource base. The research done on feeding of young children in Honduras indicated that only a small minority of families (8 percent) could not do more to improve their children's food intake (Maradiaga, Griffiths, and Nunez 1997). In cases where a household tries and still comes up short, the community is asked to take on some of the responsibility. Finally, if the community's resources are inadequate, external sources (such as food aid programs, cash transfer programs, and welfare programs) can be sought. This attitude toward family self-sufficiency is different from the common practice that assumes from the beginning that the family cannot adequately provide for its children and immediately offers food, coupons, or cash to families with malnourished children. Some- times these donations reward failure--the classic case is that of a food supplementation program that gives food only to children who are already malnourished. Some programs provide cash-equivalent coupons to clients with the condition that they use a defined pack- age of health services. These transfers devalue both the contingent services and the family. If the reason for nonuse of services was poor quality (long waiting times, inconvenient hours, lack of medical staff or drugs) or inaccessible services, then the transfers may exac- erbate the problem by creating overcrowded clinics and demand for unwanted or sub- standard services by an uninformed consumer. Where the transfer is a substitute for effective communications with families about behavior change, once the cash or in-kind Health System Innovations in Central America 191 incentive is removed, the demand for services disappears. For the family, receiving cash or in-kind incentives becomes the goal rather than improving health and the family members do not become informed consumers. Ownership. It is often said that sustained program success requires local ownership, and this has been shown in Honduras (Bossert and others 1988). AIN-C is a locally owned program because it was conceptualized and developed within the Secretariat of Health by Hondurans. From the beginning, the program had local champions who repeatedly artic- ulated the program's key concepts and reason for being. Although external assistance has played a role, it was a group of Honduran health professionals who guided the program's course. An important step in solidifying the program as part of the Honduran health land- scape was its adoption as national health policy in 2000. Not only are the policy makers the "owners," but the design and running of the pro- gram has been done in such a way as to make every community the owner of its activities and of progress made in its children's growth. AIN-C guidance does not give precise speci- fications for how community workers are to perform their jobs. Instead, the program estab- lishes goals or outcomes that must be reached. How a community or group of workers chooses to reach the goals is up to them. For example, all children under two years should be seen by a monitora each month. Whether the monitoras accomplish this by, for example, house-to-house visits, neighborhood meetings, or community-wide meetings is up to them. In AIN-C, communities decide if they want to have the program in their community, how many and who will be a monitoras, how they will reach every child under two every month, how they will create a community environment that favors adequate child growth, and how they will interact with the government's health infrastructure. Focus on tasks. The job description of the monitora is the basis for determining all other program actions because these actions function to support the community effort. Each item in the job description and community plan relates to the core objective of the program: promoting healthy growth of children. The job description is captured in prac- tical detail in the monitora's manual. In a similar fashion, the trainers' and supervisors' jobs are built upon the foundation of the monitoras' work. Their manuals and tools build on the monitoras' manuals and tools. The monitoras are trained and supervised by the local nurse from the health center using the monitoras' manual as the guide. The area nurse trains and supervises the center nurses, and her work is, once again, a further refinement of the work of those she supervises. Hence there is a seamless connection between community-based work and the formal health system. Teamwork with specialization. A key practice implemented by AIN-C is the use of a team of volunteers at the community level rather than relying on just one person. Having a team means that each member contributes different strengths. One may be good at weighing and charting while another is good at counseling. In addition, a team minimizes the effect of turnover and enables all members to help each other learn and remember lessons from the training. Working as a team means that the work is not too burdensome for any individual. Thus Honduras is able to rely on volunteers who spend about four hours per week on AIN-C tasks, in addition to periodic tasks for the Secretariat of Health (mobilizing the community for immunization days and various campaigns). 192 World Bank Working Paper Box 6.4. Periurban Communities Request Assistance on Saturdays The first indicator of AIN-C implementation is coverage. The expectation is that at least 90 percent of the children under two will be seen each month by the monitora and that no child will miss more than two sessions a year. When AIN-C monitoring data were shared with periurban com- munities, the low participation figures were an immediate cause of concern. Salaried working mothers found it difficult to bring their children to sessions during the week. They said they could attend sessions only on Saturday after working hours. The monitoras indicated that they could change the sessions to Saturday, but health center staff had been reluctant to work on Saturdays. However, when community leaders met with the health center, an arrangement was worked out for the auxiliary nurse to be available on Saturdays to attend AIN-C growth promotion sessions. Participation is now about 80 percent each month in periurban communities. Source: Tesla Gutierrez, AIN-C Region III supervisor, personal communication, 2000. Teamwork extends to the Secretariat of Health's view of how AIN-C is implemented. From the beginning, the secretariat has welcomed and encouraged NGO collaboration in the implementation of AIN-C. In fact, in 2000, when AIN-C was decreed the national child health program, the secretariat notified NGOs working in the field of child health that the basic package they must implement is AIN-C--they can add to AIN-C but not ignore it. This insistence and support from the secretariat has meant that to date at least 800 more communities have AIN because of NGO collaboration. The only drawback to NGOs' expanding program coverage is that when an NGO pulls out of an area, the secretariat must provide supervision to many more communities than they have programmed resources for. One NGO, realizing this dilemma, limited the num- ber of communities they covered and put their additional resources into supportive super- vision, additional training, and monitoring for all AIN-C communities in the area. The idea was that when they left the area, they would leave behind a strengthened permanent system that could be maintained by secretariat resources. The NGO's activities included two regional trainer supervisors, the purchase of a computer and help from a data analyst to make a computer program to compile the five AIN-C indicators for the entire region. Systematic support for volunteers. There are debates among professionals in the devel- opment field about reliance on volunteer workers. On the positive side, obviously volun- teer workers cost less than paid workers, and they often have greater motivation (beyond just a paycheck) to do well. On the negative side, volunteers are said to have limited tenures, volunteerism is thought to be exploitative, and often volunteers tend to be from the higher- status or higher-income groups in a community. Neither of these sides is always right. Honduras has a history of community spirit and volunteerism, but this alone is not enough to sustain a program over decades. AIN-C has designed its program with the vol- unteer in mind. In particular, the job of the monitora is manageable for a volunteer. AIN-C monitoras work, on average, 14.9 hours per month, of which 4.5 hours are at the monthly growth promotion session, which is a reasonable workload for a volunteer (Fiedler 2003). AIN-C also provides systematic incentives to its volunteers, and these incentives have both intrinsic value and market value. What Honduras has learned is that the incentives need to be regularly provided and planned for--just like all other operational aspects of the pro- gram. Examples of the incentives are a letter from the Secretary of Health thanking the fam- Health System Innovations in Central America 193 ily of the monitora for their generosity, an identification card with a photo of the monitora, and regular parties in honor of Mother's Day or Children's Day holidays. Training and monthly meetings at the health center are also used as incentives for volunteers. Volunteers in Honduras have done community health work for decades. Many of the monitoras are the people who volunteered years ago for sanitation or diarrhea control work, or they are traditional midwives. The number of monitoras has remained the same over time, but the individuals doing the job have changed. Because there is a team of monitoras, new people learn the job from their predecessors, and the health center nurse refreshes forgotten concepts (Griffiths 2004). The fact that over a five-year period, six to eight people might have worked in their community's AIN-C program, instead of just three, strengthens community commitment, knowledge, and ownership of the process and program. Monitoring and program development. As noted above, the information system is simple, practical, and useful for a number of different purposes: community mobilization, targeting supervision, health system reporting, and decisionmaking by the monitoras. If participation rates are low, then the community leaders may need to mobilize support among families in the community. If participation is high but too many children have chronic growth faltering, then the supervisor may need to make some visits to homes and observe the monitoras' coun- seling. If there is a widespread problem, then the community may need to take action to secure external support. In addition to the five basic indicators, AIN-C also monitors immunization rates, disease and referral rates, and distribution of micronutrient supplements. These help the auxiliary nurse to better target her own work. The community baseline and regular updates and the quarterly meetings help the community monitor its own progress in improving the well-being of its most vulnerable members and, by extension, its own development. Conclusions Health reform must extend to all communities and engage their participation to truly achieve national health goals and the MDGs. Few models exist for garnering the active participation of the community in improving the impact, efficiency, and quality of health care services. The AIN-C program offers one such model. It has benefited from careful refinement over a decade and the incorporation of best practices from reviews of other community- based programs. Growth promotion entails tailored attention to each child and frequent interaction between the community agent and the family. The World Bank (2003) says such interactions are particularly hard to monitor, yet AIN-C's five-indicator information system does just that. Built on the concept that a healthy child is a growing child, AIN-C uses the adequacy of monthly growth of children under two as the indicator of child, fam- ily, and community health. In addition, the evidence about growth is used as a tool to cat- alyze community analysis and commitment to solutions to the problems affecting all young children in the community. The program promotes equity by including all children under two and by targeting preferentially to the lowest socioeconomic groups in commu- nities. It is intimately tied in to facility-based services through the critical role of health per- sonnel in training and supervision and through referral and counter referral. Its cost is quite low, and its impact is broad and deep. 194 World Bank Working Paper Annex 6.1. Activity Flow for the Introduction and Operation of AIN-C in a District Train district health team and facilitators. Train health subdistrict and health center teams. Agreement with community to run the program and select the group of volunteers. Carry out community baseline. Train volunteers from communities under one or more health centers. Monthly growth promotion session 1. Weighs child. 2. Compares weight 3. Caregiver 4.Nurse immunizes to expected weight; counseled on children, advises informs the mother of adequate or mothers on family weight and adequacy inadequate growth planning, and sees of gain. sick children and and illness status. those with acute weight loss. Volunteers Home visit to sick complete monthly children or those report on weight with inadequate gain. weight gain. Five bars for HC. Health center nurse and volunteers hold quarterly meetings with the Meet monthly at community to discuss health center to growth of all under twos review report, check and develop projects. supplies, and get in-service education. Health System Innovations in Central America 195 Bibliography Asociación Hondureña de Planificación de Familia. 2001. "Honduras Encuesta Nacional de Epidemiología y Salud Familiar (ENESF-2001), Encuesta Nacional de Salud Masculina (ENSM-2001), Informe Resumido." Secretaría de Salud, Asociación Hondureña de Plantificación de Familia, y USAID, Tegucigalpa. Basic Support for Institutionalizing Child Survival Project (BASICS). 2001. "Reducing Severe and Moderate Malnutrition in Children." In F. Marc LaForce, ed., BASICS II. A Better Future for Children: Progress toward World Summit Goals for Health and Nutri- tion. Series of briefing papers prepared for the U.N. Special Session on Children. Arlington, Va.: BASICS II. Bossert, T., L. Godiksen, T. D. Bunce, M. Favin, C. Dabbs, and J. Massey. 1988. "The Sus- tainability of US-Supported Health, Population, and Nutrition Programs in Honduras: 1942­1986." USAID Evaluation Special Study 58. U.S. Agency for International Devel- opment, Washington, D.C. Dickin, K., M. Griffiths, and E. Piwoz. 1997. "Designing by Dialogue: Consultative Research to Improve Young Child Feeding: Support for Analysis and Research in Africa." The Manoff Group and Academy for Educational Development for the Health and Human Resources Analysis (HHRAA) Project, Washington, D.C. Fiedler, J. 2003. "A Cost Analysis of the Honduras Community-Based, Integrated Child Care Program (Atención Integral a la Niñez­Comunitaria, AIN-C)." Health, Nutrition, and Population (HNP) Discussion Paper. World Bank, Washington, D.C. Fiedler, J., and J. Suazo. 2002. "Ministry of Health. User Fees, Equity and Decentralization: Lessons from Honduras." Health Policy and Planning 17(4):362­77. Gillespie, S., J. Mason, and R. Martorell. 1996. "How Nutrition Improves." ACC/SCN State-of-the-Art Series Nutrition Policy Discussion Paper 15. U.N. Administrative Coordinating Committee Subcommittee on Nutrition, Geneva. Griffiths, M. 2004. "¿Implementación: el vinculo entre teoria y los resultados?" PowerPoint presentation at the Taller Nacional de Expansión de AIN, Secretariat of Health/BASICS II, Tegucigalpa, June 29­30. Griffiths, M., and V. de Alvarado. 1999. "Honduras. The AIN Community Experience." Paper presented at the BASICS Results Meeting, USAID, Washington, D.C., June. Griffiths, M., K. Dickin, and M. Favin. 1996. "Promoting the Growth of Children: What Works." World Bank Nutrition Toolkit, Tool 4. World Bank, Washington, D.C. Haddad, L., H. Alderman, S. Appleton, L. Song, and Y. Yones. 2002. "Reducing Child Under- nutrition: How Far Does Income Growth Take Us?" FCND Discussion Paper 137. Inter- national Food Policy Research Institute, Washington, D.C. Ho., T. J. 1985. "Economic Issues in Assessing Nutrition Projects: Costs, Affordability, and Cost-Effectiveness." Population, Health, and Nutrition Technical Note 85-14. World Bank, Washington, D.C. Kennedy, E., and H. Alderman. 1985. Comparative Analyses of Nutritional Effectiveness of Selective Interventions. Washington, D.C.: International Food Policy Research Institute. Manoff Group. 2004. "Results and Operational Framework for Community-Based Growth Promotion." Washington, D.C. Maradiaga, A., M. Griffiths, and I. Y. Nunez. 1997. Prácticas mejoradas de alimentación. Tegucigalpa: Secretariat of Health and BASICS. 196 World Bank Working Paper Martell, M., L. A. Bertolini, F. Nieto, S. M. Tenzer, R. Ruggia, and R. Belitzky. 1981. Crec- imiento y desarrollo en los dos primeros años de vida postnatal. Washington, D.C.: Orga- nización Panamericana de Salud. Martorell, R. 1990. "Importance of Childhood Retardation for Adult Body Size." Statement prepared for the Expert Meeting on Economic Consequences of Health Programs in LDCs, Committee on Population, National Academy of Sciences, Washington, D.C., June 20. Martorell, R., L. K. Khan, and D. G. Schroeder. 1994. "Reversibility of Stunting: Epidemi- ological Findings in Children from Developing Countries." European Journal of Clin- ical Nutrition 48(Suppl)(1):545­57. Mason, J., J. Hunt, D. Parker, and U. Jonsson. 2001. "Improving Child Nutrition in Asia." ADB Nutrition and Development Series 3. Manila: Asian Development Bank. Pelletier, D. L. 1994. "The Relationship between Child Anthropometry and Mortality in Developing Countries: Implications for Policy, Programs and Future Research." Jour- nal of Nutrition 124(10S):2047S­81S. Plowman, Beth, Janet Irene Picado, Marcia Griffiths, Karen Van Roekel, and Victoria Vivas de Alvarado. 2004. "BASICS II. Evaluation of the AIN Program in Honduras, 2002. Draft Survey Report." BASICS II for USAID, Arlington, VA. Secretaria del Estado en el Despacho de Salud. N.D. "Plan de gobierno para el sector de salud 2002­2006." Tegucigalpa. ------. 2003. "El proceso de reforma del sector salud en Honduras." Tegucigalpa. United Nations Children's Fund (UNICEF). 1992. Food, Health and Care: The UNICEF Vision and Strategy for a World Free from Hunger and Malnutrition. New York. Van Roekel, K., B. Plowman, M. Griffiths, V. de Alvarado, J. Matute, and M. Calderon. 2002. "BASICS II. Midterm Evaluation of the AIN Program in Honduras, 2002." BASICS II, Arlington, VA. Van Roekel, K., B. Plowman, and M. Griffiths. 2004. "Evaluación final del programa de AIN en Honduras. BASICS II." Paper presented at the Taller Nacional de Expansión de AIN, Secretariat of Health/BASICS II, Tegucigalpa, June 29­30. Villalobos, C., J. McGuire, and M. Rosenmöller. 2002. "Honduras. AIN-C Integrated Child Care: Improving Health and Nutrition at the Community Level." In M. Rosenmöller and the World Bank, eds., Challenges of Health Reform. Barcelona: Estudios y Ediciones IESE, S.L. Walker, I. 2003. "A Health Reform Strategy for Honduras." World Bank, Washington, D.C. Processed. World Bank. 1997. "Honduras--Improving Access, Efficiency, and Quality of Care in the Health Sector." Report 17008-HO. World Bank, Washington, D.C. ------. 2001a. "Honduras Poverty Diagnostic 2000." Report 20531-HO. World Bank, Washington, D.C. ------. 2001b. "Honduras--Public Expenditure Management for Poverty Reduction and Fiscal Sustainability." Report 22070. World Bank, Washington, D.C. ------. 2002. World Development Indicators. Washington, D.C. ------. 2003. World Development Report 2004: Making Services Work for Poor People. Washington, D.C. CHAPTER 7 Innovations in Improving Efficiency and Impact in Health Care Services--Lessons from Central America Maureen Lewis Overview Ensuring the sound performance and desired impact of health care delivery systems poses a challenge to all governments. In much of the developing world, including most of Latin America, the integrated public financing and service provision model prevails with line- item budgets, public employees, public infrastructure, and bureaucratic strictures. The health reforms of recent years in Latin America have attempted to address a number of the incentive problems underlying public health care delivery in both the financing and deliv- ery of health care with mixed results. The sweeping reforms in Brazil, Chile, and Colom- bia, for example, entailed ambitious, broad-gauged changes. However, their breadth and comprehensiveness, combined with a neglect of baseline data collection, made impacts hard to trace and lessons learned inconclusive. Much of the challenge in health reform involves shifting incentives to improve produc- tivity, quality, and performance. But how to accomplish those goals remains elusive in all health care systems, and those in Latin America are no exception. The experiences here demonstrate that modest reforms can have impressive effects on performance and consumer satisfaction, providing building blocks for further reforms. The foregoing chapters addressed major issues within the context of Central America, where governments have launched experiments to contract out health service delivery to nongovernmental organizations (NGOs) and devised incentives to boost productivity, strengthen the scope of public health services, often in underserved areas, and improve user satisfaction. These experiments have taken on the sensitive issue of private sector delivery of health care. Although this is a reality throughout Latin America, the interaction and com- plementarities between the public and private sectors in health have often been contentious 197 198 World Bank Working Paper and unexploited. Harnessing private entities to deliver health care to low-income popula- tions therefore has been controversial. Failures and rigidities inherent in the public health systems led governments in Central America to experiment with radical shifts in the struc- ture and functioning of health and nutrition services and reach out to NGOs and other private groups in the process. This has meant not only hiring private contractors, but implicitly having them com- pete with the public system and, in some cases, allowing private contracted personnel to provide such services as management or fiduciary functions to the ministry of health or social security system. This has also meant that for the first time, public providers face direct competition, and a number of the experiments included here specifically collected information that allows comparisons in productivity, costs, and performance. Typically lacking in benchmarks, public performance has historically been assessed only in terms of trends or claims of providers and consumers. Costs and impacts remain virtually unknown because of lack of data. The country experiences discussed in this volume raise thorny issues for governments saddled with inefficient public delivery systems, but they also offer alternatives to the tra- ditional health care delivery model--and with sufficient variety to suggest that various options can and do work. Improvements can be achieved even in difficult and far from per- fect environments. These conclusions alone offer encouragement to health system reform and to changes that can improve performance on a smaller scale. Finally, the experiments demonstrate the importance of incentives in raising perfor- mance and productivity, and the benefits to policy makers and program managers of bet- ter and more complete data. Without such innovations, health care delivery would remain a "black box" absorbing resources and producing a largely undefined set of outputs. The data and the experiments here offer new information and new ways of doing business. Alternative Approaches In Health Care Delivery Accountability in public services remains uneven in much of Central America. Indeed, recent efforts to decentralize were a hoped-for panacea for the overly centralized and rigid bureau- cracies that encountered many difficulties in ensuring that health care delivery worked and services reached the intended populations. Public failures have helped to limit the benefits of decentralization and undermine meaningful reforms in health care (Kaufman and Nelson 2004). The efforts analyzed here represent a way to address accountability through partial reforms based on experimentation, and do so with sufficient information that performance can be assessed. As a result, the experiments reflect significant departures from traditional govern- ment provision of health care. They focus on developing incentives for improved perfor- mance and efficiency in health care delivery, taking the existing public systems as the starting point. Contracting out is the most obvious initiative, but others--such as pay- ment arrangements, collecting and using costing information, and incorporating user perspectives and satisfaction--are equally innovative. The institutional and political issues have also been addressed in some of the country cases, notably in Nicaragua and, espe- cially, Guatemala, where the institutional process is laid out as a companion to the analy- sis of the experiments. Health System Innovations in Central America 199 Contracting out encompasses many different kinds of public purchasing (Loevinsohn and Harding 2004). These range from hiring contracted services for specific, nonclinical tasks, such as catering, laundry, and cleaning services (see Mills 1998), to the wholesale hir- ing of a company that takes responsibility for all aspects of health care delivery for a defined population and an agreed sum. In both cases, government remains responsible for ensur- ing that services are delivered. Yet, in the first case, their role remains managing the deliv- ery of services; in the latter case, it requires the selection and management of contractors, a very different task requiring varied sets of skills. Both, however, require that public bod- ies remain engaged in ensuring compliance and acceptable performance. Having government ensure the performance of private contractors offers an alternative to simply transferring responsibility to lower levels of government. Some of the efforts detailed in this volume, such as the Honduran growth-monitoring experiment, successfully integrated local communities with national programs without simply divesting to lower lev- els of government. In Guatemala and Nicaragua, the end of long conflict periods provided the opportunity to try and reach underserved populations with a more flexible tool that did not require long-term public investment. In Guatemala, that meant the poor and isolated; in Nicaragua, it meant middle-income earners under the Nicaraguan Social Security Insti- tute (Instituto Nicaragüense de Seguridad Social [INSS]) whose coverage had eroded under the Sandinista government. The notion of increasing flexibility and providing incentives for performance in underserved areas proved popular in both countries and resulted in not only more outreach but also a rise in suppliers willing to partner with the government. NGOs and other forms of private delivery have long been active in health care delivery in developing countries, and they have been particularly prominent in Central America over the past few decades as the countries descended into civil conflict and guerrilla warfare. Recent evidence from Uganda has suggested that NGOs provide higher-quality service at lower cost than government systems, in part as a result of altruistic motives, but the greater flexibility allows efficiency and quality to prevail (Dehn, Reinikka, and Svensson 2003). Guatemala, El Salvador, and Nicaragua relied on private groups to deliver care, particularly in rural areas, during conflict periods, and their integration in the postconflict period was a natural extension, but certainly nothing to count on. Tensions between NGOs and public provision often arise when implicit competition emerges, but when government becomes the payer, tensions subside. The Guatemala and Nicaragua examples provide details on the relationship and interaction. The experiences in Latin America--Brazil with its network of private hospitals that deliver the bulk of inpatient care (World Bank 1994), Chile with its reliance on managed care companies to provide care to the upper-income groups (Sapelli 2004), Colombia's contracting with private providers (Ramirez 2004)--all point to the emerging realization that private delivery has benefits for public programs. The Central American experiences described here simply extend those initiatives to lower-income settings with populations that are often harder to reach. Summary of Experiments and Salient Characteristics The five country experiences in Central America represent a broad set of options and, in some cases, ground-breaking designs. Table 7.1 provides a summary of the characteristics 200 World Bank Working Paper Table 7.1. Summary of Contracting Characteristics Country and experiment Delivery arrangements Financing arrangements Costa Rica: · NGOs rent clinics from the · NGOs' per capita payment Compares traditional CCSS for US$1 per year and a through a management social security provision cooperative provides agreement. (CCSS) and contracted maintenance. Purchase · CCSS historical budgets. NGO providers for inputs from CCSS at cost + · Performance guarantee bond integrated PHC services 15%. Own hiring and guaranteeing that at least using a cross-section management. 85% of performance targets analysis and multivariate · CCSS: Traditional public are met. Public providers risk analysis of time series health care delivery with 2.5% of budget for low per- for type III and IV clinics. rigid hiring and formance. management. Guatemala: · PSS (direct NGO health · Direct PSS: capitation Comparison of three service providers) payments and in-kind contracting models of contracted to reach remote vaccines. the Ministry of Health areas through community · Mixed ADMSS: capitated (MSPAS) to deliver outreach centers and payment for services. services to the rural monthly community visits. · MSPAS: budgetary allocation poor--by (1) directly via Hired all staff and managers. and in-kind vaccines. PPS; (2) indirectly through · ADMSS mixed system: administrative support to private financial the MSPAS (ASMSS); and, administrators for the (3) traditional MSPAS MSPAS. Hired most staff, services. and the MSPAS supervises. · MSPAS traditional model with volunteer physicians and midwives. Nicaragua: · MSP: Integrated care with · Monthly capitated payment Compares INSS defined benefits and specific per enrolled worker, with Provisional Health Model exclusions. Preapproval some coverage for spouse (MSP) that contracts out needed for costly tests and and children. managed care to EMPs procedures; subcontracts · EMP selected. in remote areas and the with MINSA for upgraded · INSS reinsurance financed traditional MINSA hospital care. through 5% of capitated approach. · MINSA: Traditional PHC payment. delivery with rigid hiring and · Supervisory INSS physicians management. monitor. Honduras: · AIN-C promotes community · Government nurses provide Compares communities growth-monitoring growth monitoring at participating in AIN-C programs. facilities. growth promotion · Delivery is through · Volunteers are trained and activities with control Secretariat of Health supervised by Secretariat of communities. facilities or community Health nurses. volunteers (monitoras). (continued ) Health System Innovations in Central America 201 Table 7.1. Summary of Contracting Characteristics (Continued ) Country and experiment Delivery arrangements Financing arrangements Panama: · San Miguelito Hospital: · San Miguelito Hospital: Compares the San CONSALUD purchases prospective payment system Miguelito Hospital management, medical, and with annual production experiment of an other services. Controls own ceilings on inpatient and intermediary purchaser, catchment area and has outpatient department care. CONSALUD, with two autonomy in hiring, Overproduction reimbursed traditional other management, and at a lower rate. hospitals. contracting. · Payment of San Miguelito · Other hospitals: Traditional Hospital contractors based public heath care delivery on external medical and with rigid hiring and other service audits of management practices. contracted providers and monthly user surveys. · Other hospitals: line-item budget. ADMSS health services administrator (administradora de servicios de salud); AIN-C Atención Integral a la Niñez en la Comunidad (Integrated Community Child Health Program); CCSS Caja Costarricense de Seguro Social; EMPs empresas médicas previsionales (previsional medical firms); INSS Instituto Nicaragüense de Seguridad Social (Nicaraguan Social Security Institute); MINSA Ministerio de Salud (Ministry of Health); MSP Modelo de Salud Previsional (Previsional Health Model); MSPAS Ministerio de Salud Pública y Asisten- cia Social; PHC primary health care; PSS proveedora de servicios de salud (health services provider). of each, focusing on the nature of delivery and financing arrangements. Unlike experi- ments that are national in scope, where almost everything changes, these programs con- tain rich documentation and measurement, permitting the assessment of the experiments and their components. Despite the focus on contracting, the country experiments encompass multiple ele- ments of health reform, including incentives for raising productivity, splitting providers and purchasers of care, hospital autonomy, collaborating with NGOs, using financing levers to ration health care, and working with communities to promote preventive health care. In the process, all the country experiences have incorporated key reform elements. For the most part, they evaluate natural experiments that are then compared with exist- ing or alternative approaches. Panama's San Miguelito Hospital (Hospital Integrado San Miguel de Arcángel [HISMA]) tests a radical approach. CONSALUD, a public purchasing cooperative (pooled purchasing) for the Ministry of Health and the Social Security Insti- tute, buys a comprehensive set of hospital services from HISMA to serve the insured and uninsured residents of the San Miguelito section of Panama City. The HISMA Board then purchases the medical, administrative, hotel, and other services needed to deliver health care to the catchment area population, all with virtually no intrusion by the government. This is unique in Latin America. A prospective payment system with predetermined prices for hos- pital use, physician visits, and surgery replaces line-item budgeting. Costa Rica, departing from an all-public focus that has characterized its successful health system, allows the "rental" of its public facilities by private groups with autonomy in management and provi- sion. Guatemala and Nicaragua have hired private firms and NGOs to serve hard-to-reach populations. The Honduran community growth-monitoring program compares progress 202 World Bank Working Paper in target communities with control communities to demonstrate the power of community volunteers who have received training and public supervision. A number of noteworthy elements emerge from these experiences: the willingness to allow autonomy to a private entity in the delivery of health care when public funds are involved, a major step for public health systems; collection of baseline or comparison data, including on costs and productivity; willingness of governments to compare traditional and new delivery and financing arrangements, risking the possibility that the current approach is flawed; oversight and regulation of contractors, though with significant challenges; concern for consumer satisfaction and involvement. Table 7.2 summarizes the performance measures applied in each country, and the associ- ated findings on cost and efficiency, performance and user satisfaction. Evaluation often confronts awkward political issues. If public programs don't work, will there be personal costs to public servants? The experiments suggest that the five govern- ments' health and social security leaders were willing to take the risk of bad news--and the news, although encouraging for policy, suggests that current public arrangements fall short of potential. Contractor performance was consistently better regardless of the measures or indicators used. Preventive services--notably immunizations, oral rehydration therapy and prenatal care--improved under contractor arrangements in Costa Rica, Guatemala and Nicaragua. Similar aspects were achieved in Honduras with a community based develop- ment model. Mixed public-private arrangements such as those in Guatemala also improved on traditional public health services. Efficiency gains, as reflected in lower average lengths of stay (ALOS), were consistently found in the four countries with contracted hospital services, and higher productivity was recorded in each. Indeed, in Nicaragua, productivity was three times greater in the private facilities when compared to traditional public performance. Costs of both public and private delivery need to be assessed together with perfor- mance and outcomes. In Costa Rica, costs were lower, but supervision also ensured that referrals, drugs, and laboratory tests were used with discretion, resulting in a drop in vol- ume. Guatemala and Panama posted higher management costs at the provider level, where productivity also rose. Higher costs were incurred for productivity enhancement and for reaching marginalized target groups, as was the case in Guatemala. The measures intro- duced to contain the costs and volume of care proved effective: providers remained within budget and demonstrated savings in service delivery. A major risk confronting public contractors is complacency or lack of authority on the part of the government. Contracts require clauses for government recourse in the event of noncompliance, and more than simply cutting off funding is needed. If funds are terminated, services will follow suit, leaving an unserved population. Intermediate steps and sanctions, as well as effective (dis)incentives to promote sound performance, are needed. The experi- ments here have incorporated flexibility and sets of "carrots and sticks" to encourage high performance. Costa Rica issued a guarantee bond to ensure that contractors met targets, and placed a similar if more modest tax on public performance; Nicaragua provided a virtual rein- surance fund in a 5 percent set-aside for high costs and utilization review requirements; and Panama reduced reimbursement levels for services that exceeded defined output targets to discourage overproduction. These tools apply sophisticated incentive concepts meant to Table 7.2. Selective Evaluation Findings Country and methods Performance measures Efficiency and costs Overall performance User satisfaction Costa Rica: · CCSS applied 23 core · Contractors provided more · Mortality the same for both · Waiting time: 68% said there Cross-sectional indicators visits with general practi- groups. was a short wait for care at comparisons · Analysis used 7 key indi- tioners, but fewer labora- · Comparative performance of NGOs; 55% noted a short wait and time series cators: coverage rates for tory services, drugs, and the CCSS and contractors: at the CCSS. multivariate various target groups or specialist referrals, reducing higher contractor perfor- analysis. services (for example, average costs; 30% less mance in 5 of 7 indicators, rate of Pap smears for spending overall. and difference with 2 of 7. women over age 35) · Regressions show consis- · General and infant mor- tently higher performance tality rates (higher quality and lower Volume of service indica- cost) of contractors. tors for clinics Health Guatemala: · Comparison of costs and · PSS: Highest production and · The PSS showed better perfor- · Women prefer ADMSS and the Comparison of 3 productivity of services labor productivity, and mance indicators, immuniza- PSS to HPs; shifted to mixed- System models; cross- by type of provider highest costs (travel, train- tion and prenatal coverage, service providers. sectional data · Incidence of selected ill- ing, bonuses to attract staff and outcomes compared to · MSPAS services rated lowest. Low comparators and nesses and prevention to remote areas). traditional providers. utilization where there was no Innovations logistic · Consumer use of specific · MSPAS: Lowest costs and · ADMSS showed better prena- physician on staff. regression. services and their satis- productivity: tal care coverage, tetanus and · Frequent consumer shifts from faction, by type of Per capita cost Average cost typhoid immunization, and MSPAS to ADMSS or the PSS. provider PSS Q 5.70 Q 40,000 micronutrient distribution. in ADMSS Q 8.50 Q 30,000 · Management of childhood Central HPs Q 7.60 Q 24,000 diarrhea was best in ADMSS · Indirect overhead costs are and more similar in the PSS 20% for central MSPAS ver- and MSPAS. America sus 7% for PPS. Direct over- heads: PPS 3%, MSPAS 0%. (continued ) 203 204 World Bank Working Paper Table 7.2. Selective Evaluation Findings (Continued ) Country and methods Performance measures Efficiency and costs Overall performance User satisfaction · Coverage of immuniza- tion, prenatal care use, and growth monitoring, by type of provider Honduras: · 5 indicators related to · Costs are entirely for super- · Knowledge and participation · Program has continued to grow, Comparison caretakers' exposure or vision because services are increased threefold; among attracting new participants. between participation and use of run by volunteers. First-year the control group. Across 10 participating key health services costs are 39% for supervi- categories, change was negli- communities and · 5 indicators of caretak- sion and 54% for training of gible except in the case of Control group of ers' health-promoting volunteers and auxiliary knowing about CBGP. communities. practices personnel. · Iron supplements for chil- · AIN-C annual recurrent costs dren rose from 2% to 47%; are US$2.73 per child under no change in control group. age 5. · Oral rehydration therapy and feeding improved by 50%; little or no change in control group. Nicaragua: · Comparison of MSP and · Productivity in EMPs was · EMPs' penalties have · Increasing patient satisfaction Comparison of INSS financial perfor- 3 times higher than in included 25 minor infrac- with EMPs, no MINSA data. contractor and mance; growth in cover- MINSA in the outpatient and tions, 31 verbal reprimands, · Mean waiting time: 55 minutes at MINSA age, revenues, and emergency departments. 22 fines, 10 suspended, and an EMP, 78 minutes at MINSA. performance in expenditures; volume of · Almost 50% more efficiency 7 terminated. service delivery. services provided, by in inpatient care. · External drug purchases by type of services · The MSP's ALOS is half that patients: 16% in EMPs and · 23 consumer satisfaction of MINSA's. 40% in MINSA. indicators · Costs declined by 39% over · MSP market share rose from 15 years. 17% in 1993 to 33% in 1998 Panama: · 7 output indicators, · San Miguelito Hospital · Summary accreditation · Based on random sample of Comparison of including discharges, showed overall greater effi- compliance score for San patients, much higher satisfac- new hospital and number of visits, and ciency. Miguelito Hospital more tion with San Miguelito Hospital. 2 other hospitals. surgeries · ALOS is shorter. than 50% higher than for · 9 efficiency indicators, · Most productive for emer- other hospitals. (Some com- including ALOS and bed gency services and surgery; parisons were difficult to occupancy rates second on discharges. make because of lack of · 10 unit cost measures for · Lower unit costs for all but data from other hospitals.) Health inpatient and outpatient maternity. services · Higher management costs. System · 8 quality indicators cover- ing accreditation, death rates and caesarean births Innovations ALOS average length of stay; Q quetzal(es) (Guatemalan currency). in Central America 205 206 World Bank Working Paper ration care and control costs, but they have been adapted to simpler environments, and they work. The evolution of regulation is clearly laid out for Guatemala, and the ability of low- income governments to adopt a viable regulatory arrangement, however imperfect, shows that it not only can be done, but can produce measurable improvements among providers. Oversight and regulation of contractors posed a serious challenge to the five countries. Instead of being charged with delivery and management of public servants and public inputs, ministries of health were required to redefine their roles and functions. Although in some countries controlling corruption initially took center stage, it was rapidly replaced with more complex regulatory functions. The Guatemalan experience outlined in this volume suggests that the process was fitful and iterative, with constant adjustments responding to new requirements. Although often frustrating and time-consuming, this is not an unexpected pat- tern for the Central American ministries of health, given the newness of contracting out. Consumers tend to be overlooked in program design for health care reform. All of the country programs discussed in this volume have incorporated serious efforts to elicit feed- back. These range from waiting times in Costa Rica and Nicaragua, to expressed prefer- ences in Nicaragua and Panama, to tracking enrollment patterns and changes in Guatemala and Honduras. They have relied on interviews, enrollment numbers, household surveys, and user surveys, bringing to bear a solid arsenal of data and information. Based on per- ceived quality and convenience, consumers favored the private alternatives in all cases and, in many cases, "voted with their feet," shifting their health care utilization to new providers. Conclusions and Policy Implications The Central American innovations provide a rich set of lessons for the countries themselves, as well as for other developing countries contemplating reforms. Expanding services, even to remote areas, can be achieved and often improved with private contractors. It requires political will and the authority to oversee and enforce rules. The startup and learning process may seem like time lost in the short run, but as the Guatemala example shows, it is critical to long-term feasibility. Learning by doing characterized most of the innovations, partly because of their newness, limited experience, and lack of a blueprint for governments. Documentation of successful program implementation is possible only because of the data collection efforts and the introduction of performance objectives, both key to sound evaluation and policy impact. Effectively managing the process of reform requires the gov- ernment to continue to monitor and document trends and impacts, and to adjust the incentives and the rules over time in response to changes in circumstances. For example, capitation payments must keep up with inflation, or contractors will depart as costs outstrip revenue, as occurred in Nicaragua and Guatemala. Contractors also need to be compensated for new or different activities and the inclusion of costly groups. Otherwise, the system is jeopardized. Some of the shortcomings of the studies have to do with the structure of the evalua- tions, the nature of new initiatives, and sustainability. The lack of randomized allocation of enrollees in contracted programs leads to self-selection of dissatisfied users of public health clinics biasing the returns to the alternative investment. New facilities also tend to attract users for no other reason than the quality of the new surroundings, thus growth in number of clients and satisfaction may be inflated. Similarly, the lack of baseline data in all but the Health System Innovations in Central America 207 Honduran case compromises the measured impacts. All substitute comparisons with exist- ing public alternatives, which is a second-best option. Availability of time series data greatly enhances the amassed evidence, and the multivariate analyses complement the qualitative and cross-sectional information. Ideally, all studies should be able to control for both selec- tion bias and the multitude of factors that affect health care production and consumption. Each of the innovations represents a relatively small niche in a largely public health delivery system, and there is no evidence of intentions to expand beyond the initial partic- ipating communities. Consolidation of experiments remains an essential part of the process, but the lack of plans to expand successful approaches raises some concerns regarding true reform of lackluster public performance. Nicaragua's program has shown the greatest expansion, but it covers only 8 percent of the population. All of the experiments have been ongoing for at least five years, but sustainability remains a concern. The San Miguelito Hospital's dramatic successes could well fall prey to political dissatisfaction from other actors in the public health system. The lack of plans to extend the model raises concerns about political survival. The volunteer driven monitora program in Honduras could suffer from "monitora fatigue," where volunteers tire of the effort, government interest wanes, communities decide that child growth-monitoring con- cerns should be replaced by other issues, or some combination of these. Is it realistic to expect communities to constantly devote their meetings to the same set of issues around child growth? If bigger issues with a broader constituency or of a more immediate nature (for example, a natural disaster) come along, the child growth issue could easily be replaced. All successful start-up enterprises require nurturing to ensure their survival, as well as polit- ical support. Political pressures either to loosen the oversight or to abandon the innovation are often strong, which makes the persistence of the innovations in Central America all the more impressive. Although they are not perfect, the programs offer a rich array of possibilities for test- ing alternatives to publicly financed and delivered services. They demonstrate that com- plicated sets of incentives can be adapted to less sophisticated environments with felicitous, if not ideal, results. Most important, these studies indicate how important data and evalu- ation are to health policy. Without them, policy makers are grappling in the dark. Luckily, Central America has a body of work that can guide and inform policy in health care reform on both delivery and financing. Bibliography Dehn, J., R. Reinikka, and J. Svensson. 2003. "Survey Tools for Assessing Performance in Service Delivery." In F. Bourguignon and L. Pereira da Silva L., eds., The Impact of Eco- nomic Policies on Poverty and Income Distribution: Evaluation Techniques and Tools. Washington, D.C., and New York: World Bank and Oxford University Press. Kaufman, R. and J. Nelson. 2004. Crucial Needs, Weak Incentives: Social Sector Reform, Democratization, and Globalization in Latin America. Washington, D.C.: Woodrow Wilson Center Press and The Johns Hopkins University Press. Loevinsohn, B., and A. Harding. 2004. "Contracting for the Delivery of Community Health Services: A Review of Global Experience." Health, Nutrition and Population Discus- sion Paper. World Bank, Washington, D.C. 208 World Bank Working Paper Londoño, Juan Luis, and Julio J. Frenke. 1997. "Structured Pluralism: Towards an Innov- ative Model for Health System Reform in Latin America." Health Policy 41(1):1­36. Mills, A. 1998. "To Contract or Not to Contract? Issues for Low and Middle Income coun- tries." Health Policy and Planning 13(1):32­40. Ramirez, P. 2004. "A Sweeping Health Reform: The Quest for Unification, Coverage, and Efficiency in Colombia." In R. Kaufman and J. Nelson, eds., Crucial Needs, Weak Incentives: Social Sector Reform, Democratization, and Globalization in Latin America. Washington, D.C.: Woodrow Wilson Center Press and The Johns Hopkins University Press. Sapelli, C. 2004. "Risk Segmentation and Equity in the Chilean Mandatory Health Insur- ance System." In W. Jack and M. Lewis, guest eds., Special Issue on "Adjusting for Market Failure: Challenges in Public Health Alternatives." Social Science and Medicine 58(2):259­65. World Bank. 1994. The Organization, Delivery and Financing of Health Care in Brazil: Agenda for the 90s. World Bank Report 12655-BR. Washington, D.C.: World Bank. Contributors R icardoA.Bitrán is a health economist with 24 years of experience in Chile, the United States, and more than 35 developing countries. He received a Ph.D. in economics and an M.B.A in finance from Boston University, and an M.S. in industrial engi- neering from the University of Chile. His areas of expertise include the prioritization of health spending; the economic evaluation of health investment and reform projects; the analysis of equity, efficiency, and financial sustainability; and the formulation and evaluation of health policy. Since 1995, he has been President of Bitrán & Asociados, a leading international con- sulting firm based in Santiago, Chile, that specializes in health finance and economics. Previ- ously, he worked for 10 years with Abt Associates Inc. of Cambridge, Massachusetts, first as a health economist and then as senior scientist. While at Abt, he was research director for the HealthFinancingandSustainabilityProject.BeforejoiningAbt,Dr.Bitránwasafinancialana- lyst at the Massachusetts Department of Public Health. For the past 20 years, he has been a consultant for the World Bank, the Inter-American Development Bank, the World Health Organization, and the governments of two dozen developing countries. His academic experi- ence also spans more than 20 years, with teaching posts or assignments in Chile, the United States, France, China, Democratic Republic of Congo, and a dozen other developing coun- tries. Since 2000, Dr. Bitrán has been academic director of the World Bank Institute's Flagship Program on Health Sector Reform and Sustainable Financing for Latin America and the Caribbean. He is the author of academic articles, book chapters, and research papers. RodrigoBriceño is a health economist with Sanigest International, a health care consult- ing firm. Mr. Briceño's current research interests include the analysis of macroeconomics applied to the health sector, the performance of the health sector in Central American coun- tries, and the study of the new kinds of organizations for the provision of health services. He 209 210 World Bank Working Paper has worked on projects related to the health reform process in Latin America and Eastern Europe, and he has undertaken several economic evaluations of projects about HIV/AIDS and tuberculosis, family health projects, and the education sector in Costa Rica. One of his latest works is development of a framework for the economic analysis of hand-washing projects, based on the experiences of Central America and Peru. Mr. Briceño is also a professor of sta- tistics and macroeconomics in the National University of Costa Rica. He received his M.Sc. in development economics from the National University of Costa Rica in 2002. James Cercone is an economist and president of Sanigest Internacional, a health care management and consulting company. After graduating from the University of Michigan with a degree in mathematical economics, Mr. Cercone developed his interests in health eco- nomics and health reform through engagements with the World Bank (from 1990 to 1994) and the government of Costa Rica (from 1994 to 1997). Mr. Cercone has more than 14 years' experience in the health sector, with particular emphasis in Latin America and transition countries. He has worked in more than 40 countries, writing and developing solutions for the impact of corruption in the health sector, hospital restructuring, purchasing strategies, provider payment mechanisms, performance improvement, economic analysis of health projects, monitoring and evaluation, development of HIV/AIDS strategies, and management of health care services. Mr. Cercone has worked extensively on the design, implementation, and evaluation of projects for the World Bank, Inter-American Development Bank, Inter- national Labour Organisation and other development organizations. Carmen Cerezo is a Guatemalan physician who has worked extensively in developing health care models for underserved rural populations in her native country. She graduated as a Physician in 1984, from the national university San Carlos de Guatemala and went on to specialize as a Pediatrician, performing rotations both in Guatemala as well as in the United States. After practicing clinical medicine for some time, Dr. Cerezo shifted her pro- fessional interest to public health endeavors, co-founding in 1988 a non profit organiza- tion, GUATESALUD. This organization provides basic health care to farm and migrant workers living and working in large plantations in the south coast and northwestern part of the country. She was the Medical Director of GUATESALUD for eight years. Between 1997 and 2002 she worked for the Ministry of Health on the outreach, SIAS program. Upon leaving the Ministry of Health, Dr. Cerezo worked as an independent consultant up to year 2004 when she joined the HIV-AIDS, Global Fund Initiative for Guatemala as the Coor- dinator of the Planning and Programming Unit. Isabella Danel is a senior public health specialist detailed to the World Bank by the U.S. Centers for Disease Control and Prevention (CDC). She received an M.D. from Albany Medical College and an M.Sc in community health in developing countries from the London School of Hygiene and Tropical Medicine (LSHTM). She has worked at the CDC as an epidemiologist and public health specialist since 1993, focusing initially on maternal, child, and reproductive health. She has collaborated extensively with the Pan American Health Organization and other development agencies on monitoring and eval- uation issues. At the World Bank, she provides technical assistance to the Latin America and the Caribbean Regional Office on maternal and child health, noncommunicable dis- eases, and essential public health functions, with a focus on public health surveillance. She Health System Innovations in Central America 211 is currently working on projects in Argentina and Brazil. Formerly she was a research assis- tant at the LSHTM, working primarily in Central America. She has published a number of articles on health in the United States and internationally. John L. (Jack) Fiedler, senior health economist, has worked in the Latin America and the Caribbean Region of the World Bank since 2003. He earned a Ph.D. from Vanderbilt University, where he specialized in health and development economics and minored in Latin American studies. Before working at the World Bank, he was an international health consultant for 20 years in more than 30 countries. His work has covered a variety of topics, including demand and cost studies, cost-effectiveness analysis, designing and evaluating nutrition programs, assessing user fee systems, conducting longitudinal assessments of the performance of ministries of health, analyzing the private health care market and the pri- vate insurance industry, and designing insurance and privatization schemes. He has pub- lished more than 35 articles and a book on a variety of topics in health economics. VirunGauri is an economist in the Development Research Group (public services team). Mr. Gauri's current research interests include the uses and impact of social and economic rights, the political economy of government responses to HIV/AIDS, and the governance of nongovernmental organizations in developing countries. He has worked on and led a variety of operational tasks in the World Bank and the International Finance Corporation, including operational evaluations, investments in privately owned hospitals in Latin America, a social sector adjustment loan to Brazil, several health care projects in Brazil, and a study of the decen- tralization of health care in Nigeria. Mr. Gauri has previously taught courses in health policy, bioethics, and education policy in the United States and Chile. He received his Ph.D. in pub- lic policy from Princeton University in 1996. Paulina Gómez is a medical doctor who received her degree from the Catholic Univer- sity of Chile and holds a master's degree in public health from the University of Chile. She spe- cializes in public health and health economics and has more than 15 years of experience in Chile and elsewhere in Latin America. She is currently chief medical officer for Integramédica, Chile's largest private health provider network. Previously she was the director of San Borja- Arriarán Public Hospital. Her recent policy research work has dealt with financial equity, eval- uation of health care networks, costing and pricing of health care services, design of a social security health reform project, and training of policy makers in health policy for the World Bank Institute. She also designed the price schedule for Chile's public insurer. Marcia Griffiths is president of The Manoff Group, a company specializing in public health programs and social marketing. Ms Griffiths received a BA degree in anthropology and an M.Sc degree in nutritional sciences from the University of Wisconsin. For more than 25 years, she has provided technical assistance to a variety of programs in Latin America, Africa, the Middle East, and Asia. A special focus of her work is maternal and child health, particularly in developing systems to help communities participate in health and nutrition service delivery and in helping families and individuals improve health-promoting prac- tices. For the past decade, Ms. Griffiths has been dedicated to working with governments and nongovernmental organizations to develop growth promotion programs such as Hon- duras's AIN-C (Atención Integral a la Niñez en la Comunidad [Integrated Community 212 World Bank Working Paper Child Health Program). Currently there are 11 countries in which such schemes are ongo- ing. Ms Griffiths is the author of numerous publications, including the World Bank's "Pro- moting The Growth of Children: What Works" and the World Health Organization's IMCI Nutrition Food Box Adaptation Guide. GerardM.LaForgiaisaleadhealthspecialistattheWorldBank.HereceivedaB.S.degree from Northwestern University and an Sc.D. from School of Public Health at the University of Pittsburgh. He currently works and resides in Brazil, where he oversees the World Bank's health lending and research operations there. Before moving to Brazil, he spent 10 years con- ducting policy research and leading lending operations in Central America. He formerly was a health specialist at the Inter-American Development Bank and a Research Associate at the Urban Institute. He has published a number of articles and prepared many technical reports and monographs on health systems issues in Latin America. At present he is involved in pro- jects and research to improve primary care and hospital performance in Brazil. Maureen Lewis is a senior fellow at the Center for Global Development; she earned her Ph.D. at Johns Hopkins University. Dr. Lewis was formerly chief economist of the Human Development Network of the World Bank, and before that, she managed a unit in the Bank dedicated to economic policy and human development research and programs in Eastern Europe and Central Asia. Before joining the World Bank, Dr. Lewis established and directed the International Health and Demographic Policy Unit at the Urban Insti- tute. She has published widely on health and population in development and is currently an adjunct professor in the Graduate Program at George Washington University. Cecilia Má is a medical doctor from the National University of San Marcos in Peru. She holds a master's degree in public policy from the University of Chile and an M.S. in health administration from Peruvian University Cayetano Heredia. She specializes in pub- lic health and public policy and in health project evaluation, with work experience in more than a dozen countries in Latin America and the Caribbean. She is currently an adviser to the Inter-American Development Bank in Panama and in El Salvador, and to the govern- ments of both countries, to assist with the design and implementation of their health reform projects. Her professional experience in Peru involved the design and implemen- tation of national public health policies, the development of strategic planning to increase human capital resources in the health area, and the realization of innovative strategies for health education. For five years she was a manager at the Child Health Institute and the Cayetano Heredia General Hospital in Lima. She also has two years of experience as a con- sultant for health facilities and health care networks in Peru. Judith S. McGuire is currently an independent consultant based outside Washington, D.C. She has worked extensively on nutrition programs in Latin America, Sub-Saharan Africa, North Africa, and Asia. In 2001, she retired as senior nutritionist from the World Bank, where she had worked for 12 years. Before joining the World Bank, Dr. McGuire was a resident fellow at Resources for the Future. She received a B.A. in biology from Welles- ley College and a Ph.D. in nutritional biochemistry and metabolism from the Massachu- setts Institute of Technology. In 2005, she received an M.S. in social and organizational Health System Innovations in Central America 213 learning from the School of Public Policy at George Mason University. She was awarded the International Affairs Fellowship from the Council on Foreign Relations in 1985. Patricia H. Mintz is an independent consultant who has worked on health care issues with a variety of public and nongovernmental organizations over the past 25 years in Latin America and the United States. She currently serves as project director and consultant to the California Health Care Foundation's Local Coverage Expansion Initiative, working with local governments to increase coverage to low-income, uninsured children and to low-wage workers. Before her current consulting work, she served as a consultant to the World Bank and the Inter-American Development Bank on issues of insurance reform and extension of health insurance coverage to the poor. She began her health career with Kaiser Permanente. Ms. Mintz has a B.A. from Yale University and an M.B.A. from Stanford Uni- versity Graduate School of Business. Eco-Audit Environmental Benefits Statement The World Bank is committed to preserving Endangered Forests and natural resources. We print World Bank Working Papers and Country Studies on 100 percent postconsumer recy- cled paper, processed chlorine free. The World Bank has formally agreed to follow the rec- ommended standards for paper usage set by Green Press Initiative--a nonprofit program supporting publishers in using fiber that is not sourced from Endangered Forests. For more information, visit www.greenpressinitiative.org. In 2004, the printing of these books on recycled paper saved the following: Trees* Solid Waste Water Net Greenhouse Gases Electricity 307 14,387 130,496 28,262 52,480 * Pounds Gallons Pounds KWH 40" in height and 6-8" in diameter Health System Innovations in Central America is part of the World Bank Working Paper series. These papers are published to communicate the results of the Bank's ongoing research and to stimulate public discussion. Similar to developing countries elsewhere, during the 1990s Central American countries faced pressures to improve the performance of their health systems. For most countries, the consensus was that the systems were failing to live up to their potential. Rather than take on system-wide change, each country opted to step into reform through launching innova- tions to address specific problems or deficiencies in a particu- lar program, function, or intervention of the system. The studies that constitute this volume report on how these experiences fared--a hospital in Panama, a nutrition program in Honduras, primary care extension in Guatemala, a subset of hospitals and primary care units in Costa Rica, and a social security-managed health care program in Nicaragua. The studies report on the performance of the innovations, the policy environment in which they were developed as well as nuts-and-bolts features and processes incorporated into their design and implemantation. World Bank Working Papers are available individually or by subscription, both in print and online. ISBN 0-8213-6278-X TMxHSKIMBy362785zv,:<:#:(:) THE WORLD BANK 1818 H Street, NW Washington, DC 20433 USA Telephone: 202 473-1000 Internet: www.worldbank.org E-mail: feedback@worldbank.org