Better Spending, Better Care A Look at Haiti’s Health Financing COVER PHOTO CREDIT: LOGAN ABASSI UN/MINUSTAH Health Nutrition and Population Global Practice Latin America and Caribbean Region World Bank March 2017 Standard Disclaimer: This volume is a product of the staff of the International Bank for Reconstruction and Development/ The World Bank. The findings, interpretations, and conclusions expressed in this paper do not necessarily reflect the views 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 any judgment on the part of The World Bank concerning the legal status of any territory or the endorsement or acceptance of such boundaries. Copyright Statement: 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 to reproduce portions of the work promptly. 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, telephone 978-750-8400, fax 978-750- 4470, http://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, e-mail pubrights@worldbank.org. Table of contents ACKNOWLEDGMENTS 5 ABBREVIATIONS 6 EXECUTIVE SUMMARY 8 INTRODUCTION 16 BACKGROUND 20 HEALTH OUTCOMES AND THE HEALTH SYSTEM 26 HEALTH FINANCING 34 ACCESS TO HEALTH SERVICES 50 EFFICIENCY ANALYSIS 60 MAIN FINDINGS AND RECOMMENDATIONS 78 APPENDIX 84 BIBLIOGRAPHY 97 All dollar amounts are U.S. dollars unless otherwise indicated. BETTER SPENDING, BETTER CARE: A LOOK AT HAITI’S HEALTH FINANCING 3 ACKNOWLEDGMENTS T his report was prepared by Eleonora Cavagnero, Marion Cros, Ashleigh Dunworth and Mirja Sjöblom. Significant contributions were also made by Nicolas Collin Dit De Montesson, Alexo Esperato, Louise Estavien, Nelta Joseph, Francois Staco, and Isabelle Simeon. We are grateful for comments on an earli- er version of this report by Pierre Bonneau, Daniel Dulitzky, Michelle Keane, Sunil Rajkumar, Raju Singh, and Kanae Watanabe. Kassia Antoine and Ibrahim El Ghandour provided valuable support on interpre- tation and understanding of the BOOST data set and other national data sources. We also appreciate comments on an earlier draft of this report by three peer reviewers: Sarah Alkenbrack, Jean Kagubare, and Ajay Tandon. This study was conceived in partnership with the Ministry of Public Health and Population (Ministère de la Santé Publique et de la Population, MSPP) in Haiti. Special thanks go to the Minister of Public Health and Population, Dr. Marie Greta Roy Clément and her team, as well as the Director of the Evaluation and Programming Unit (UEP), Dr. Jean-Patrick Alfred, and those of the Unit of Contractualization (UC) and the Project Management Unit (PMU) of PASMISSI, Dr. Johnny Calonges and Dr. Wedner Pierre for their invaluable support. We are also very grateful to the Technical Committee on Health Financing and the many professionals and managers involved in the process who provided technical and logistical support throughout this study. We acknowledge with thanks the financial and technical support received from the World Bank’s Global Solutions Group on Health Financing. 5 ABBREVIATIONS AIDS Acquired immune deficiency syndrome ALOS Average length of stay ANC Antenatal care ANOVA Analysis of variance ASC Agent de santé communautaire (Community health worker) BOR Bed occupancy rate BSC Balanced Score Card CAL Centre de santé avec lit (Health center with bed) CDAI Centre Departemental d’Approvisionement en Intrants CEmOC Comprehensive Emergency Obstetric Care CHE Catastrophic health expenditure CNMP Commission Nationale des Marchés Publics (National Procurement Commission) CONAM Coordination National de l’Assurance Maladie (National Coordination of Health Insurance) CSL Centre de santé sans lit (health center without bed) DALY Disability-adjusted life years DASH Développment des Activités de Santé en Haïti (Development Activities and Services for Health) DDS Directions departementales sanitaires (departmental health directorates) DEA Data envelopment analysis DH Departmental hospital DHS Demographic and Health Survey DTP Diphtheria, tetanus, and pertussis ECVMAS Enquête sur les Conditions de Vie des Ménages après le Séisme (Survey on the Living Conditions of Households after the Earthquake) EPHS Essential package of health services GAVI Global Alliance for Vaccines and Immunizations HIS Health information system HIV Human immunodeficiency virus HR Human resources IMR Infant mortality rate LAC Latin America and the Caribbean LIC Low-income country MIF Multilateral Investment Fund MMR Maternal mortality ratio MPCE Ministère du Plan et de la Coopération Extérieure (Ministry of Planning and External Cooperation) MSH Management Sciences for Health MSPP Ministère de la Santé Publique et de la Population (Ministry of Public Health and Population) NCD Noncommunicable disease BETTER SPENDING, BETTER CARE: 6 A LOOK AT HAITI’S HEALTH FINANCING NGO Nongovernmental organization NHA National Health Account ODA Official development assistance OFATMA Office d’Assurance Accidents du Travail, Maladie et Maternité (Office of Insurance for Work Accidents, Illness and Maternity) OOP Out-of-pocket ORS Oral rehydration solution ORT Oral rehydration therapy PAHO Pan American Health Organization PDS Plan Directeur de Santé (Health Master Plan) PER Public expenditure review PES Package of essential services PFM Public financial management PHC Primary health care PIP Programme d’Investissement Public (Public Investment Program) PNS Politique Nationale de Santé (National Health Policy) RBF Results-based financing SARA Service Availability and Readiness Assessment SCD Systemic Country Diagnostic SDG Sustainable Development Goal SDI Schéma Directeur Informatique (IT Master Plan) SDI Service delivery indicator SDSH Santé pour le Développement et la Stabilité d’Haïti (Health for the Development and Stability of Haiti) SH Small hospital SPA Service Provision Assessment TE Technical efficiency THE Total health expenditure U5MR Under-5 mortality rate UAS Unité d’arrondissement de santé (district health unit) UH University hospital UHC Universal health coverage UN United Nations UPE Unité de Planification et d’Evaluation (Planning and Evaluation Unit) USAID U.S. Agency for International Development WASH Water, sanitation, and hygiene WDI World Development Indicators (database) WHO World Health Organization 7 EXECUTIVE SUMMARY ASC T his report seeks to formulate a long-term vision for Haiti’s health sector to accelerate progress toward universal health coverage (UHC), a key objective of the government’s National Health Policy (Politique Nationale de Santé, PNS)–MSPP (2012). Progress toward this goal has been hindered by political instability and frequent natural catastrophes. Most recently, in October 2016, Hurricane Matthew wreaked havoc on Haiti’s health system. It has been estimated that at least 1,000 peo- ple died and 1.4 million Haitians were directly affected by the hurricane. Such disasters have influenced Haiti’s government and development partners by demand- ing a short-term focus on acute need priorities. This study aims to take a step back, assess Haiti’s health financing system, and identify critical constraints and opportunities to accelerate progress toward UHC and the health-related United Nations’ Sustainable Development Goals (SDGs) in the long term. The re- port compiles existing studies and information, and it PHOTO  CREDIT: VICTORIA HAZOU UN/MINUSTAH provides new analysis of larger data sets, as well as hospital financing data. To our knowledge, it is the first attempt to assess systematically the health financing system in Haiti. BETTER SPENDING, BETTER CARE: 8 A LOOK AT HAITI’S HEALTH FINANCING EXECUTIVE SUMMARY 9 Findings similar or lower maternal and infant mortality ratios, such as Rwanda ($125) and Eritrea ($51).2 This finding Although Haiti has made significant progress on also highlights issues of low efficiency in Haiti’s health key health outcomes since the 1990s, it still fares sector. worse than many low-income countries in terms of service coverage of key interventions and in pro- The efficiency of health providers could be greatly viding equitable access to health. Between 1990 improved. Service readiness is an issue across all fa- and 2015, maternal and child mortality fell by about cilities, and present levels of health worker productiv- half. And yet the maternal mortality ratio and the un- ity is very low. An analysis of how efficiently health in- der-5 mortality rate have to decline further–by 80 per- puts are turned into health services reveals that Haiti cent and 64 percent, respectively, by 2030–to attain has very low technical efficiency scores compared the SDGs. Compared with other low-income countries with those of other LICs (Zere et al. 2006; Akzaili et (LICs), Haiti has low coverage rates of basic services. al. 2008; Sebastian and Lemma 2010; Marshall and For example, according to the 2012 Demographic and Flessa 2011; Hernandez and Sebastian 2013; Kirigia Health Survey (DHS) in Haiti, the coverage of institu- and Asbu 2013; Jehu-Appiah et al. 2014; Osmani tional deliveries was 37 percent–the Low and Middle- 2015). Dispensaries are the most inefficient type of Income Countries (LMICs) average is 70.5 (Joseph et health facility, and the inefficiency of the remaining al. 2016)–and the percentage of children under 24 facility types–health centers without bed (centres de months who received all three diphtheria, tetanus, and santé sans lit, CSLs), health centers with bed (cen- pertussis (DTP) vaccine doses. Meanwhile, service cov- tres de santé avec lit, CALs), and hospitals–follows erage was dramatically lower for the poorest wealth accordingly. Thus primary care level units are partic- quintiles–for example, deliveries in health care facilities ularly inefficient. Other measures of efficiency at the were eight times more frequent (76 percent) for the hospital level, such as bed occupancy rate, confirm highest wealth quintile than for the lowest quintile (9 the low productivity of hospitals. One reason facili- percent). The disparity in utilization mirrors the inequal- ties are inefficient is low staff productivity levels. For ity in health outcomes in Haiti. For example, growth example, medical staff see only six patients a day (less was stunted in 31 percent of children in the lowest than one patient per hour). Productivity is also neg- wealth quintile but only 6 percent of children in the atively influenced by absenteeism, which contributes highest wealth quintile (DHS 2012). to the waste of approximately $3 million per year (MSPP forthcoming3), moonlighting, and limited ser- The overall health expenditure in Haiti is high rel- vice readiness. A recent study of health facilities in ative to those of the LICs, but health outcomes three departments revealed that the medical staff are not significantly better, which points to low in primary health care (PHC) facilities work only four overall efficiency in the health sector. Haiti’s total hours a day but are actually paid a full-time4 salary health expenditure (THE) as a proportion of its gross (World Bank, USAID, and MSPP 2013). Furthermore, domestic product (GDP) is 7.6 percent, which is higher only 32 percent of health facilities provide essential than the average for the LICs (5.7 percent) and com- medicines,5 and only 31 percent possess basic medi- parable to the average for the Latin America and the cal equipment. Other key factors contributing to low Caribbean (LAC) region (7.2 percent). Haiti’s THE per productivity at the hospital level are poor functioning capita is $131 in international dollars, which is much of the referral system and poor utilization rates. The higher than the LIC average ($93) but much lower fact that the Ministry of Public Health and Population than the LAC region average ($1,113).1 Nevertheless, (Ministère de la Santé Publique et de la Population, value-for-money is low because the level of spending MSPP) allocates 90 percent of its operating budget to in Haiti is much higher than in other countries with personnel costs means that operational budgets are 1 These figures are in international dollars (at constant 2011 prices, purchasing power parity–adjusted). 2 World Development Indicators (database) 2016, World Bank, http://data.worldbank.org/products/wdi. 3 This publication, developed in partnership with the U.S. Agency for International Development (USAID), has not been released, but it was drafted in September 2014. 4 Here, “full-time” refers to a workday of eight hours. 5 Facilities were considered to have basic access to essential drugs if at the time of the survey they dispensed at least half of the 14 medicines in the Service Availability and Readiness Assessment (SARA) list of the World Health Organization (WHO 2010b). BETTER SPENDING, BETTER CARE: 10 A LOOK AT HAITI’S HEALTH FINANCING too tight to ensure an adequate supply of essential of the earthquake, several capital investments in in- drugs and equipment. frastructure were funded by development partners in the form of donations to the MSPP. Since then, the Even though it would be more cost-effective to in- MSPP has found the operational costs necessitated by vest in primary care, large allocations of resources these capital investments to be unaffordable–a situa- to hospital care persist, which is one reason why tion that has posed further challenges to funding the value-for-money is low. Currently, Haiti only spends health sector. In other words, the post-catastrophe re- 19 percent of its total health expenditure on preven- sponse has often taken the form of construction or re- tive care, whereas 54 percent is spent on curative care. habilitation of hospitals without planning for how the Furthermore, the number of dispensaries per capita running costs will be met after the initial emergency (the dispensary is the key facility for the provision of has passed. Consequently, hospitals are currently lack- primary care) is much lower than the average of other ing the basic resources to ensure service delivery, and LICs, while the number of community referral hospitals the MSPP is unable to meet these increasing opera- (hôspitaux communautaires de référence, HCRs)6 per tional costs, which is affecting its capacity to ensure capita is much higher (MoHSW 2008; Awate 2014; staff recruitment, training, and the provision of medi- Ujoh and Kwaghsende 2014). However, the three cal equipment and commodities. leading causes of disability-adjusted life years (DALYs) in Haiti are the human immunodeficiency virus (HIV), Meanwhile, for the poorest Haitians health care acute respiratory infections, and diarrhea, all which is unaffordable. After the 2010 earthquake, out-of- could be addressed by preventive and primary health pocket expenditures as a fraction of total health expen- care interventions. This evidence on Haiti’s disease bur- diture fell to 26 percent (2011), which is about 10 per- den indicates that it would be much more cost-effec- cent lower than in 2009.8 However, this study shows tive to increase coverage of promotional and preven- that out-of-pocket expenditures increased steadily in tive health services at the primary care level than to the years that followed and reached 35 percent in maintain the current density of hospitals per capita. 2014.9 The incidence of catastrophic health expendi- tures (CHEs)10 has also increased, and vulnerable pop- Inefficiencies in both domestic and external fund- ulations, such as those hospitalized, the unemployed, ing are exacerbated by the fragmentation and lack and households with more than three children under of coordination of external aid. After the 2010 earth- 5, are the most affected.11 Almost all health facilities quake,7 it appears that a large share of external emer- (93 percent) charge user fees; this financial burden falls gency funding focused on strengthening infrastruc- heaviest on the poorest segments of the population. ture, particularly the construction and rehabilitation of In fact, nearly two-thirds (63 percent) of households in hospitals. Because Haiti did not have a strong coordi- the lowest wealth quintile do not consult a health pro- nation mechanism in place at that time and 90 percent vider because they cannot afford to do so. of external funding is off-budget, it has been difficult to track, monitor, and plan how these resources are Haiti’s health financing system has undergone pro- applied to the health sector. As a consequence, this found change over the last two decades, partic- funding has not been maximized to facilitate long-last- ularly since the 2010 earthquake. Government fi- ing and positive impacts. In the immediate aftermath nancing of health care has also declined sharply in 6 The density of dispensaries and community referral hospitals (hôspitaux communautaires de référence, HCRs), was estimated using the 2013 SPA data set–Service Provision Assessment (Évaluation de la Prestation des Services de Soins de Santé, EPSSS), Haitian Institute of Childhood and ICF International, http:// dhsprogram.com/what-we-do/survey/survey-display-442.cfm. The density of community hospitals included small hospitals. Although a small hospital is not classified as a community referral hospital, these hospitals have a similar bed capacity and staff, and thus could be regrouped. According to SPA, there were 40 HCRs and 65 small hospitals in 2013. 7 A catastrophic 7.0 magnitude earthquake struck Haiti in 2010. Over 100,000 Haitians died, and millions were displaced. The infrastructure damage was extensive; the earthquake destroyed approximately 105,000 homes and damaged more than 208,000. It also left more than 1,300 educational establishments and 50 health centers and hospitals completely unusable (World Bank 2010a). 8 Global Health Expenditure Database (GHED), World Health Organization, http://apps.who.int/nha/database/Select/Indicators/en. 9 Global Health Expenditure Database (GHED), World Health Organization, http://apps.who.int/nha/database/Select/Indicators/en. 10 A household that allocates at least 25 percent of its nonfood consumption to health is considered to be encountering catastrophic health expenditures or financial hardship related to health (WHO and World Bank 2015). 11 Survey on the Living Conditions of Households after the Earthquake 2013 (Enquête sur les Conditions de Vie des Ménages après le Séisme, ECVMAS), Haitian Institute of Statistics and Data Processing, http://catalog.ihsn.org/index.php/catalog/5360. EXECUTIVE SUMMARY 11 Haiti over the last two decades, while external financ- to measure progress toward UHC should be incor- ing has increased. Between 1995 and 2014, public porated into the investment case. health expenditure as a fraction of total health expen- diture decreased by half, lowering from 41 to 21 per- 2. Increase equitable access to quality care. Update cent.12 External health financing reached record lev- and implement a facility mapping tool by re-classi- els of about 70 percent of THE in 2011 as a result of fying health facilities to enhance service readiness the large inflow of emergency aid in response to the and facilitate a practical referral network. Facilities earthquake. Nevertheless, because external financing should be properly (re)classified and a popula- has decreased sharply in recent years and domestic fi- tion-based carte sanitaire (facility mapping) devel- nancing is not increasing in proportion to this decline, oped to ensure systematically that all facilities in- households are bearing a growing burden of health cluded in the referral network meet the minimum costs, with grim implications for the poorest segments criteria in terms of service readiness, which will vary of the population. by type of facility. The MSPP should therefore de- velop a facility mapping tool to (1) identify the ex- isting public and private facilities; (2) establish their Recommendations: Seven Strategic service readiness (mostly in terms of staff and in- Shifts puts); and (3) determine the population coverage of each facility. The first step would build on the ex- isting carte sanitaire that emerged from the Service Based on these findings, we identified seven stra- Provision Assessment (SPA) survey, which was a tegic shifts that would accelerate the progress to- census of all health facilities in Haiti and a mapping ward universal health coverage in Haiti: of the services actually being delivered in each fa- cility. The findings of such a mapping tool would 1. Prioritize primary health care. Realign resourc- identify service gaps or redundancies and trigger es from hospital to primary health care and cost a re-categorization of certain facilities. However, it and prioritize the existing Health Master Plan (Plan would not necessarily mean building new dispen- Directeur de Santé, PDS) to guide future financ- saries. Taking into consideration the investment ing. As Haiti undergoes epidemiological transition, priorities defined in the Plan Directeur (see Shift it also takes on the double burden of disease that 1), certain inefficient community referral hospitals accompanies this change –the main causes of mor- could be transformed into health centers that offer bidity and mortality are now attributable to both health promotion services and primary care. In oth- communicable and non-communicable diseases. er cases, certain facilities could be converted into Since primary care models and preventive health primary health care units, or upgraded to hospitals, services target the root causes of both communi- or given special attention to ensure service readi- cable and non-communicable diseases, they would ness. Merged facilities would be better equipped yield the highest rate of return on investment. The with drugs and medical equipment. For this exer- MSPP and development partners should spearhead cise, it would be crucial to have a well-defined es- the development of a joint investment case (or stra- sential package of health services to be financed at tegic plan) to guide investments in the sector and the primary care level. to shift resources to the primary care level. Such a document would use the existing Plan Directeur 3. Spend more wisely on hospitals. In the short run, and the essential package of health services (EPHS) consider placing a moratorium on new hospital as starting points and would prioritize and cost a construction until the existing infrastructure can be few focus areas or interventions on which MSPP mapped and a hospital licensing program has been and development partners could focus their financ- developed. The MSPP should also encourage de- ing. Innovative and cost-effective models for de- velopment partners to finance technical assistance livering health care, particularly at the level of the for hospitals. The ongoing externally financed wave community, should be considered. And, indicators of hospitals construction was not accompanied by 12 Global Health Expenditure Database (GHED), World Health Organization, http://apps.who.int/nha/database/Select/Indicators/en. BETTER SPENDING, BETTER CARE: 12 A LOOK AT HAITI’S HEALTH FINANCING plans to sustain hospitals’ operational costs and improvements in efficiency. Linking financing for in- maintain service delivery. Consequently, hospitals dividual staff and facilities to outcomes through re- are currently lacking the basic resources to en- sults-based financing (RBF) mechanisms is one pos- sure service delivery, and the MSPP does not have sible way to strengthen accountability and thereby enough financing to meet the increasing opera- lift productivity. Thus RBF could serve as an efficient tional costs, thereby affecting its capacity to ensure tool for improving the productivity of human re- staff recruitment, training, and the provision of sources and making health facilities more account- medical equipment and commodities. In the short able in terms of results, as demonstrated by the first term, no new hospital should be built unless it re- findings from the promising pilot of the national sponds to the urgent functional or geographical RBF program now being implemented. The avail- needs that will remain beyond the emergency pe- ability of medicines could also be improved by re- riod. Technical assistance should focus on business vamping supply chain management. Considerable plans that can financially sustain hospital infrastruc- savings could result from enhancing the coordina- ture that is being or has been handed over to the tion of the distribution network and focusing on government. Revenue generation strategies that last-mile distribution, potentially by outsourcing to might entail, for example, luxury wards for patients local transport companies, which has been suc- who have a high willingness to pay, or cost-cutting cessfully piloted in Haiti. strategies for hospital care, or alternative sources of revenue, such as from very wealthy individu- 5. Better use of external funding. To increase impact als, diaspora, or religious organizations, should be and enforce adherence to a costed and prioritized considered. Plan, Haiti should have an adequately staffed and well-functioning donor coordination unit that pur- 4. Improve technical efficiency at PHC level. Value- sue donor tracking and transition planning. The do- for-money in service delivery should be increased nor coordination unit would, among other things, by reforming human resources, having better avail- maintain the national database of cooperation proj- ability and use of inputs (particularly medicines) and ects and ensure that there is complementarity and serving more patients, especially at the first level of that transition plans (especially when donors are care. While facilities are being recategorized and withdrawing) match health system needs with the basic equipment and medicines are being better available resources. The MSPP should enforce reg- distributed (Shifts 1 and 2), it is vital to improve istration of development partners with the donor technical efficiency. Increasing value-for-money coordination unit (other countries have enforced will require increasing patient flow and reforming that practice by decree). In the short term, develop- human resources (among other things, the decen- ment partners should begin to pool external financ- tralization of certain decisions) in order to reduce ing virtually around the essential package of health absenteeism and improve recruitment and work- services and key interventions identified in the cost- ing conditions. Primary care facilities in Haiti are ed and prioritized Plan Directeur (or the investment less efficient than those in other low-income coun- case). Some partners have launched this process for tries. Low productivity characterizes health facilities a limited set of services in the context of the RBF across all categories–primary care dispensaries and program. Meanwhile, all donors should follow a health centers with and without beds are already standard reporting format, which would be devel- known to be especially inefficient. Low productivity oped by the donor coordination unit (together with can be explained in part due to high levels of ab- the development partners). At the same time, the senteeism and moonlighting by health personnel. MSPP and development partners should strength- This situation is likely exacerbated by low levels of en public financial management (PFM) structures to demand from prospective patients in poor commu- make it possible to set up a SWAp13 mechanism to nities. Facilities are not properly classified in terms pool external financing in the future and strengthen of the minimum criteria, and referral networks the capacities at the departmental level (including are not in place (see Shift 2), all of which impede planning, budgeting, monitoring, and reporting). In 13 Sector Wide Approach in health policy. EXECUTIVE SUMMARY 13 the short term, harmonized procedures and agree- toward UHC. On the external revenues side, Haiti ments among partners on levels of per diems and should work toward increasing external financing salaries could slash transaction costs. To this end, the and rally external partners around a more sustain- health ministry and development partners should able contribution in line with the Plan Directeur, draft and sign a memorandum of understanding to which implies working on long-term financing strat- identify minimum standards for emergency financ- egies to achieve UHC. Finally, vaccines in Haiti are ing–for example, including requirements that major now entirely financed by donors – unlike in most capital investments such as hospitals be supported low-income countries – and this needs to change. by long-term plans. Without significant government cofinancing, key donors will be unable to continue financing vac- 6. Increase resources for health. In light of decreas- cines in the country for much longer. es in external financing, it is particularly important to increase domestic resources for health and to 7. Increase affordability of health services for the ensure that domestic financing allocations address poor. The feasibility of removing user fees for se- key priorities while leveraging donor financing for lected services or target populations (for example, essential items such as vaccines. Despite pressing children under 5 and pregnant women, especially health care needs, Haiti has seen a sharp drop in in rural areas) should be assessed. User fees nega- government expenditure in the health sector over tively affect not only equity in access but also effi- the last two decades, with a consequent increase in ciency of health facilities and ultimately health out- donor dependency. In the past, Haiti’s health sec- comes. Almost all health facilities charge user fees to tor received allocations of between 9 and 14 per- bridge the gap in funding. As a result, out-of-pock- cent of the national budget. In 2014, the share of et spending and thus catastrophic health expendi- government expenditure going to health was just tures are increasing. In 2013 almost one-quarter 6.1 percent of the total government expenditure, of households reported not consulting a provider well below the Abuja Declaration’s recommended when sick, and, among those, 49 percent could allocation of 15 percent14 and has since fallen to not afford care. However, because user fees are just 4.5 percent in the 2016-17 budget. In addition, currently an important part of the operating bud- donor financing is decreasing, and thus the gov- gets of health facilities, their removal needs to be ernment urgently needs to plan for increasing do- carefully assessed so it will not affect the availability mestic financing for health to avoid a spike in out- or worsen further the quality of the services pro- of-pocket expenditures. Increasing public spending vided. Mechanisms to increase the affordability of on health may imply an increase in domestic re- health services for the poorest should be pursued. source mobilization as a whole or specifically for These include a transportation voucher program or the health sector. One way of achieving the lat- the revival of the equity fund at the facility level ter is by introducing earmarked taxes for health. to protect the poorest from the direct and indirect Either way, the MSPP should build a strong case costs of health care. The mobile clinics and services for the Ministry of Economy and Finance (Ministère provided by community health workers are mostly de l’Economie et des Finances, MEF) to invest in used by the poor and should be strengthened. As the health sector. For that, it is essential to show discussed in Shift 1, more resources should be allo- enhanced value-for-money, improved budget ex- cated to expand and strengthen community care in ecution rates, and a vision to accelerate progress order to move further toward UHC. 14 In 2011 African heads of states approved the Abuja Declaration, which sets a target of allocating15 percent of a government’s total expenditure to health. This target can be regarded as aspirational, as it is currently reached only by some countries. BETTER SPENDING, BETTER CARE: 14 A LOOK AT HAITI’S HEALTH FINANCING CHAPTER 1 INTRODUCTION T his report describes how Haiti can accel- erate and sustain progress toward univer- sal health coverage (UHC). A key objective of Haiti’s National Health Policy (Politique Nationale de Santé, PNS) is to attain uni- versal health coverage (MSPP 2012).15 However, be- cause of Haiti’s political instability and high frequen- cy of natural catastrophes–most recently, Hurricane Matthew in October 2016 in which reportedly at least 1,000 people died, with 1.4 million directly affected and 175,000 internally displaced16–both national and international development partners have tended to fo- cus on emergency needs and short-term measures to improve the health sector. This analysis aims to redirect that approach toward a long-term vision for the sector. 15 The 2012 National Health Policy establishes the vision of attaining over the next 25 years the universal delivery of an essential package of health services (MSPP 2012). 16 This report was written largely before Hurricane Matthew struck Haiti in PHOTO CREDIT : SOPHIA PARIS UN/MINUSTAH 2016. However, the systemic challenges to Haiti’s health system have not changed. Moreover, pre-hurricane trends indicated that external funding, which surged after the 2010 earthquake, had dropped sharply, and economic growth was slowing in 2016. Since Hurricane Matthew, the prospects for economic growth in 2017 are even lower, and the domestic revenues and the budget available for all sectors, including health, will decrease. Although there has been a temporary–and modest–spike in emergency financing for the hurricane response, external financing is expected to approach pre-hurricane trends in 2017. Therefore, the analysis and policy recommendations in this report remain valid in the post–Hurricane Matthew period and are therefore relevant to government and partners in shaping the reconstruction efforts. BETTER SPENDING, BETTER CARE: 16 A LOOK AT HAITI’S HEALTH FINANCING Chapter 1 • INTRODUCTION 17 In doing so, it identifies a set of critical constraints to an efficient way. Our study places a special emphasis overcome and opportunities to seize to move toward on measuring value-for-money in Haiti’s health sector UHC. The recommendations are intended to guide not by examining the ability of the health system to turn only Haiti’s government but also its development part- resources into health services that result in improved ners, who play an important role in advancing Haiti’s health outcomes for the population. health care system. This study compiles existing information and pulls UHC is a moving target, and it includes dimensions together new analysis of recent data. The questions such as coverage and quality of services as well included in the Health Financing System Assessment as financial protection. For countries like Haiti, with template20 were used as a starting point for the study. low coverage of basic health services, UHC is achieved It also builds on the analysis carried out for the pov- gradually. The first step is to prioritize and strength- erty assessment and public expenditure review (PER) en the primary level of health care to enable a con- in Haiti. Additional analysis includes study of the de- tinual scale-up of essential services for the vulnerable terminants of catastrophic health expenditures (CHEs), and poor populations. Increasing the number of indi- the drivers of inefficiency (including human resources), viduals with access to health services is an important and health-seeking behaviors. Meanwhile, new data dimension of UHC. Quality of services is another im- were collected on hospital financing, and an efficien- portant aspect, as well as financial protection for all. cy analysis was carried out for all facilities. New analy- Countries must avoid placing those needing health sis was also conducted using the 2013 Survey on the services in the position of having to choose to forgo Living Conditions of Households after the Earthquake health care because of financial issues or accept the (Enquête sur les Conditions de Vie des Ménages impoverishment that may result from out-of-pocket après le Séisme, ECVMAS) and the BOOST data set.21 (OOP) expenditures. The focus of this study is aligned with recent com- This report describes these important dimensions, mitments to UHC at the global and country lev- including the level of health care coverage,17 equity els. Its objective is consistent with the United Nations’ in access to health services,18 and financial protec- Sustainable Development Goals (SDGs) and the World tion19 in Haiti. It also discusses the three basic func- Bank’s strategy of eliminating extreme poverty and tions of health care financing: (1) revenue collection– boosting shared prosperity. The achievement of UHC, to raise enough revenue to provide individuals with a in which all people are effectively covered by essen- package of health services that ensures, in an equita- tial health services and no one suffers undue financial ble, efficient, and financially sustainable manner, finan- hardship because of illness, is key to reaching these cial protection against catastrophic health expenses twin goals. The focus of this study is also consistent arising from illness and injury; (2) pooling–to manage with the Systematic Country Diagnostic (SCD) and the these revenues to pool health risks equitably and ef- Country Partnership Framework for fiscal years 2016– ficiently; and (3) purchasing–to ensure that the pay- 19 for Haiti, particularly in the strategic area of building ment for or purchase of health services is carried out in human capital, with the objective of increasing access 17 Coverage includes indicators for preventive care such as family planning requirements, at least four antenatal consultations, vaccinations, and improved water sources. In addition, health care coverage includes indicators of curative services such as hypertension treatment, diabetes treatment, TB detection, skilled birth attendance, and antiretroviral therapy. 18 Equity in coverage is measured by assessing prevention and treatment service coverage by wealth quintile. 19 Financial protection is assessed by examining the proportion of households who spend a certain threshold (in this report 25 percent, which is usually used) of their nonfood expenditures on health care or are impoverished because of out-of-pocket payments. 20 This template was recently developed by the World Bank’s Health Financing Global Solutions Group. 21 The following data sets and surveys were used in this study and are cited throughout in shortened form: BOOST–Database of Public Budget Expenditures, World Bank, http://wbi.worldbank.org/boost/boost-initiative; DHS–Demographic and Health Survey, U.S. Agency for International Development, http://www.dhspro- gram.com/; ECVMAS–Enquête sur les Conditions de Vie des Ménages après le Séisme (Survey on the Living Conditions of Households after the Earthquake), Haitian Institute of Statistics and Data Processing, http://catalog.ihsn.org/index.php/catalog/5360; GHED–Global Health Expenditure Database, World Health Organization, http://www.who.int/health-accounts/ghed/en/; GBD (Global Burden of Disease) Compare–Institute for Health Metrics and Evaluation, https:// vizhub.healthdata.org/gbd-compare/; MGAE–Module Gestion de l’Aide Externe (External Aid Management Module, Haiti): MPCE–Ministére de la Planification et de la coopération externe (Ministry of Planning and External Cooperation); SNPPGD–Systéme national de planification, de programmation et de gestion du développement (National System of Planning, Programming and Development Management), https://haiti.ampsite.net/portal/; SPA–Service Provision Assessment (Évaluation de la Prestation des Services de Soins de Santé, EPSSS), Haitian Institute of Childhood and ICF International, http://dhsprogram.com/what-we-do/ survey/survey-display-442.cfm; WDI–World Development Indicators, World Bank, http://data.worldbank.org/data-catalog/world-development-indicators. BETTER SPENDING, BETTER CARE: 18 A LOOK AT HAITI’S HEALTH FINANCING to health services. The study was conceptualized with system, and chapter 4 then turns to health financing the Ministry of Public Health and Population (Ministère and discusses resource mobilization (domestic, exter- de la Santé Publique et de la Population, MSPP) and nal, and private financing), pooling, and purchasing. key health system stakeholders in Haiti, and comple- Chapter 5 examines access to health services, and ments other ongoing analytical activities. chapter 6 describes the efficiency of the health system in producing the services needed by the population. This report is organized in seven broad chapters. Chapter 7 concludes by discussing the main findings Chapter 2 provides context for the overall report by of the study, describing the key strategic shifts need- describing the macro and fiscal situations in Haiti. ed to move towards UHC in Haiti, and offering policy Chapter 3 describes health outcomes and the health recommendations. Chapter 1 • INTRODUCTION 19 CHAPTER 2 BACKGROUND H aiti is one of the most unequal countries in the world, and most of the population is poor. Haiti ranks 163rd out of 187 coun- tries on the Human Development Index and remains the most unequal country in the Latin America and the Caribbean (LAC) region (Gini, 0.6). Overall, the poverty headcount is about 59 percent, and 24 percent of the population lived in extreme poverty in 2012, indicating that almost 6.3 million Haitians cannot meet their basic needs, and 2.5 million cannot even cover their food needs (World Bank 2016b) Based on the international pov- erty lines, 54 percent of the population lives on less than $1.90 a day and 71 percent on less than $3.10 a day.22 In 2014 only 25 percent of the population had access to electricity, which is lower than the average of low-income countries (LICs) overall, and Haiti’s un- employment rate remains one of the highest in the LAC region at 30.1 percent (World Bank 2015g). Haiti also has the lowest rate of labor force participation in the region: only 60 percent of working-age individ- uals participate in the labor market, compared with, PHOTO CREDIT : VICTORIA HAZOU UN/MINUSTAH for example, 70 percent in the nearby Dominican Republic (World Bank 2015g). Ninety-three percent of the population works in the informal sector (Herrera et al. 2014), making it difficult to set up a national and 22 In constant 2011 prices, purchasing power parity–adjusted. The global poverty lines are now set at $1.90 and $3.10 a day, using 2011 prices. Previously, the values for extreme and moderate poverty were $1.25 and $2.50 a day, respectively. BETTER SPENDING, BETTER CARE: 20 A LOOK AT HAITI’S HEALTH FINANCING Chapter 2 • BACKGROUND 21 FIGURE 2.1: Annual Trends in GDP, 2013–15, and Forecasts, 2016–18: Haiti 5 14 12.3 Percent change in the real GDP 12 4 Percent rate of inflation 10.7 10 7.5 3 8.6 8 6.8 2 6 3.9 4 1 4.2 2.8 1.7 0.9 1.9 1.7 2 0 2013 2014 2015 2016f 2017f 2018f 0 Real GDP Inflation (average) Sources: Ministry of Economy and Finance, Bank of the Republic of Haiti, International Monetary Fund, and World Bank staff calculations. public health insurance system because those mech- the 2014 rate, thereby limiting the fiscal space for ex- anisms require levying taxes on a formal workforce. panding funding to the health sector. Only 5 percent of the population is enrolled in a com- pulsory health insurance program (see chapter 4), and Improved tax collection is one way to increase do- they are primarily formal sector workers. There is no mestic revenues for health. However, although tax government policy to protect vulnerable populations mobilization rose after the earthquake, it was likely to from health-related financial losses. fall in 2016. From 2005 to 2015, revenue as a share of GDP increased by nearly 50 percent, from 13.1 to In 2016 economic growth slowed in Haiti. Although 18.3 percent of GDP (World Bank 2016a). This im- the economy may rebound in 2017, gross domestic provement stemmed primarily from external grants, product (GDP) growth will remain low. In 2014 Haiti’s which increased from 3.5 percent of GDP in 2005 to gross national income (GNI) per capita was $800, mak- 12.1 percent in 2010. The fiscal revenue picked up ing it a low-income country. Between 1999 and 2014, as well, moving from 9.6 percent of GDP in 2005 to Haiti’s average GDP annual growth rate was 1.27 13.6 in 2015, but it was expected to decline to 13.5 percent, but after the 2010 earthquake (in 2011 and percent of GDP in 2016. Since 2015, the fiscal deficit 2014), the average growth rate increased to 3.85 per- has remained below 3 percent of GDP (World Bank cent. This growth rate exceeded that in the LAC region 2016a). Public expenditures jumped from 13.5 per- (2.99 percent), but was below the average growth rate cent of GDP in 2005 to 23.2 percent of GDP in 2015 (6.31 percent) of other LICs (WDI 2015). In response to (World Bank 2016a). Public expenditures and rev- inflation and erosion of the international reserves (fig- enues increased initially after the 2010 earthquake, ure 2.1), GDP growth began slowing in Haiti in 2014 but both are expected to fall to 18.6 percent in 2018. and continued to decelerate to 0.9 percent in 2016. In addition, the recent decrease in domestic revenue mobilization is forcing a substantial decline in public The decline in GDP growth is affecting domestic investment (expected at 6.3 percent of GDP this fiscal revenues and shrinking the budget available for all year compared with 9.6 percent last fiscal year) –see sectors, notably health. The slowing economy is lim- figure 2.2. iting the scope for increasing public financing for the health sector in the short term. However, projections Haiti raises little tax revenue given its economic indicate that GDP may rebound in 2017 (figure 2.1), status, but there is scope to raise more. Haiti has which would present an opportunity to increase the the second-lowest tax-to-GDP ratio (13.7 percent) of government’s contributions to equitable and efficient all countries in the LAC region and one that is only health financing for universal health coverage (UHC) slightly better than the average for LICs. Its tax-to-GDP in the medium term. That said, the forecasts for 2017 ratio is 1.07 times higher than that of the LICs, but and 2018 indicate that GDP growth will remain below its GDP per capita is 1.36 times higher than the LIC BETTER SPENDING, BETTER CARE: 22 A LOOK AT HAITI’S HEALTH FINANCING FIGURE 2.2: Fiscal Account as Percentage of GDP: Haiti, 2013–18 35 30 25 20 Percent 15 19.1 24.1 25.5 28.3 27.7 25.8 10 5 0 0 -2.2 -1.5 -5 -4 -7.1 -6.3 -10 2013 2014 2015 2016f 2017f 2018f Fiscal balance Public debt Total revenues Total expenditures Sources: Ministry of Economy and Finance, Bank of the Republic of Haiti, and World Bank staff calculations. average, which indicates that Haiti raises relatively lit- government’s reliance on donor financing over time. tle tax given its economic status and it should have a In 2010, 16 percent of financing for the social sectors higher tax-to-GDP ratio. If the country increases its tax- was foreign assistance, rising to 45 percent in 2012 to-GDP ratio to 15 percent,23 it could increase its fiscal (Singh and Barton-Dock 2015). Health, education, revenue by $18 per capita or 2 percent of GDP (IMF and social protection are the most aid-dependent sec- 2016)–see appendix A for a more detailed discussion tors in Haiti. The transition from high levels of exter- of domestic revenues. As indicated in the public expen- nal financing for the health sector post-earthquake to diture review (PER) for Haiti (World Bank 2016a), its tax the lower levels of external financing observed now system should undergo large-scale reforms. It is cur- needs to be managed. Although large efficiency gains rently regressive because the country’s fiscal revenues are possible in the health sector (see chapter 6), the rely heavily on indirect taxes, which affect consumers government should begin to plan to increase domes- independently of their income level. Haiti also may not tic financing for health to compensate for the drop in be exploiting its full revenue potential from corporate external aid and protect the poor from growing out- and personal income taxes (World Bank 2016a). Thus of-pocket expenditures. there is scope to raise more taxes, but that may not be feasible in the short term. Sin taxes on alcohol and tobacco are an interesting option for raising a substantial amount of revenue The health sector is highly dependent on external for the health sector while discouraging consump- financing. Because it is now decreasing, the govern- tion of these goods. Several countries are using taxes ment needs to plan to increase domestic financing for on alcohol and cigarettes to reduce the prevalence rate health to avoid a spike in out-of-pocket expenditures. of tobacco and alcohol use and to raise revenue for In both fragile states and LICs, net official develop- the health sector. Currently, Haiti has no tax on tobac- ment assistance (ODA) as a percentage of GDP fell co, and the tax rate is 4 percent for locally produced over the last decade (figure 2.3) By contrast, ODA in- spirits and 16 percent for imported alcohol. On aver- creased by 50 percent over the same period in Haiti. age, taxes account for 31 percent of the retail price of It peaked after the 2010 earthquake, but it has been cigarettes in LICs and 47 percent in the Latin America falling ever since. Although the availability of donor and the Caribbean region (WHO 2015). Thus there is assistance has enabled Haiti to finance an expan- scope for raising taxes on these products in Haiti. An sion in the social sectors, it has also increased the estimated $8.2 million a year, at a minimum, could be 23 Although 15 percent is an arbitrary choice, it is often suggested as minimum benchmark to reach. Tax shares of 20 and 25 percent may be difficult to achieve because of administrative and capacity constraints (Heller 2005, 2006; IMF 2011). Chapter 2 • BACKGROUND 23 FIGURE 2.3: Net Official Development Assistance as Percentage of GDP: Haiti, 2004–13 50 45 40 35 30 Percent 25 20 15 10 5 0 2011 2004 2005 2006 2007 2008 2009 2010 2012 2013 Haiti Fragile states Low-Income Countries Latin American and Caribbean Region Sources: WDI and World Bank staff calculations Note: The share of gross national income represented by net official development assistance in LAC countries fell below 1 percent at each annual interval on the chart. The share began at 0.34 percent in 1995, peaked at 0.37 percent in 1996, and ended at 0.17 percent in 2014. LAC = Latin America and the Caribbean; LICs = low-income countries. raised for the health sector if Haiti were to increase the goods, earmarking tax revenues for the health sector tax on alcohol24 to 25 percent and earmark the tax rev- can be justified. enue for health (see table A.3 in appendix A). The pro- ceeds from such a tax would represent a growth rate Earmarking taxes for the health sector raises tech- of almost 11 percent in per capita government health nical and political issues that warrant a thorough spending, or $0.76 per capita. It is difficult to estimate assessment. Such a step could be instrumental in rais- how much revenue could be generated from an in- ing domestic revenues for that sector, but success in crease in tobacco taxes because the sales numbers are levying such taxes will require sufficient administrative unknown (Josephson and Bode 2013). Sin taxes could capacity and information as well as alignment from to- increase the predictability of financing for the health bacco and alcohol corporations and lobbies. That said, sector, while reducing the consumption of alcohol and administrative capacity is an issue for the implemen- cigarettes and thereby improving the health of the tation of several possible tax reforms in Haiti. A more population and reducing health care costs. Because in-depth study should be conducted to assess the po- the health sector incurs a disproportionate cost com- litical feasibility of such reforms and to avoid potential pared with other sectors in the consumption of these negative impacts such as cross-border smuggling. 24 Estimates for revenue are based on the sales data for rum and beer for selected brands because countrywide data on alcohol sales are not readily available. BETTER SPENDING, BETTER CARE: 24 A LOOK AT HAITI’S HEALTH FINANCING CHAPTER 3 HEALTH OUTCOMES AND THE HEALTH SYSTEM Key Health Outcomes Despite Haiti’s progress on meeting the 2015 health-related Millennium Development Goals (MDGs) over the last decade, much work remains to reach the 2030 health-related Sustainable Development Goals (SDGs). Haiti’s maternal mor- tality ratio (MMR) fell from 670 maternal deaths per 100,000 live births in 1990 to 359 in 2015 (46 percent decline), and its infant mortality rate (IMR) and under-5 mortality rate (U5MR) fell by 48 percent and 52 per- cent, respectively (table 3.1). The SDGs aim to reduce the MMR to less than 70 maternal deaths per 100,000 live births and the U5MR to 25 or lower deaths per PHOTO CREDIT : LOGAN ABASSI UN/MINUSTAH 1,000 live births by 2030. To achieve these goals by 2030, Haiti will need to reduce the current MMR by a further 80 percent and the U5MR by 64 percent. Haiti sustained an average annual percentage change in maternal mortality of 2.2 between 1990 and 2015, and reduced its MMR by 29 percent between 2000 and 2015 (figure 3.1). Based on these trends, Haiti is not currently on track to achieve the SDG goal for the MMR in 2030. BETTER SPENDING, BETTER CARE: 26 A LOOK AT HAITI’S HEALTH FINANCING Chapter 3 • HEALTH OUTCOMES AND THE HEALTH SYSTEM 27 TABLE 3.1: Comparing Health Outcomes in Haiti, LICs, and LAC Region: 1990, 2000, 2013, 2015 % change, 1990 2000 2013 2015 SDGs 2030 1990–2015 Maternal mortality ratio <70 Haiti 670 510 380 359 –46% LICs 900 740 450 495 –45% LAC region 110 81 68 69 –37% Infant mortality rate – Haiti 100 85 54 52.2 –48% LICs 104 74 52 53.1 –49% LAC region 33.7 21.7 12.4 15.9 –53% Under-5 mortality rate 25 Haiti 144 104 72 69 –52% LICs 166 134 76 76.1 –54% LAC region 42 36 14 18.8 –55% Sources: WHO 2016; DHS 2000, 2005–06, 2012. Note: – = not available; LAC = Latin America and the Caribbean; LICs = low-income countries; SDGs = Sustainable Development Goals. FIGURE 3.1: Trends in MMR, U5MR, and IMR: Haiti, 1990–2015 250 700 225 600 200 NMR/U5MR per 1,000 live births MMR per 100 000 ive births 175 500 150 400 125 100 300 75 200 50 100 25 0 0 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 Maternal mortality ratio Under-5 mortality rate Infant mortality rate Source: WHO 2016. Note: IMR = infant mortality rate; MMR = maternal mortality ratio; U5MR = under-5 mortality rate Despite substantial improvements, inequalities re- second, and third quintile groups.25 However, the main, with the poorest economic quintiles having 2012 Demographic and Health Survey (DHS) reveals worse health outcomes than the wealthier quin- substantial inequalities in health outcomes, with the tiles. For the years 2005–06 and 2012, major gains lowest quintiles faring worse. In 2012, 17 percent of in child mortality were achieved among the lowest, those in the lowest wealth quintile were underweight, 25 Surprisingly, the 2012 Demographic and Health Survey shows that child mortality increased in the fourth and highest wealth quintile groups. This may be explained by the 2010 earthquake, which affected the metropolitan area, where households are relatively richer than those in the rest of the country. BETTER SPENDING, BETTER CARE: 28 A LOOK AT HAITI’S HEALTH FINANCING Trends of Key Health Preventive and Treatment Service Indicators, by Coverage Rate: Haiti FIGURE 3.2: Demographic and Health Survey (DHS), 1994–2012 30 33 Immunization 41 45 31 Diarrhea 41 treatment 44 58 Institutional 16 17 delivery 22 36 Skilled birth 21 24 attendance 26 37 0 10 20 30 40 50 60 70 Percent of population coverage for each service or treatment 1994-5 2000 2005-6 2012 Sources: DHS1994–95, 2000, 2005–06, 2012. and 31 percent were stunted, compared with 3 per- YLLs in 1990, it ballooned to 74 percent in 2010, and cent and 6 percent, respectively, of those in the high- then declined again to 10 percent in 2013 (figure 3.3). est wealth quintile (DHS 2012). Furthermore, the num- Access to key health preventive and treatment ser- ber of deaths from cholera was much higher among vices has improved in Haiti over the last two decades, households in the poorer wealth quintile than among and other than the anomaly distribution following the households in the highest wealth quintile. Of those 2010 earthquake, the proportion of YLLs attributable households in the poorest quintile, 2.4 percent had a to communicable diseases has declined accordingly, member who died from cholera, but only 0.1 percent while the percentage of YLLs attributed to noncom- of those in the richest wealth quintile reported such municable diseases is increasing dramatically. an outcome (DHS 2012). Thus a household member in the lowest wealth quintile was 24 times more like- Compared with other low-income countries (LICs), ly to die from cholera than one in the highest wealth Haiti still has much to achieve on several univer- quintile. sal health coverage indicators related to child and maternal health as well as water and sanitation. Coverage of key health services has increased over Three key maternal health indicators are important in the last two decades, and the burden of disease is monitoring the progress toward alleviating maternal shifting from communicable to noncommunicable mortality: (1) the percentage of unmet needs for fam- diseases. Between 1994–95 and 2012, deliveries by ily planning, (2) the percentage of pregnant women skilled birth attendants increased by 76 percent, deliv- receiving all four recommended antenatal care (ANC) eries in a health facility (also called institutional deliver- visits, and (3) the percentage of pregnant women un- ies) by 125 percent, treatment of diarrhea by 87 per- dergoing institutional delivery or with skilled birth at- cent, and immunization coverage by 50 percent (figure tendants. Haiti performs weakly on all three indicators: 3.2). The proportion of years of life lost (YLLs) attribut- 67 percent of pregnant women in Haiti receive four able to communicable diseases was still dominant in ANC visits, compared with 48 percent in LICs and 90 2013, but from 1990 to 2013 it decreased from 75 percent in the Latin America and the Caribbean (LAC) percent to 56 percent (figure 3.3). By contrast, the pro- region; 35 percent of women 15–49 report unmet portion of YLLs attributable to noncommunicable dis- needs for family planning in Haiti, compared with 22 eases (NCDs) increased from 19 percent to 34 percent percent in LICs and 10.7 percent in the LAC region; between 1990 and 2013. However, the 2010 burden and only 37 percent of pregnant women in Haiti have of disease in terms of YLLs exhibited a dramatic shift institutional deliveries, compared with 70.5 percent in toward causes associated with injuries from the earth- low- and middle-income countries and more than 75 quake; whereas this measure represented 6 percent of percent in rural areas and 90 percent in urban areas in Chapter 3 • HEALTH OUTCOMES AND THE HEALTH SYSTEM 29 FIGURE 3.3: Attributable Years of Life Lost (YLLs), by Cause: Haiti, 1990, 2000, 2010, 2013 a. 1990 b. 2000 6% 8% 19% 22% 75% 70% c. 2010 d. 2013 9% 10% 17% 34% 74% 56% Communicable diseases Noncommunicable diseases Injuries Source: IHME 2015. the LAC region (WHO 2015; Joseph et al. 2016; UNFPA 18 percent of deaths in children under 5 still are from 2016). Furthermore, only 68 percent of children un- diarrheal diseases, which leaves significant room for im- der 24 months of age in Haiti receive all three diphthe- provement. All these indicators would improve great- ria, tetanus, and pertussis (DTP) vaccine doses, com- ly with strong primary health care interventions. Thus pared with 90 percent in the LAC region (WHO 2015). these indicators support the finding that inadequate Pertussis, which could easily be prevented by DTP vacci- resources are allocated to preventive health services. nation, still causes 3 percent of under-5 deaths in Haiti (WHO 2013). Furthermore, across Haiti only 62 percent Health inequalities persist in the coverage of pre- of people use improved drinking water sources, and ventive and treatment services. As table 3.2 shows, 24 percent use improved sanitation practices. Water, the distribution of fully immunized children ages 12– sanitation, and hygiene (WASH) remains fifth in the 23 months by wealth index quintile improved between 2013 global burden of disease (GBD) ranking of top 2005–06 and 2012 but inequalities still persist for oth- risk factors for disability-adjusted life years (DALYs) in er services. In 2012 about 52 percent of children with Haiti. Relative to the LICs and countries in the LAC re- acute respiratory infections (ARIs) in the highest wealth gion, Haiti performs poorly on WASH indicators, which quintile received treatment versus 23 percent of those is a concern because of the country’s cholera epidemic in the lowest wealth quintile. Furthermore, institutional (World Bank 2015g). Of the children under 5 with di- deliveries were eight times more frequent (76 percent) arrhea in Haiti, 58 percent receive treatment, which is in the highest wealth quintile than in the lowest quin- slightly higher than the LIC average (50 percent) and tile (9 percent). The disparity in utilization mirrors the just below the LAC region’s average (59 percent). Yet inequality in health outcomes described earlier. BETTER SPENDING, BETTER CARE: 30 A LOOK AT HAITI’S HEALTH FINANCING Maternal and Child Health Coverage Rates, by Wealth Index: Haiti Demographic and Health TABLE 3.2: Survey (DHS), 2005–06, 2012 Q1 Q2 Q3 Q4 Q5 Total DHS 2005–06 Immunization 34 40 45 37 56 41 ARI treatment 27 31 41 40 40 35 Diarrhea treatment 34 38 47 54 54 44 Skilled birth — — — — — 54 Skilled attendant — — — — — 26 Delivery in health facility 5 8 17 35 58 22 DHS 2012 Q1 Q2 Q3 Q4 Q5 Total Immunization 43 46 52 42 41 45 ARI 23 32 36 52 52 38 Children with diarrhea who received ORT (or ORS at home) 57 52 59 61 62 58 Mothers who received 4+ ANC visits from any provider — — — — — 67 Births assisted by a provider skilled in obstetrics — — — — — 37 Births delivered in a health facility 9 20 38 51 76 36 Sources: DHS 2005–06 and 2012. Note: Immunization refers to fully immunized children ages 12–23 months. – = not available; ANC = antenatal care; ARI = acute respiratory infection; ORS = oral rehydration solution; ORT = oral rehydration therapy. Key Aspects of Haiti’s Health System hospitals, and referral/teaching hospitals. However, the Organization referral system is not functional (only 6 percent of re- ferrals are carried out properly), and thus the theoreti- The Ministry of Public Health and Population cal roles ascribed to each level in this pyramidal struc- (Ministère de la Santé Publique et de la Population, ture are not fulfilled. MSPP) is struggling to coordinate and oversee health service delivery, and the referral system is functioning poorly. The MSPP’s oversight role of the Efforts to Deliver an Essential entire health system needs to be strengthened. The Package of Health Services MSPP owns and runs 38 percent of health facilities, and another 20 percent are owned and managed jointly by the MSPP and nongovernmental organizations (NGOs), Recurrent cycles of natural catastrophes and po- for a total of 58 percent (figure 3.4). Sixty-one percent litical instability have shifted the Haitian govern- of all health facilities are located in rural areas, and, of ment’s focus toward emergency responses and these, most are owned by the MSPP. Both NGO and away from key structural health reforms. Because private for-profit facilities are overrepresented in the 96 percent of Haiti’s population remains at risk of ex- metropolitan area (figure 3.5). However, 50 percent of posure to two or more hazards (World Bank 2014) private, for-profit facilities and 55 percent of NGO facil- and because natural disasters occur frequently in Haiti, ities are located in rural areas. Overall, private facilities most of the attention and financing of the MSPP and are not accountable for the services they provide, and development partners are being diverted from follow- they do not systematically submit reports to the MSPP. ing up with structural health system reforms toward The formal health service delivery system has three firefighting. Another key impediment to health sys- levels: health centers and dispensaries, departmental tem reforms is political instability. Since 1986, Haiti has Chapter 3 • HEALTH OUTCOMES AND THE HEALTH SYSTEM 31 FIGURE 3.4: Health Facilities, by Ownership: Haiti, primary care level: community referral hospitals (hôpi- 2013 taux communautaire de référence, HCRs), health cen- ters with bed (centres de santé avec lit, CALs), health centers without bed (centres de santé sans lit, CSLs), and dispensaries. However, establishment of the UAS has been slow, and in practice the organization of the Private service delivery system has not changed significantly in 24% recent years. MSPP 38% A key problem in improving health care delivery is the absence of a prioritized and costed package Mixed of health services. In 2015 the MSPP commissioned 20% a task force to update the 2006 minimum package of ONG 18% health services to a package of essential services (PES). However, the services offered in the PES are not priori- tized and include most services already provided by the Source: SPA 2013 system (and even others that are not currently provid- Note: MSPP = Ministère de la Santé Publique et de la Population (Ministry of Public Health and Population); NGO = nongovernmental orgtanization ed such as fertility treatment). In addition, it has not been costed. Therefore, the PES is not useful because Haiti cannot realistically provide all health services to faced a succession of short-lived government adminis- all its citizens, at least not in the short run, given bud- trations, and a transition government has been in place getary constraints. The PES, then, does not offer much since February 2016. The newly elected incoming gov- guidance to service providers in terms of what priority ernment has just taken office in February 2017. services should be delivered where. The PES also does not include updated requirements for equipment and Meanwhile, reforms of the organization of the standards for each level of care, which will make its im- health delivery system have been enacted in re- plementation difficult. cent years, but they have yet to be implemented. In 2012 the National Health Policy stipulated the cre- Another constraint is the poor quality of care, ation new health units to improve the service delivery which is considerably worse in the area of preven- system. District health units (unités d’arrondissement tive clinical care. Only 62 percent of pregnant wom- de santé, UAS) were to be established to oversee the en receive physical examinations that meet minimum FIGURE 3.5: Health Facilities, by Ownership and Location: Haiti, 2013 100 90 80 55 50 70 65 61 69 60 Percent 50 14 40 14 30 20 23 20 24 36 31 10 19 7 11 0 MSPP ONG Privée Mixte Nationale Rural Urban Metropolitan area Source: SPA 2013. Note: MSPP = Ministère de la Santé Publique et de la Population (Ministry of Public Health and Population); NGO = nongovernmental orgtanization. BETTER SPENDING, BETTER CARE: 32 A LOOK AT HAITI’S HEALTH FINANCING standards of care, which should include uterine and with the necessary medicines (13 percent), equipment fetal height measurements. Furthermore, 3 out of 10 (54 percent), and infrastructure (7 percent).28 For the health providers fail to inquire about pregnancy risk infrastructure and drug dimensions, service readiness factors during patient interactions. Only 20 percent of increases at higher levels of care. At the dispensary medical consultations with pregnant women incorpo- level, patient volumes are extremely low; the average rate preventive care services such as counseling on the patient flow rate through dispensaries is about one minimum number of ANC visits or dispensing essential per hour or eight patients a day (World Bank 2015a). nutritional interventions such as folic acid supplemen- Based on these findings, dispensaries with a full staff tation (SPA 2014). The recent Balanced Score Card team are likely, for lack of essential inputs, to fail to de- assessment26 of health facilities in Haiti found that liver basic health services to the patients who seek care health facilities overall score low on internal manage- there. Because dispensaries represent 4 in 10 health ment processes (World Bank 2015a). These indicators facilities (n=359), they must be prioritized as Haiti ad- measure how supervisors or managers conduct differ- dresses service readiness in the short term. ent processes within the health facility. Low scores on facility management processes may explain the low Haiti has far less basic infrastructure and equip- readiness of staff to deliver care along clinical guide- ment than other low-income countries. The propor- lines. Many health facilities (46 percent) operate with- tion of health facilities in Haiti that scores satisfactorily out any data collection system, which makes moni- across internationally benchmarked standards is very toring and evaluation, as well as quality supervision, low both in absolute terms and against international challenging. standards. Analysis of service provision data collected in Haiti, Kenya, and Uganda support this finding. It also highlights uneven levels of service readiness in Haiti’s Service Readiness health facilities across the individual core service readi- ness indicators. The availability of basic infrastructure– Dispensaries, which are essential for the provision electricity, water, and toilets–is far lower in Haiti (31 of primary care in Haiti, score very poorly on sev- percent) than in Kenya (86 percent) and Uganda (64 eral key service readiness indicators. The Balanced percent) for all facility types for which data are avail- Score Card report assessed readiness to deliver essen- able (dispensaries through HCRs). Likewise, the avail- tial health services across four dimensions: minimum ability of the minimum equipment in Haiti (49 percent) personnel, basic infrastructure, basic equipment, and is about half of that observed in Kenya (77 percent) drugs (World Bank 2015a).27 A key finding of this as- and Uganda (82 percent).29 By contrast, 73 percent of sessment is that dispensaries suffer from a severe lack health facilities in Haiti score satisfactorily in the service of service readiness; only 2 percent meet the definition readiness assessment of personnel staffing (IHE and of full service readiness. More broadly, 6 percent (n=54) ICF International 2014). Pinpointing the cause of these of all health facilities (n=907) in Haiti are fully ready to performance weaknesses will be an essential step in provide minimum essential services. Dispensaries are the process of improving the technical efficiency of less likely than other facility types to be fully equipped Haiti’s health facilities. 26 This report uses a so-called Balanced Score Card (BSC) to assess the readiness of the inputs of service delivery. For comparison purposes, it can be applied to different countries. 27 The minimum personnel definition was based on Haiti’s essential package of health services and varies by facility type. Aligned with the World Bank’s Service Delivery Indicators (SDI) methodology (World Bank 2014), infrastructure was considered adequate if the facility had access to electricity, an improved water source, and an improved bathroom. Consistent with the SDI, facilities met the basic equipment requirement if they had a stethoscope, thermometer, sphygmo- manometer, and weighing scale. Finally, facilities were considered to have basic access to essential drugs if at the time of the survey they dispensed at least half of the 14 medicines in the Service Availability and Readiness Assessment (SARA) list of the World Health Organization (WHO 2010b). 28 Unexpectedly, 98 percent of dispensaries have the minimum requirement of personnel, far more than any other facility type. This finding likely stems from the various definitions of “minimum personnel.” For example, whereas one auxiliary/nurse is considered the minimum sufficient personnel for dispensaries, the rest of facilities require at least one medical doctor, who are scarce in Haiti. 29 World Bank staff estimates, 2016, based on the Service Provision Assessment (SPA) 2013–14 data set. Chapter 3 • HEALTH OUTCOMES AND THE HEALTH SYSTEM 33 CHAPTER 4 HEALTH FINANCING Resource Mobilization Haiti’s health sector enjoys a high level of resourc- es compared with those in other low-income countries (LICs), but it is still much lower than that in the Latin America and the Caribbean (LAC) re- gion. In 2014 Haiti’s total health expenditure (THE) per capita was $131 in international dollars,30 whereas in low-income countries and the LAC region the averag- es were $91 and $1,112, respectively. For the same year, if measured in constant dollars (2010), Haiti was spending $61 per capita on health, whereas the LICs were spending on average $37 (figure 4.1) and the LAC countries $714. When measuring THE as a per- centage of the gross domestic product (GDP), Haiti has a higher value than the average for the LICs. In 2014 THE in Haiti represented 7.6 percent of GDP, which is above the average of 5.7 percent for the LICs and comparable with the 7.2 percent average for the LAC PHOTO CREDIT : WORLD BANK PHOTOCOLLECTION countries. However, even if Haiti has a relatively high level of THE (as a share of GDP) compared with that of other LICs, there has been a downward trend since the post-earthquake period in 2011 when THE (as a share of GDP) reached 10.4 percent (figure 4.2). This re- veals the high volatility of the total health expenditure 30 Constant 2011 prices, purchasing power parity (PPP)–adjusted. BETTER SPENDING, BETTER CARE: 34 A LOOK AT HAITI’S HEALTH FINANCING Chapter 4 • HEALTH FINANCING 35 FIGURE 4.1: Total Health Expenditure (THE) per Capita: Haiti and LICs, 1995–2014 125 THE (2010 constant US$) 100 75 50 25 0 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 Haiti Low-income country Source: GHED 2016. Note: LAC = Latin America and the Caribbean; LICs = low-income countries. FIGURE 4.2: Total Health Expenditure (THE) as Percentage of GDP: Haiti, LICs, and LAC Region, 1995–2014 12 10 8 THE (% GDP) 6 4 2 0 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 Haiti Low-income country Latin America and the Caribbean Source: Data retrieved from the 2016 WDI database and GHED. Note: LAC = Latin America and the Caribbean; LICs = low-income countries. following the country’s economic fluctuations and nat- THE), followed by the government (41 percent of THE) ural catastrophes. and then NGOs (14 percent of THE)–see figure 4.3. Between 1995 and 2014, the proportion contributed Considering a longer-term perspective, the total by government decreased substantially, from 41 per- health expenditure has increased over the last 20 cent to 21 percent. During the same period, the share years, driven by external financing to nongovern- of out-of-pocket expenditures also decreased, from 46 mental organizations (NGOs) while the government percent to 35 percent, while the share of NGO financ- has played an increasingly marginal role in financ- ing increased dramatically, from 14 percent to 45 per- ing the sector. The majority of the external financ- cent, mainly because of the increase in external fund- ing has been channeled through nonprofit institutions ing. NGO participation began expanding in 2004 after serving households such as NGOs. Thus the increase in the Aristide coup, reaching 35 percent in 2006. It then external financing has changed the structural compo- went up again, reaching 62 percent of THE in 2012 af- sition of the total health expenditure. In 1995 house- ter the earthquake, and it has decreased substantially holds were the main financiers of the health system in recent years. Concurrently, the decline in the general through out-of-pocket (OOP) payments (46 percent of government health expenditure (GGHE) accelerated: it BETTER SPENDING, BETTER CARE: 36 A LOOK AT HAITI’S HEALTH FINANCING FIGURE 4.3: Trend in Share of Total Health Expenditure, by Finance Source: Haiti, 1995–2014 70 60 50 40 Percent 30 20 10 0 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 General government expenditure on health Out-of-Pocket expenditure Nonprofit institutions serving households (e.g. NGOs) Private insurance Source: GHED 2016. Note: In the National Health Accounts (NHA) methodology followed by the World Health Organization, the total health expenditure is composed of the government health expenditure and private expenditure. The latter includes out-of-pocket expenditures, nonprofit institutions serving households (such as nongovernmental organizations, NGOs), and private insurance. Thus the sum of the four variables shown in figure 4.3 is the total health expenditure. represented 36 percent of THE in 2004 and fell to 21 If used efficiently, Haiti’s current THE per capita percent in 2015. In 2014 OOP payments in Haiti were would be sufficient to provide a basic package of similar to those in other LICs. On average, however, health services. A costing exercise utilizing the mini- these payments have been steadily declining in LICs, mum package of health services developed in 2006 by whereas in Haiti OOP payments decreased broadly fol- the Ministry of Public Health and Population (Ministère lowing the influx of external funding but have been de la Santé Publique et de la Population, MSPP)31 rising since 2012. This increase stems mainly from the finds that the actual unit cost for delivering the pack- reduction in external funding to NGOs, together with age32 is $5 per capita, and the normative cost33 is the fact that the government did not increase resourc- $9.50 per capita. Financial health records capture both es enough to make up for the fall in external aid. the reported provision and expenditure by health facili- ties, as well as the costs associated with drugs donated The growth rate of the total health expenditure through vertical programs. If $9.50 per capita is need- has been very volatile, and the lack of predictability ed to provide a basic package of health services at the poses a challenge to long-term planning to reach first level of care, the total amount would represent health sector goals. The annual real growth rate of 8 percent of THE. Similarly, global estimates from the THE per capita greatly fluctuated from 2003 to 2013, World Health Organization (WHO) indicate that $50 with peaks in 2004, 2006, and 2010 corresponding per capita would ensure access to an essential health to political crises or natural catastrophic events. From care package (Cavagnero et al. 2008; WHO 2001).34 2003 to 2013, the average annual real growth rate of Even if these higher global estimates are used, Haiti the health expenditure in Haiti was 16 percent, vary- has sufficient resources to finance a basic package of ing from –27 percent to 53 percent. The high volatil- health services if all sources of funding (public, private, ity in THE (mainly due to the volatility of donor fund- and external) were pooled and used efficiently for this ing) makes it difficult to create a long-term investment purpose. plan. 31 This costing would have to be updated using the recently developed package of essential services (PES). 32 The actual unit cost reflects what health facilities spend today to treat patients at the first level of care. 33 The normative cost is the cost of the minimum package of services needed for all people to receive access to treatment and prevention. The estimate would be based on prevalence rates. For example, if the prevalence rate of malaria is 3 percent, it is assumed that 3 percent of the total population would be treated for malaria. 34 WHO estimated $34 in 2001 prices, which would correspond to $50 in 2014 prices, calculated using a methodology similar to that used by Cavagnero et al. (2008). Chapter 4 • HEALTH FINANCING 37 Public Financing Government Health Expenditure as FIGURE 4.4: Share of General Government Expenditure: Haiti, Compared with other LICs, Haiti has very low lev- LICs, and LAC Region, 2014 els of public financing as a share of the total health 16 expenditure. As shown previously, public funding as 14 13 a share of THE is the smallest source of financing for the health sector apart from private insurance, which 12 is negligible in Haiti. In the 1990s, the share of public 10 10 funding was higher, but it has been steadily decreas- Percent ing. In 2014 the total public health expenditure was 8 about 6,000 million Haitian gourdes (HTG)35 (counting 6 6 operational budget, public treasury, and on-budget ex- ternal financing), or about 1 percent of GDP. The gen- 4 eral government health expenditure (GGHE) as a per- 2 centage of GDP has been hovering at 1–2 percent and 0 is currently below that of the average LIC. Because the Haiti Low-Income Latin American and GGHE includes both public financing and on-budget Countries Caribbean Region external financing, it is possible that part of the de- Source: GHED 2016. Note: GGE = general government expenditure; GGHE = general government crease in the GGHE as a percentage of THE is driven by health expenditure; LAC = Latin America and the Caribbean; LICs = low-income a decline in on-budget external financing. In 2014 Haiti countries. spent 6 percent of its total government expenditure on Per Capita Government Health FIGURE 4.5: health–that is, about half the spending of the average Expenditure: Haiti, LICs, and LAC Region, 2014 LIC (figure 4.4). Thus there is scope for the government 400 to increase public financing and double this ratio to 336 350 reach the average level for a LIC. Furthermore, in 2014 Haiti’s per capita GGHE was $13, which is lower than Per capita Expenditure 300 the LIC average of $15 (figure 4.5). (current US$) 250 The share of the budget allocated to the health 200 sector has been decreasing over time. In the past, 150 Haiti’s health sector received domestic allocations of between 9 and 14 percent of the national budget. 100 Between 2000 and 2005, the government health ex- 50 13 15 penditure as a share of the general government bud- get was 14 percent on average. This was similar to 0 Haiti Low-Income Latin American and the average for the LAC region, which has been rel- Countries Caribbean Region atively stable between 12 and 13 percent (figure 4.6). Source: GHED 2016. Note: GGHE = General government health expenditure; LAC = Latin America However, during 2006–10 the same indicator dropped and the Caribbean; LICs = low-income countries. to 9 percent in Haiti. Primarily because of donor fund- ing displacement in the post-earthquake period, na- fall; some sources indicate that the share of total gov- tional budget allocations to health in 2012 were fur- ernment expenditure going to health in the 2016-17 ther reduced to 3.4 percent. In Haiti, the government budget is just 4.5 percent. expenditure on health represented just 6.1 percent of the total government expenditure in 2014, well below The majority of domestic financing is allocated to the Abuja Declaration recommended allocation of 15 salaries. The public health budget is composed of the percent. In addition, this percentage has continued to operating budget and on-budget external funding, 35 This figure was annualized to calendar year 2014 because a fiscal year extends from October to September in Haiti. From October 2013 to September 2014, the operational budget, public treasury, and external funding on-budget amounted to $3,059 million, $612 million, and $3,418 million, respectively, and from October 2014 to September 2015 the operational budget, public treasury, and external funding on-budget amounted $3,321 million, $534 million, and $1,760 million, respectively (Le Moniteur 2014, 2015). BETTER SPENDING, BETTER CARE: 38 A LOOK AT HAITI’S HEALTH FINANCING Government Health Expenditure as Share of General Government Expenditure: Haiti and LAC FIGURE 4.6: Region, 2000–2014 18 16.0 16.6 16 14.7 13.4 13.1 14 12.8 12.1 12.4 13.1 12.9 12.8 13.2 12.0 11.9 12.2 12.5 12 12.3 11.8 Percent 10 11.7 11.8 12.6 8 9.2 9.5 6.2 6.1 6.1 8.2 5.5 5.5 6 3.4 4 2 0 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 Haiti Latin American and Caribbean Region Source: GHED 2016. Note: LAC = Latin America and the Caribbean. both executed by the MSPP, and the public invest- are critical elements of the health system, are very low ment budget (public treasury), which is executed by in Haiti. the Ministry of Planning and External Cooperation (Ministère du Plan et de la Coopération Extérieure, For the last several years, key items such as vac- MPCE). The government budget is composed primar- cines have been fully financed by external donors ily of the operating budget, which amounted to HTG without any cofinancing by the government unlike 3,059 million, or $68 million,36 in fiscal 2014-15, and in most other LICs. From a strategic standpoint, this represented on average 84 percent of the public bud- needs to change; otherwise, some of these donors are get over the last five years. From fiscal 2010-11 to fiscal likely to stop (or substantially reduce) their financing 2013-14, the operating budget increased by 50 per- for vaccines in Haiti. For major vaccine financiers such cent, from HTG 2,040 million to HTG 3,059 million, as the Global Alliance for Vaccines and Immunizations and the investment budget increased by 7 percent, or (GAVI), cofinancing of vaccines from the government’s only from HTG 570 million to HTG 612 million. The domestic budget is a requirement for the program to MSPP uses a large share of its operating budget for continue. The required cofinancing amounts are small– personnel costs, which represented about 90 percent indeed, much smaller than GAVI’s contributions–and on average between fiscal 2006/07 and fiscal 2013- most LICs comply with this requirement. In the case 14. Other low- and middle-income countries allocate of Haiti, exceptions have been made to allow the pro- a smaller share of their operating budgets (excluding gram to continue even without any cofinancing by the the investment budget from public funds) to human government, but it is unlikely that these exceptions will resources. Examples are Honduras (65 percent), Ghana continue much longer. Thus it is essential for the gov- (58 percent), Tanzania (58 percent), Uganda (53 per- ernment to begin providing the needed cofinancing cent), Burkina Faso (43 percent), and Benin (22 per- through a dedicated budget line in which the need- cent)37–see figure 4.7. Because of the high share of ed funds are actually made available for spending. In domestic resources allocated to the payroll, financing the past, such a line was at times introduced into the for nonsalary operating expenditures, including spend- budget, but ultimately with no funds actually provided ing on items such as vaccines and medicines, which under the budget line. Essentially, domestic financing provided in this manner would leverage a much larger 36 Using an exchange rate of HTG 45 per U.S. dollar. 37 These figures are World Bank calculations based on the public expenditure reviews (PERs) of Tanzania (2012), Ghana, (2010), Uganda (2006–07), Burkina Faso (2005), and Benin (2004). The PERs for some countries such as Honduras, Ghana, Burkina Faso, and Tanzania do not specify whether donor funding was included in the investment budget. Thus these comparisons should be interpreted with caution. Chapter 4 • HEALTH FINANCING 39 Human Resources Salary Payment as Share of Government Operating Budget: Haiti and Selected FIGURE 4.7: Countries, Various Years 100 80 60 Percent 91 40 65 58 56 53 20 43 31 0 Haiti Honduras Tanzania Ghana Uganda Burkina Faso Benin Source: Adapted from World Bank 2016a. Note: This excludes the domestic investment budget. amount of financing from GAVI (and other donors). There is no clear or enforceable guiding principle Similar arguments apply to other items that the gov- on the amount of public spending to be devot- ernment considers essential. ed to the health sector. The 2012 National Health Policy stipulates, using the 2001 Abuja Declaration An increasing share of the public health budget is that the government should allocate 15 percent of the allocated to the central level. Departmental health general government expenditure to the health sec- directorates (directions departementales sanitaires, tor. However, in practice there is no legal framework DDSs) captured 61 percent of the operational budget38 for public spending on the health sector. The bud- of the MSPP in 2006–07, but their budget fell consider- get ceiling for all sectors is determined by the assess- ably in 2013–14, mainly because resources were shift- ment of the Program of Public Investment (PIP). Each ed to the central directorates. In 2006–07, 17 percent year in October, the Ministry of Economy and Finance of the operational budget was allocated to the central (Ministère de l’Economie et des Finances, MEF), level, compared with 51 percent in 2013–14. To date, it Ministry of Planning and External Cooperation, and the is unclear what exactly led to this shift in budget alloca- Office of the Prime Minister (la Primature) discuss the tions. However, according to recent studies, the central structure of the budget. The outcome of this meeting government has limited stewardship capacity in health is a “framework letter” that sets indicative credit ceil- service delivery (Durham et al. 2015), and a greater pro- ings (both investment and operational) for each sector portion of funds are consumed by administrative and based on preliminary technical work. The prime min- stewardship costs when they are allocated at the cen- ister sends the framework letter to each sector. The tral level rather than departmentally (Josephson and MPCE then consolidates the PIP in February based on Vinyals 2012). Therefore, increased centralization of the framework letter and communicates it to the MEF, the MSPP’s operational budget poses a risk to funding which finalizes the overall budget for “budgetary con- for service delivery. University and specialized hospitals ferences” in which all line ministries participate. During also received a smaller portion of the operating bud- these conferences, the MEF reviews the ceilings of get between 2006–07 (22 percent) and 2013–14 (12 each sector before they are submitted to the Minister percent)–see figure 4.8. The budget for central direc- Council. After submission to the Minister Council, the torates, university hospitals, and DDSs is not allocated prime minister submits the budget to Parliament for based on need, population, and poverty factors, and, as a vote in May. A justification for additional funding is noted earlier, about 90 percent is used to pay salaries. guided by the priorities of the government in office. 38 No information is available on the distribution of the public treasury among central directorates, DDSs, and university/specialized hospitals. BETTER SPENDING, BETTER CARE: 40 A LOOK AT HAITI’S HEALTH FINANCING FIGURE 4.8: Percentage Distribution of MSPP Budget, by Directorate: Haiti, 2006–07 to 2013–14 100 90 80 37 46 Percent of total budget 70 58 54 53 53 54 61 60 50 40 51 30 22 27 29 30 40 17 29 20 10 22 20 18 17 16 16 14 12 0 2006-07 2007-08 2008-09 2009-10 2010-11 2011-12 2012-13 2013-14 Departmental directorate Central directorate University and department hospitals Source: World Bank 2016a. Note: The budget includes operating budget only. The public treasury is not broken down by directorates. MSPP = Ministère de la Santé Publique et de la Population (Ministry of Public Health and Population). Because external donors have increased their contri- an inverse relationship between the OOP expenditure bution to the health sector over the last decade, the and donor/NGO financing. Haitian government has had little incentive to increase the share of the budget allocated to the health sector. The out-of-pocket expenditure increased by 36 percent from 2011 to 2014, driven mainly by ex- penditures for medicines (82 percent) and consul- Private Financing tation/user fees (60 percent). Individuals who had to pay out-of-pocket spent on average HTG 664 in 2012. Private financing includes household out-of-pock- By 2013, this figure had increased by 55 percent, to et spending, nonprofit institutions serving house- HTG 1,032 (ECVMAS 2012, 2013)–for more details, holds (such as NGOs), and private insurance, which see chapter 5. This increase in OOP expenditures not includes direct service payments by private corpo- only deters and delays utilization of health services rations. The private health expenditure is the main when needed but also risks pushing people into or source of financing in Haiti, and it represented about deeper into poverty. OOP spending also constrains the 80 percent of THE in 2014. About 35 percent was the redistributive capacity of health financing systems to out-of-pocket expenditure, less than 1 percent was enable resources to be allocated based on need rath- private insurance, and the remaining 44 percent was er than on ability to pay. An analysis of the impact of from nonprofit institutions serving households. The lat- the OOP expenditure on the welfare of households ap- ter mainly represents financing from NGOs in Haiti be- pears in chapter 5 on access to health services. cause they play an important role in both the financing and provision of health services. The share of NGO fi- nancing increased from 14 percent of THE in 1995 to External Financing 44 percent in 2014. The out-of-pocket expenditure fell from about 50 percent in the early 2000s to 26 per- A large surge in external financing39 followed the cent in 2011. But in 2014, because of the drop in ex- 2010 earthquake in Haiti, but it is now declining ternal financing after the earthquake, the OOP expen- dramatically. External funding began to increase in diture went up to 35 percent. Thus there seems to be 2003–04, reaching its peak in 2011 in the aftermath 39 External financing is all the external aid available in a country. It can finance, among other things, on-budget government contributions (that is, it is part of the variable “general government health expenditure” in figure 4.3) and the private sector–for example, through nonprofit organizations such as NGOs (another variable in figure 4.3). Thus external financing is part of the public and private financing discussed in previous sections. However, because of the importance of external financing in Haiti, this subsection was deemed necessary. Note that because the vast majority of external aid is channeled through NGOs, the pattern found in the variable “external financing” in figure 4.9 is similar to the variable “nonprofit institutions serving households” (such as NGOs) in figure 4.3. Chapter 4 • HEALTH FINANCING 41 FIGURE 4.9: External Financing as Share of Total Health Expenditure: Haiti, LICs, and LAC Region, 2003–14 100 90 80 70 Percent of THE 60 50 40 30 20 10 0 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 Haiti Low-Income Countries Latin American and Caribbean Region Source: GHED 2016. Note: LAC = Latin America and the Caribbean; LICs = low-income countries; THE = total health expenditure. On- and Off-Budget External Financing for Health as Percentage of Total External Financing for TABLE 4.1: Health: Haiti, 2010–11 to 2014–15 HTG, millions 2010–11 2011–12 2012–13 2013–14 2014–15 Operating budget 2,040 2,209 2,591 3,059 3,321 Public treasury, PIP (on-budget) 570 372 532 612 434 External funding, PIP (on-budget) – – 8,947 3,418 1,760 External funding, MGAE (off- 47,110 117,016 234,884 185,241 9,060 budget) On-budget external funds for health, as % of total external – – 4% 2% 19% funds for health (off-budget) Off-budget external funds for health, as % of total external – – 96% 98% 81% funds for health Sources: Le Moniteur; BOOST 2015; MGAE database for health, https://haiti.ampsite.net. Note: – = not available; MGAE = Module Gestion de l’Aide Externe (External Assistance Management Module); PIP = Programme d’Investissement Public (Public Investment Program). of the earthquake, when it represented 69 percent of to 2014–15, the off-budget contributions have de- THE. From 2003 to 2014, external financing as a share clined by 25 times and the on-budget contributions of THE significantly increased in Haiti, much more so by five times (table 4.1), representing a massive loss than in LICs, but it also showed a more volatile pattern for the health system. Public treasury funds have also (figure 4.9). External financing has declined dramatical- decreased but at a slower pace, whereas the operat- ly since 2011, falling from 69 percent of THE in 2011 ing budget has increased slightly, but not enough to to 33 percent in 2014, which is lower than the LIC av- compensate for the sharp decrease in external fund- erage of 38 percent. ing. A large portion of external resources is currently used to finance operating costs such as vaccines, the The drop in external financing raises the issue of health workforce, and medical products (MSPP 2014). sustainability of investment programs. Both on-bud- The recurrent/capital expenditure ratio is slightly high- get and off-budget donor contributions have been fall- er for programs in the health sector (73 percent) com- ing since 2012–13. From the highest levels in 2012–13 pared with those in other sectors. With the withdrawal BETTER SPENDING, BETTER CARE: 42 A LOOK AT HAITI’S HEALTH FINANCING of external funding, the Haitian government needs to Share of External Financing for Health FIGURE 4.10: start paying these recurrent expenses to ensure the Committed to Disease-Specific Programs: Haiti, maintenance of capital investment and the functioning 2014 of the health system. Large financing gaps for recur- rent costs are currently emerging, and they are likely to continue to occur because external financing is de- creasing rapidly in Haiti. Faced with lack of a system for tracking donor resources and how they are used and with limited public financing, the government may not be able to plan and take over the costs of maintenance and operations for health facilities. External health funding is largely off-budget with limited oversight by the government. Neither the treasury nor the MEF monitor the expenditures fi- nanced by donors. In an effort to improve its monitor- ing, the Ministry of Planning and External Cooperation has started to aggregate information on the projects financed by donors through its External Assistance Management Module (Module de Gestion de l’Aide Externe, MGAE), which was set up in 2009 and has data from 2010. On average, 95 percent of external financing was off-budget40 over the last three years, al- Others 1% HIS 2% though that percentage trended downward, from 96 AIDS 7% Hospitals 16% percent to 81 percent from 2012–13 to 2014–15 (ta- ble 4.1).41 In the absence of a functioning donor coor- Ancillary functions 9% MCH 12% dination mechanism, it is difficult for the MSPP to ef- Basic health 5% Nutrition 7% fectively channel this external financing in accordance Cholera 4% Research 0% with the Health Master Plan (Plan Directeur de Santé), also because that plan has been neither costed nor Social services 6% Handicapped 1% prioritized. Training 2% Health systems strengthening 11% Vaccination 3% External financing needs to be more coordinated and aligned to Haiti’s global burden of disease. It Health care delivery 3% WSS 11% is challenging to track how external financing is used Source: Estimates based on MGAE database, 2016. Note: AIDS = acquired immune deficiency syndrome; HIS = health information in Haiti. Each donor has its own policy for disburse- system; MCH = maternal and child health; WSS = water supply and sanitation. ment, procurement, and audit systems, and many dis- burse directly to service providers or NGOs (that work rehabilitation captures the highest share of external directly with service providers), thereby bypassing the funding, 16 percent (one-fifth)–see figure 4.10. And MSPP and with high transaction costs. After compil- yet, the three leading causes of disability-adjusted life ing information on key donor programs in the health years–human immunodeficiency virus (HIV), acute re- sector, this study found that external financing could spiratory infections, and diarrhea–could be addressed be better aligned to Haiti’s burden of disease. Hospital by preventive and primary health care interventions 40 Off-budget project support includes funding for joint projects with the private sector and NGOs, as well as some projects implemented by donors and multilat- eral organizations. As the MEF is not involved in the administration of this kind of support, it is excluded from any planning or collaborative discussion, which makes it easier for donors to control selection of the type of projects to be funded and the timetable for implementation. Financiers choose this approach to circumvent existing regulations on budget execution, to avoid macroeconomic budget ceilings, to conform to guidelines of donor countries, to strengthen civil society, or because of the belief that government agencies are inefficient or corrupt. Since this funding approach utilizes financial management systems outside the existing government structures, it may create parallel implementation units (Stierman et al. 2013) 41 This reduction is explained by the fact that the off-budget funding decreased by 25 percent, while the on-budget external funding decreased by 5 percent and the public treasury funding by 1.2 percent. Thus the on-budget funding represented a higher share of the total external funding for the health sector. Chapter 4 • HEALTH FINANCING 43 instead of services that are provided at the hospital workers by location), the Table Sectorielle has func- level. tioned as a mechanism for ad hoc and emergency re- quests. As a result, it has not led to joint decision mak- The MSPP should capitalize on and scale up the ing between the MSPP and donors. However, some existing examples of donor alignment around spe- positive experiences of donor alignment around spe- cific programs, so that Haiti can harmonize ex- cific programs have emerged. For example, the MSPP ternal sources of financing for health. At the mo- has led the development of a national manual for re- ment, the activities of external partners are not fully sults-based financing (RBF), which aligns key donors coordinated, which makes planning and implement- such as the World Bank, U.S. Agency for International ing sector-wide health programs difficult. The Table Development (USAID), and the Global Fund (as of Sectorielle, which convenes the minister of health and 2016) around a results-based purchasing mechanism donors to discuss health system challenges, does not for primary care. The same mechanisms and indicators meet regularly. Although there have been intermittent are used. Although the RBF project is a good example efforts to coordinate donor funding (such as the deci- of donor alignment, a weakness of this approach is sion to map donor contributions to community health that it is still 100 percent donor-financed. Pooling The current prepayment mechanisms are largely number enrolled and the target population. Voluntary insufficient in Haiti and only cover a limited share health insurance initiatives launched by Partners in of the population. An important aspect of universal Health (PIH) and the Development Activities and health care (UHC) is a move away from reliance on OOP Services for Health (Développment des Activités de payments as the primary source of financing to pooled Santé en Haïti, DASH) are the best-known commu- and prepaid sources (table 4.2). Pooling includes pre- nity health insurance mechanisms. With funding from paid insurance as well as government spending–a form the Inter-American Development Bank (IDB), DASH of- of implicit pooling where prepayment is in the form of fers a market-based, low-cost, prepaid health card that taxes. However, as discussed earlier, public spending gives 100,000 low-income Haitians limited access to is a relatively small source of financing, representing quality basic health care (IDB 2013). In 2013 DASH42 only 6 percent of the general government expenditure covered 40,000 inhabitants (Le Nouvelliste 2013) or (GGE). Prepaid insurance schemes are underdeveloped 0.04 percent of the population, but it aimed to cover and cover less 5 percent of the population. The majori- 100,000 or 1 percent of the population. The coverage ty of the population who is covered by health insurance rate of prepaid insurance schemes is similar to that in lives in the metropolitan area. The Office of Insurance many LICs and far below that of neighboring countries for Work Accidents, Illness, and Maternity (Office d’As- (see table 4.3 for a review of existing schemes). surance Accidents du Travail, Maladie et Maternité, OFATMA) covers about 5 percent of the population, OFATMA is the main health insurer in Haiti. It is and other community health financing mechanisms an autonomous public institution under the adminis- cover less than 1 percent. Little information is avail- trative supervision of the Ministry of Social Affairs.43 able on commercial insurance other than that nine pri- OFATMA’s mission is to manage insurance regimes vate voluntary insurance schemes exist in Haiti (Wright against health, accident, and maternity risks. Initially, 2015). However, there is no clear information on the OFATMA offered work accident insurance (since 1967) 42 In 2003 the Multilateral Investment Fund (MIF), a member of the Inter-American Development Bank Group, approved an $827,807 donation for a project with DASH (MIF 2013). In 2013 DASH enrolled 40,000 patients. IDB funds have been particularly helpful in marketing DASH and informing the population about the advantages of prepayment mechanisms. 43 OFATMA is also overseen by the entity National Coordination of Health Insurance (Coordination National de l’Assurance Maladie, CONAM) and by CAROSSE, which is the administration board. BETTER SPENDING, BETTER CARE: 44 A LOOK AT HAITI’S HEALTH FINANCING TABLE 4.2: Health Financing System Characteristics, Haiti General revenue– Voluntary health financed Private health Household out-of- OFATMA insurance scheme government insurance pocket payment (DASH) spending All the population About 5 percent of population Target (although roughly More than 93 (440,000 persons–that is, 88,000 Not known Not known but population 40 percent of the percent of the civil servants and their dependents), but assumed assumed to be <1 and share of population lacks population; those private sector employees and their to be <1 percent; typically, population access to essential not covered by dependents, as well as 2,000 persons percent informal sector covered health and nutrition other schemes from the informal sector services) Voluntary, Mandatory for civil servants; Mode of Open to the entire depending on voluntary for the private and informal Voluntary Voluntary participation population the willingness of sectors household to pay Nonrisk-related health insurance contribution Different packages of services at a cost of Two plans offering various services: Taxes (tax system is between HTG 68 ($1) one plan for employees earning less currently regressive and HTG 685 ($10) than HTG 25,000 and one plan for because the per month Voluntary: those earning more than HTG 25,000 Revenue country’s fiscal Avantage santé households’ Civil servants: 30 percent funded by Premiums (pilot): fee of HTG source revenues rely disposable income civil servants and 70 percent funded 1,500 for six months heavily on indirect and saving by government taxes) plus external financing Private sector: 3 percent levied from employee’s salary and 3 percent paid IDB grant covering by firm; same for informal sector marketing costs Informal sector Level of National but mainly covering the West and no No interpersonal pooling and National separate pooling between civil servants and private Scheme level pooling redistribution sector schemes HTG 50 when going to general Copay of HTG 75 to 93 percent of health Entirely cofinanced practitioner and HTG 100 when Varies across have access to free ge- Cofinancing facilities charge user (there is no going to specialist; 20 percent for schemes nerics and lab analysis fees prepayment) paramedical services (lab, drug, x-ray) (Advantage santé) Health insurance: PHC consultation, Benefit package is laboratory exam, drugs, hospital not implemented, stay, regular hospital visits, maternity expenses, dental care and prevention, Preventive care and so it varies by ambulance transportation, optical Varies across basic PHC services as Benefits region, depending n.a. care; preventive care not covered schemes well as surgery on availability and readiness. Implicit Occupational insurance: accident, loss rationing. of a member, funeral costs. Benefits vary by professional status. Referral system No referral Gatekeeping in theory but not Referral not required Not known Referral required required implemented Source: World Bank 2016a. Note: n.a. = not applicable; DASH = Développment des Activités de Santé en Haïti (Development Activities and Services for Health (DASH); HTG = Haitian gourde; IDB =Inter-American Development Bank; OFATMA = Office d’Assurance Accidents du Travail, Maladie et Maternité (Office of Insurance for Work Accidents, Illness, and Maternity); PHC = primary health care. Chapter 4 • HEALTH FINANCING 45 TABLE 4.3: Health Insurance Coverage in Haiti and Other Countries Population covered Financial protection (level of out-of- Health insurance or services Benefits package pocket expenditure subsidized by government Target as % of total health expenditure) Insured (%) Uninsured (%) Ghana Entire population 54 46 Comprehensive 27 Indonesia Entire population 63 37 Comprehensive 38 Rwanda Entire population 92 8 Comprehensive 22 Vietnam Entire population 42 58 Comprehensive 58 Civil servants, expanding to Haiti 5 95 Not comprehensive 35 formal private sector Inpatient (with pilot India People below the poverty line 8 92 61 outpatient) Inpatient (with pilot Kenya Formal sector 20 80 43 outpatient) Mali Entire population 3 97 Comprehensive 53 Civil servants, expanding to Nigeria 3 Comprehensive 59 informal sector Source: World Bank 2016b; adapted from Lagomarsino et al. 2012. for the private sector and expanded to maternity insur- employees’ salaries, and it is matched by another 3 ance in 1975, but the latter was only put into practice percent paid by employers. This tax covers employ- in 2014. This social insurance scheme is compulsory for ees and their dependents. Civil servants contribute all workers, according to a 1967 law. However, not all up to 30 percent of the premium, and the MEF con- firms have enrolled their employees. Today, OFATMA tributes up to 70 percent. Informal workers pay HTG covers up to 5 percent of the population and targets 100 per month for an individual membership only. four populations: (1) 88,000 civil servants and their This funding is supposed to be complemented by a dependents (about 440,000 persons in total, assum- public allocation of HTG 600 per person (OFATMA ing households include five members)–this scheme is 2015), but it has not yet been fixed (Lamaute-Brisson compulsory and was moved from the private sector to 2015). Although OFATMA manages four popula- OFATMA management in 2014; (2) employees from tions, it is organized in two pools. The civil servant private companies with an agreement–the accident scheme is one pool, and the pool regrouping contri- insurance is compulsory, but not all firms enroll their butions from the formal and informal sectors is man- employees and the number of employees benefiting aged separately. Because there is no cross-pooling from accident and health insurance is not known; (3) between these two funds, risk equalization arrange- employees from the informal sector who are organized ments are limited. The financial viability of OFATMA- in organization/trade unions–this scheme is voluntary managed schemes is also questionable. For example, and covers only 400 persons working at the airport the premium for the civil servant schemes has not (red caps) and 100 women and their dependents;44 changed in 15 years, and OFATMA does not know and (4) employees from the formal sector (voluntary) exactly how many civil servants are in the program (OFATMA 2015). because this information has not been provided by the government. Thus OFATMA does know wheth- The financing of OFATMA is fragmented. A er the program is financially sustainable based only monthly tax of 3 percent is levied on private sector on the contributions from employees and the MEF. 44 This scheme, known as Kat wòz konbit solidarity or Carte rose, will not be extended because OFATMA did not receive the subsidy from the government for this purpose (2 percent of the government’s budget). BETTER SPENDING, BETTER CARE: 46 A LOOK AT HAITI’S HEALTH FINANCING Because OFATMA is perceived to be a management have to pay 20 percent of the total amount of para- entity more than an insurer, it does not have any medical services (lab, x-ray, drugs). mechanism to improve the civil servants’ scheme. Furthermore, the MEF is often delayed in reimburs- OFATMA’s health insurance excludes the unem- ing OFATMA, which poses solvency issues and delays ployed and is less inclusive for informal workers’ in payments to providers. This situation is affecting dependents. This health insurance is offered only to OFATMA’s image because providers perceive this de- those who are employed, which is problematic be- lay as being OFATMA’s responsibility. cause the unemployment rate in Haiti is 30.1 percent (Singh and Barton-Dock 2015). And yet the unem- OFATMA offers several benefits packages that ployed are the most likely to encounter catastrophic vary, based on the salary of the insured. There are health expenditures (CHEs). OFATMA has experiment- two plans for the private sector: one for employees ed with contracting associations of informal workers. earning HTG 25,000 or less a month and another for However, it is questionable whether this program could employees earning more than HTG 25,000 a month. be scaled up, and there are issues with how informal The three plans for civil servants are based on the workers’ dependents are included in the program.45 salary of the insured. Overall, both the formal and informal sectors would be entitled to the following Building capacity in effectively managing existing services: primary health care consultation, laborato- insurance schemes would be essential for future ry exam, drugs, hospital stay, regular hospital visits, expansion of coverage. In 2013 OFATMA was able maternity expenses, dental care and prevention, am- to reduce the time required to reimburse the provid- bulance transportation, and optical care (OFATMA er from six months to 45 days thanks to a firm that 2015). Preventive care is not covered, however. set up software to manage insurance claims. However, Those earning less than HTG 25,000 have lower re- there are no in-house insurance and actuarial skills imbursement than those earning more. Patients with that would allow OFATMA to estimate the cost of the OFATMA insurance are offered services in OFATMA health services it covers and the per capita premium facilities, which comprise three hospitals and a net- based on the number of enrollees. In 2013 OFATMA work of 37 health facilities. To avoid supplier-induced relied on a firm to conduct such an analysis, but ideally demand, there is a copay of HTG 50 per outpatient such work would be done internally or in partnership consultation and HTG 100 for a visit to a specialist. with an organization that could strengthen OFATMA’s Civil servants and employees from the private sector technical capacities. Purchasing Strategic purchasing is limited in Haiti because of departments conducted by the World Bank in 2013 the fragmentation of health financing flows and found that facilities typically rely on three sources of the service delivery system. All health facilities, re- funding: various donors (main source), MSPP/public fi- gardless of their ownership status, report to depart- nancing, and internal revenue/user fees (figure 4.11). mental health directorates. Most facilities receive in- Because of the complex ownership structures and the put-based financing from a number of sources. A small fragmented financial flows to facilities, strategic pur- study of 45 primary health care (PHC) facilities in three chasing is limited in practice. 45 The children of workers in the informal sector and those of workers in the formal economy do not have equal access to health insurance. Civil servants and workers in the formal economy are entitled to cover four dependents for free. By contrast, workers in the informal sector have to pay a monthly fee of HTG 55 for each additional child. Because most workers in the informal sector are poor (they live on $2.25 or less per day) and have on average five dependents (ECVMAS 2012), paying HTG 100 per month plus HTG 275 in total (HTG 55 per dependent) would push them to spend 10 percent of their salary on health care per month, which is far above the 3 percent tax levied on salaried workers. Chapter 4 • HEALTH FINANCING 47 FIGURE 4.11: Funding (HTG) of Primary Health Care Facilities, by Source: Haiti, 2013 100 80 60 du financement total 60 51 52 40 33 34 33 34 30 Percent 24 24 20 14 10 0 HCRs CALs CSLs Dispensaries Percentage internal revenue Percentage MSPP Percentage donors Sources: World Bank, USAID, and MSPP 2013. Note: CALs = centres de santé avec lit (health centers with bed); CSLs = centres de santé sans lit (health centers without bed); HCRs = hôpitaux communautaire de référence (community referral hospitals); HTG = Haitian gourde; MSPP = Ministère de la Santé Publique et de la Population (Ministry of Public Health and Population). Because there is no prioritized benefits package, between purchasers and providers. MSPP facilities each provider delivers whatever services it deems do not receive budget allocations for their use. Instead, necessary. In the absence of strategic purchasing, NGO the MSPP pays the salaries of health staff at the depart- and private providers do not necessarily deliver services mental level via the departmental finance directorates. aligned with the minimum package of health services For the rest, MSPP facilities have to make requests to defined in 2006 by the MSPP. In theory, MSPP facilities their departmental health directorate to pay nonsalary are supposed to implement this package, but no one expenses such as for utilities, medicines, and supplies. monitors compliance. As noted earlier, the MSPP re- MSPP facilities usually rely on user fees to purchase cently developed a package of essential services (PES), medicines and medical supplies. but that package has not been prioritized or costed. Usually, an essential package of health services (EPHS) Because these facilities have no control over a bud- focuses on reproductive health, maternal and child get, they are limited in their ability to use resourc- health (MCH), noncommunicable diseases (NCDs), and es to improve services delivery. MSPP second-level mental health. In Haiti, the PES goes beyond this list facilities such as the HCRs and hospitals have more of basic services (WHO 2008). For example, the newly leverage because they receive funding for nonsalary proposed PES offers fertility and cancer treatments.46 expenses from the departmental finance authorities. In addition, the PES does not provide an overview of For NGO facilities, a line-item budget is the typical pro- the required input (especially drugs and equipment) by vider payment mechanism used across local and inter- level of care–dispensary, health center, community re- national NGOs in charge of delivering health services. ferral hospital (hôpital communautaire de reference, Traditionally, an NGO’s headquarters has managed the HCR). The PES thus differs from its predecessor, which financial aspects of health facilities and paid staff, as at least included a list of drugs and equipment by lev- well as managed disbursement for specific line items. el as well as the minimum number of staff by level. Health workers hired by NGOs receive higher salaries Because there is not enough funding in any health sys- than staff recruited by the MSPP, in particular medi- tems, including Haiti’s, to cover all services for all citi- cal doctors. In a study conducted in three departments zens, it is very unlikely that the PES can be implemented (Center, North-East, and North-West), medical doc- in practice unless it is prioritized and costed. tors recruited by NGOs made 58 percent more than medical doctors recruited by the MSPP. This incentive Health facilities have limited financial autono- may contribute to the higher productivity of medi- my, and there is little accountability for results cal staff in NGO facilities, as confirmed by the Service 46 The package of essential services was launched and made available to the team only in September 2016. BETTER SPENDING, BETTER CARE: 48 A LOOK AT HAITI’S HEALTH FINANCING Provision Assessment (SPA–Évaluation de la Prestation mechanisms. It worked with each NGO to set ser- des Services de Soins de Santé, EPSSS) data set (see the vice delivery targets and incentives based on the or- discussion of human resources in chapter 6). ganization’s historical performance and to agree on a budget. The MSH disbursed 95 percent of budgeted Strategic purchasing is also limited within OFATMA. funding to NGOs on a quarterly basis after receiving There is no accreditation system between OFATMA and the required information (such as data reports and an its contracted health facilities and hospitals. Ninety- action plan), and it retained 5 percent of budgeted three medical doctors are affiliated with OFATMA, as funding as incentives. Thus an NGO received the low- well as four laboratories, three pharmacies, and two est percentage (95 percent) of its budgeted funding if eyeglass manufacturers (OFATMA 2015). Health facili- it met none of its service targets and the highest per- ties and medical doctors receive a copayment from pa- centage (105 percent) if it achieved all the targeted tients at the point of service delivery. OFATMA then goals set by SDSH by the end of each year. An eval- reimburses the provider the difference. OFATMA pro- uation in 2009 found that NGO health facilities en- vides a financial incentive to medical doctors who re- rolled in the scheme performed better than those in ceive insured patients at an OFATMA hospital. the rest of Haiti in complete immunization coverage, prenatal care, assisted deliveries, and postnatal care The MSPP recently strengthened the tools that will (Eichler, Auxila, and Pollock 2001; Eichler and Levine encourage provider-purchaser responsiveness. It 2009). A more recent study confirmed those findings has developed several tools to control resources and and showed that the addition of performance-based the production of health services at the national level. incentives, training, and technical assistance for non- Each facility, regardless of type, has to submit a report governmental health facilities in Haiti increased key on the utilization of PHC services to the departmental services over a three-year period by 39 percent. For health directorates, and those reports are forwarded to children under 1 year and pregnant women, the in- the Planning and Evaluation Unit (Unité de Planification creases in services were both statistically significant et Evaluation, UPE) at the MSPP for a quality check and large in magnitude –1.7 to 2.2 times the baseline and monitoring. Using its new health management in- rates (Zeng et al. 2013). formation system, SISNU, the MSPP can better oversee the delivery of care. For example, the UPE will follow The MSPP has just begun to implement RBF in up with the departmental health directorates if data about 10 percent of PHC facilities and will begin are inconsistent or there is an alarmingly low utilization to pay providers based the coverage and quality of of key health services coverage indicators. care. With the implementation of RBF in March 2016, 80 PHC facilities (50 sponsored by the World Bank and Several USAID-funded NGOs have improved ser- 30 sponsored by USAID) are being paid through out- vice coverage by contracting providers based on put-based payments with the implementation of RBF. results-based financing (RBF). With USAID fund- The 80 PHC facilities include dispensaries, health cen- ing, the Management Sciences for Health (MSH) ters, and community referral hospitals. Fifty percent of program contracted out the provision of health ser- the funding will be allocated to improve the operating vices in all 27 NGOs through the project Health for and quality of health services, and up to 50 percent of the Development and Stability of Haiti (Santé pour total funding will be used to pay health staff based on le Développement et la Stabilité d’Haïti, SDSH) as a grid of indicators combining measures of both quan- of 2007. The MSH not only contracted out services, tity and quality of care (MSPP 2014)–for more details, but also implemented performance-based financing see chapter 6. Chapter 4 • HEALTH FINANCING 49 CHAPTER 5 ACCESS TO HEALTH SERVICES T he outpatient utilization rate in Haiti was low in 2013, at approximately 0.5 visits per per- son per year. The MSPP has estimated that there were 0.58 visits per person in 2014– 15 (MSPP 2016), confirming the relatively weak utilization. However, there are wide variations by department and only in the West, North, and North- East departments is utilization higher than the nation- al mean. Two departments, South-East and Artibonite, have very low utilization rates, with less than 0.4 visits per person (IHE and ICF International 2014). Affordability Affordability is one of the main causes of low out- patient utilization. In 2013, 24 percent of households reported not consulting a health provider when sick (ECVMAS 2013). Among those, 49 percent did not consult a provider because they could not afford care PHOTO CREDIT : LOGAN ABASSI / UN/MINUSTAH (figure 5.1). Low affordability has also been found in other studies in Haiti. For example, in one study the majority of the sample reported not using tradition- al birth assistance or a hospital for birthing because of the cost (Urrutia et al. 2012). The removal of user fees for maternal and child health services in several facilities in Grand’Anse led to a 200 percent increase in attendance as compared with cost sharing (Altaras 2009). BETTER SPENDING, BETTER CARE: 50 A LOOK AT HAITI’S HEALTH FINANCING Chapter 5 • ACCESS TO HEALTH SERVICES 51 TABLE 5.1: Health Behavior and Financial Barriers to Access to Health Care, by Wealth Quintile: Haiti, 2013 Did not seek care for Did not seek care because Reported health problems Wealth quintile reported health problems “too expensive” in last 30 days (%) (%) (%) Lowest 16 35 66 Second 13 25 46 Middle 20 28 53 Fourth 18 17 42 Highest 20 17 39 National 18 24 49 Source: World Bank estimates based on ECVMAS II 2013, World Bank, and ONPES 2014. Cited Barriers to Health Services Utilization FIGURE 5.1: wealth quintile did not seek medical attention be- as Share of Survey Respondents: Haiti, 2012 cause of lack of money, compared with 39 percent in the highest wealth quintile and 49 percent at the national level (table 5.1). Further analysis confirmed that wealthier households were more likely to consult a provider when sick (World Bank estimates based on ECVMAS 2013).47 Because the poor consult a health provider less often, they encounter a lower level of catastrophic health expenditures (CHEs)48 than the highest wealth quintile. Overall, the incidence of CHEs increased for all socioeconomic and geospatial categories from 2012 to 2013, except for the lowest quintile (figure 5.2). Indeed, from 2012 to 2013 the lowest wealth quintile saw a decrease in its out-of- pocket (OOP) payments, while all other wealth quin- tiles saw an increase in OOP payments, in particular the third and fourth wealth quintiles (table 5.2). The decrease in OOP in the lowest quintile could be theo- Not necessary 20% Not enough money 49% retically explained for two reasons. One reason could be that the poorest are exempted from paying and Too far 1% Bad quality 1% therefore more protected against CHEs. Another rea- son could be that the poorest are directly not using Automedication 22% Others 6% health services when needed since they know they Too busy 0% Cultural reason 1% cannot afford to pay. The latter seems to be explain- Sources: World Bank estimates based on ECVMAS, World Bank, and ONPES ing that decrease since 66 percent of households in 2012; DHS 2012. the lowest quintile forgo health services due to its cost. Further analysis confirms that belonging to the Affordability is particularly acute for the poor, highest and third quintiles was associated with CHEs two-thirds of whom did not seek care in 2013. in Haiti. Studies from other countries, such as Bolivia In 2013, 66 percent of households in the lowest (Aguilera Rivera, Xu, and Carrin 2006) and Argentina 47 See appendix B for a more detailed analysis of access. 48 Catastrophic health expenditures are one way to measure financial protection. Health spending is labeled catastrophic if households spend a certain threshold of their income or nonfood consumption on health. CHEs are considered a key measure of financial protection because they indicate whether a health system is able to protect its citizens from extreme financial hardship (Murray et al. 2003). In this study, CHEs were estimated using the ADePT software based on the Survey on the Living Conditions of Households after the Earthquake (Enquête sur les Conditions de Vie des Ménages après le Séisme, ECVMAS). A household encounters CHEs when allocating 25 percent or more of its nonfood consumption to health. BETTER SPENDING, BETTER CARE: 52 A LOOK AT HAITI’S HEALTH FINANCING FIGURE 5.2: Incidence Rate of Catastrophic Health Expenditures, by Wealth Quintile: Haiti, 2012 and 2013 Lowest 2.7 4.1 Second 9.8 6.1 Third 8.4 3.2 Fourth 9.4 2.1 Highest 8.1 1.5 Urban 5.2 1.6 Rural 9.8 5 National 7.7 3.4 0 1 2 3 4 5 6 7 8 9 10 Incidence rate as percentage 2013 2012 Source: World Bank estimates using AdePT based on ECVMAS I and II. Out-of-Pocket Payments for Health, Household and per Capita, by Wealth Quintile: Haiti, 2012 TABLE 5.2: and 2013 Household Per capita Wealth quintile % change 2012 2013 % change 2012 2013 Lowest 1,498.57 1,398.12 –7% 189.46 162.23 –14% Second 2,588.97 3,409.05 32% 340.49 405.99 19% Third 2,698.28 5,773.08 114% 437.14 797.51 82% Fourth 2,999.27 5,751.42 92% 543.92 863.47 59% Highest 6,907.17 10,845.31 57% 1,608.06 2,249.90 40% Source: World Bank estimates based on data from ECVMAS 2012 and 2013. Note: Out-of-pocket payments include user fees for consultations, costs of hospitalization, medicines, medical furniture and ancillary services, prostheses, and eyeglasses. (Cavagnero et al. 2006), confirm these results and Mean Out-of-Pocket Payments per TABLE 5.3: find that households with higher incomes are more Capita: Haiti, 2012 and 2013 likely to face CHEs than the poorest households sim- 2012 2013 % change ply because the richer are willing to spend more mon- ey on health services. Total health 664 1,032 55% Hospitals 1,197 844 –29% Affordability is an issue for specific services. Eyeglasses 867 721 –17% Households face CHEs because of visiting a private Medical supply 511 631 23% clinic or undergoing hospitalization. A first analysis Consultation 205 327 60% shows that the mean out-of-pocket expenditures are the highest for hospitals and medicines (table 5.3). A Medicines 758 1,380 82% more in-depth regression analysis (see details in table Lab exams 302 349 16% B.2 in appendix B) reveals two of the main charac- Source: World Bank estimates based on ECVMAS I and II. Note: Currency is in Haitian gourdes teristics associated with CHEs. First, households visit- ing a private clinic are almost three times more like- ly to encounter CHEs, which is statistically significant. private providers and hospitalization is well document- And, second, households in which a member went to ed in the existing literature (Adhikari, Maskay, and a hospital within the last 12 months are three times Sharma 2009; Li 2012; Brinda, Andres, and Enemark more likely to encounter CHEs (p < .001) than those 2014). Households consulting a traditional healer who did not. The association between CHEs and are also twice as likely to incur CHEs compared with Chapter 5 • ACCESS TO HEALTH SERVICES 53 TABLEAU 5.4: Utilisation des services de santé par quintile de richesse: Haïti Pour cent Community Wealth Public Public Traditional Private Ambulant health Pharmacy Others Total quintile dispensary hospital healer provider drug seller worker Lowest 23 23 13 6 17 1 9 5 100 Second 18 29 6 9 20 3 10 8 100 Third 18 27 6 6 28 4 7 5 100 Fourth 18 28 4 5 31 5 4 5 100 Highest 14 26 3 3 40 5 6 5 100 Total 18 27 6 6 28 4 7 5 100 Source: Estimations de la Banque mondiale sur la base de l’ECVMAS II, 2013. households visiting a public dispensary or health cen- about twice (p < .01 and p < .05, respectively) as like- ter (p < .05). This finding could be of concern because ly to encounter CHEs. Furthermore, households with households from the lowest wealth quintiles consult two or more children under the age of 5 years are 1.5 traditional healers more often than households from times (p < .001) more likely to seek care and 1.7 times the highest wealth quintile (table 5.4). And yet, the (p < .05) more likely to encounter CHEs than those performances of traditional healers are neither regu- without any children. In addition, households headed lated nor monitored and could pose a health risk for by a member who failed to complete a primary ed- the poor. There is also an association between CHEs ucation are 1.3 times more likely (p < .005) to seek and accessing specialized services (such as eye care). care and 2.1 times more likely (p < .001) to encoun- Accessing specialized services can imply high costs be- ter CHEs than households headed by a member with cause of the purchase of medicines and the consulta- no education. Households headed by members with tion fees in hospitals. health insurance are 3.5 times (p < 0.001) more likely to seek care than those headed by uninsured mem- bers. However, health insurance status is not associ- Socioeconomic Factors ated with a greater likelihood of encountering CHEs. Households located within urban areas are about half Socioeconomic factors also influence health-seek- (p < 0.05) as likely to encounter CHEs than those lo- ing behavior and the likelihood of incurring CHEs, cated in rural areas. Although about one-fifth of the whereas health insurance only has an impact on determinants of CHEs are explained by this statistical health-seeking behavior. Households headed by an modeling, the determinants of health-seeking behav- inactive person are 1.5 times more likely to consult a ior are far more complex, and it is difficult to include all health care provider when sick than those headed by the explanatory variables, such as distance from health an employed person (p < .001). This finding reflects the facilities and cultural factors.49 fact that such households have more household mem- bers over the age of 65, thereby requiring more health Distance is another key obstacle to care in Haiti services (see tables B.1 and B.2 in appendix B for more and could be addressed by an improved commu- details). Households with one elderly (65 and older) nity health system and reliable referral system. The member are 1.6 times more likely to encounter CHEs 2012 Demographic and Health Survey (DHS) pointed (p < .05) than those without. Compared with house- out that transportation was the second reason, af- holds headed by an employed member, those headed ter finance, for the low access of women ages 15– by an inactive or unemployed household member are 49 to health services; 43 percent did not seek care 49 Fewer conclusions can be drawn from the statistical model on determinants of health-seeking behavior than from the model on incidence of CHEs. Overall, R-squared in the former model remains weak at 8 percent. BETTER SPENDING, BETTER CARE: 54 A LOOK AT HAITI’S HEALTH FINANCING FIGURE 5.3: Reported Obstacles to Access to Health Care Services, by Wealth Quintile: Haiti, 2005–06 and 2012 a. 2005–06 b. 2012 19 21 24 43 60 Highest 76 11 9 24 31 15 24 79 Fourth 57 16 8 29 17 44 31 83 Middle 77 17 9 37 20 61 44 89 Second 83 21 9 40 26 72 61 92 Lowest 86 22 10 28 32 43 74 78 Total 90 17 11 0 20 40 60 80 100 0 20 40 60 80 100 Percent of respondents Percent of respondents Not willing to go alone Distance to health provider Not having money for treatment Not having permission to go for treatment Sources: DHS 2005–06, 2012. because of distance (figure 5.3).50 Several studies on do not always capture the role of cultural practices in health-seeking behaviors related to maternal and child influencing health-seeking behavior. Several qualitative health confirm that distance impedes access to health studies highlight the role of religion, voodoo, and oth- services, particularly in rural areas. Of 2,030 individuals er cultural factors in health-seeking behavior in Haiti. who reported diarrhea in the North department, two- For example, a study examining the determinants of thirds did not consult a provider because of distance. seeking care for mental health problems in rural Haiti As for maternal health, only 55 percent of rural house- revealed that 32 percent of respondents selected God holds live within 10 kilometers of a basic emergency as the first choice of care, followed by clinics and hos- obstetric facility versus 90 percent of households in ur- pitals (Wagenaar et al. 2013). A descriptive analysis ban areas. Alexandre et al. (2005) also confirm that of the health-seeking behavior of pregnant women long travel time and greater distance to centers in ru- found that their health-seeking decision making was ral areas are barriers to repeated antenatal care (ANC) guided by underlying core beliefs about wind/bad air visits. This does not mean, however, that the system and the need to obtain permission for any action from needs more health facilities, but that the referral sys- their husbands. In other cases, even though mothers tem with appropriate transportation needs to be im- recognized that umbilical cords infection pose a poten- proved. Strengthening the community system could tial health threat to newborns, they often misattribut- also improve access to basic primary health care (PHC) ed that threat to a mythical or voodoo-based explana- services, including maternal and child health services. tion rather germ theory (Walsh et al. 2015). Religious and cultural practices seem to influ- When encountering a health problem, almost ence health-seeking behaviors as well. Data from half of the population goes to a public provider. ECVMAS (2013) indicate that cultural aspects play a According to the 2013 ECVMAS, on average 45 per- marginal role in influencing health-seeking behavior in cent of households visit a public facility when sick, Haiti. According to ECVMAS, of 438 sick patients, 1 per- with little variation among economic gradients (ta- cent did not consult a provider because of cultural rea- ble 5.5). Another significant portion consults a private sons. However, quantitative surveys, such as ECVMAS, provider (28 percent), with a much higher prevalence 50 Distance was not perceived to be a key barrier to access to health care services in the ECVMAS survey for two reasons. First, the population in the two surveys is different. In ECVMAS, it was the head of household who was interviewed, whereas in the Demographic and Health Survey it was women ages 15–49. Second, ECVMAS offered more options for answers on why households did not go to a doctor when sick, whereas there were very few categories of answers in the DHS. Chapter 5 • ACCESS TO HEALTH SERVICES 55 TABLE 5.5: Participation Incidence Rates for Health Services, by Wealth Quintile: Haiti, 2013 Percent Lowest Second Fourth Highest Third wealth wealth wealth wealth wealth Total quintile quintile quintile quintile quintile Public dispensary 18 17 22 21 22 100 Public hospital 13 17 22 22 26 100 Community health worker and traditional 34 15 21 14 16 100 birth attendant Traditional healer 14 26 28 19 13 100 Private provider 9 10 18 24 39 100 Pharmacy 5 11 23 28 34 100 Drug seller 18 16 33 10 23 100 Source: World Bank estimates based on ECVMAS 2013. among the fourth and highest wealth quintiles. This assessment (World Bank 2014). In addition, more ben- could explain the higher prevalence of CHEs among eficiaries in the fourth and highest wealth quintiles go the richer. Although private providers have higher user to public hospitals. The only services that seem clear- fees than facilities under the Ministry of Public Health ly pro-poor are those provided by CHWs at nongov- and Population (Ministère de la Santé Publique et de ernmental organizations (NGOs) and the MSPP–more la Population, MSPP), one-fifth of the lowest wealth than a third of the beneficiaries of these services are quintile goes to a private facility when sick. Meanwhile, from the lowest wealth quintiles (table 5.5). Finally, a the same quintile relies more often on a community matter of concern is the fact that among households health worker (CHW) or consults mobile clinics more buying medications from street vendors, an important than the richest wealth quintiles. The lowest and sec- proportion (almost 70 percent) comes from the three ond wealth quintiles also have a higher incidence of us- lowest wealth quintiles (World Bank estimates based ing street vendors for medications than the fourth and on SPA 2013). Equally worrisome is the high share of highest wealth quintiles, certainly because their pric- people from the lowest, second and third wealth quin- es are cheaper than those at pharmacies. This finding, tiles who go to traditional healers. The services of both together with the fact that 22 percent of households street vendors and traditional healers are not moni- do not go to health facilities because of self-treatment tored, and little is known about their fees. Therefore, practices, could explain why medicine is the highest there is a risk that some of the poor pay a significant OOP expense (see table 5.3). The prevalence of con- amount of money for ineffective treatments. sulting a traditional healer is also higher among the lowest wealth quintiles (lowest, second, and third) A larger proportion of publicly managed facilities than the highest wealth quintiles (fourth and highest). than NGO facilities charge user fees. The proportion of MSPP facilities charging routine user fees (almost 94 And yet, public facilities, including public dispen- percent) is higher than that of NGO facilities (85 per- saries, receive a higher portion of richer beneficia- cent), which explains why MSPP facilities are not neces- ries, demonstrating that public health facilities are sarily pro-poor (IHE and ICF International 2014). In addi- not pro-poor. Among households going to the public tion, a higher share of dispensaries (95 percent) charge dispensaries, 18 percent belong to the lowest wealth routine user fees than hospitals (90 percent) and health quintile, 17 percent to the second quintile, 21 percent centers (93 percent)–(SPA 2013). This difference could to the fourth quintile, and 22 percent to the highest reflect the fact that hospitals and health centers gen- quintile (table 5.5). This finding is striking because dis- erally receive more funding from the government and pensaries are in general thought to be pro-poor as donors, whereas dispensaries usually have very small they are located in rural areas where the majority of operational budgets and so often rely exclusively on the population is poor, according to the latest poverty user fees for all operating expenses (except salaries). BETTER SPENDING, BETTER CARE: 56 A LOOK AT HAITI’S HEALTH FINANCING TABLE 5.6: Impoverishment Resulting from Health Expenditures, by Poverty Line: Haiti, 2012 Percentage point Pre-health payments Post-health payments % change (pp) change Moderate poverty Poverty headcount 58.6%a 59.3% 0.7 pp 1% Poverty gap (HTG) 7,361 7,548 187 pp 3% Extreme poverty Poverty headcount 23.7% 24.7% 1 pp 4% Poverty gap (HTG) 1,168 1,226 58 pp 3% Source: AdePT, using ECVMAS 2012. Note: HTG = Haitian gourde. Pre-health payments include both recurrent and exceptional health expenditures. Because of the high poverty rate in Haiti, any FIGURE 5.4:Mean Amount of User Fees Paid by amount of user fees, even very low ones, can deter Patients for Antenatal Care (ANC) Services: Haiti, the poor from seeking care. According to World Bank 2013 estimates for 2013 (table 5.4), dispensaries and MSPP 200 facilities charged lower user fees than hospitals and 181 private facilities. On average, user fees for ANC visits 180 were lower among MSPP facilities than those charged 160 by the private for-profit sector (figure 5.4). However, the presence of user fees at the public dispensary level, Mean amount of user fees (HTG) 140 even low, can deter the extremely poor–representing 120 24 percent of the population in 2012–from consulting 120 a health provider. On average, an ANC visit costs HTG 100 60 ($1.50) at a dispensary and HTG 77 ($1.60) at a hos- 77 pital.51 The extremely poor live on less than $1.25 a day. 80 65 60 59 Thus the user fee at the dispensary level corresponds to 60 more than one day of consumption, without counting other costs (such as transportation) and the opportuni- 40 ty cost of seeking care (such as the revenue lost from absence at work because of illness). In addition, these 20 data show that there is not a coherent user fee policy 0 because fees are almost the same for dispensaries and Public For-profit hospitals, thereby giving patients the wrong incentive Dispensary Health centers Hospitals to seek care directly at the hospital level and contribut- Source: World Bank estimates based on ANC data set, 2013. ing to the inefficiencies in the system. Note: HTG = Haitian gourde. Because the poorest segments of the population poverty (see table 5.6).52 Such low figures reflect the cannot afford health care, they forgo care instead low level of CHEs in Haiti. However, in the face of do- of paying. Thus catastrophic health expenditures nor withdrawal, the proportion of households impov- push only a small proportion of the population into erished by CHEs is likely to increase. poverty. According to the data, 0.7 percent of house- holds do transition into moderate poverty because of Among the various strategies used by households high health expenditures, and 1 percent into extreme in Haiti to cope with health expenditures, drawing 51 Based on the 2013 exchange rate. 52 The analysis used $1.25 and $2.50 for the extreme and moderate poverty lines, respectively. The analysis will have to be updated with the new poverty lines of $1.90 and $3.10 for extreme and moderate poverty recently developed by the World Bank. Thus impoverishment from health expenditures will be greater than that presented here. Chapter 5 • ACCESS TO HEALTH SERVICES 57 Share of Survey Respondents Reporting Various Strategies to Cope with Financial Losses Arising TABLE 5.7: from Health Problems, by Wealth Quintile: Haiti, 2012 Percent Wealth quintile Lowest Second Third Fourth Highest Total Savings 20 22 28 42 42 32 Food support from parents/friends 3 6 3 2 3 3 Financial support from parents/friends 13 10 13 10 10 11 Reduction in food consumption 4 4 5 3 1 3 Loan from parents/friends 20 17 13 13 18 16 Loan from banks/shops 4 4 3 4 1 3 Sale of livestock and farming assets 14 13 8 3 4 8 Committed to spiritual activities 5 1 2 2 2 2 No strategy 5 5 7 4 7 6 Other strategies 13 17 16 17 12 15 Total 101 100 100 100 100 100 Source: World Bank calculations based on ECVMAS I 2012. on savings and borrowing money from friends and from CHEs and health shocks. Households whose family are the main ones adopted. In 2012, overall, heads are unemployed, have no education, or are re- 32 percent of households who experienced a health tired, and households with elderly and children seem to problem used savings to pay for health care services, be the main populations affected by CHEs. These popu- 11 percent received financial support from parents or lations are not covered by any social protection mecha- friends, and 16 percent borrowed money from par- nism. Only wage employees working in the formal sec- ents or friends (table 5.7). The proportion of house- tor have access to the limited social insurance schemes holds who cope with health problems using no par- (health, pension, disability) in Haiti (World Bank 2016a). ticular strategy could reflect the prevalence of CHEs. Meanwhile, only 2.6 percent of the elderly population These households pay out-of-pocket and see a direct has access to a pension (old age, disability), but 92 per- reduction in their nonfood expenditures. On average cent of them live in urban areas, whereas most of the in 2012, 6 percent of the population had no particular poor live in rural areas. The social assistance system– strategy to cope with health problems (table 5.7). mainly composed of scholarships, food aid, and other transfers–is weak and covers only 8 percent of the pop- Very few social protection and assistance mecha- ulation. Support is highly fragmented, and little support nisms are in place to protect vulnerable populations exists for children under 5 (World Bank 2016a). BETTER SPENDING, BETTER CARE: 58 A LOOK AT HAITI’S HEALTH FINANCING CHAPTER 6 EFFICIENCY ANALYSIS Allocative Efficiency Compared with other low-income countries, Haiti’s health sector is relatively well resourced but with- out significantly better health outcomes, pointing to low overall efficiency. According to recent model estimates, the 2015 maternal mortality ratio (MMR) in Haiti is 359 deaths per 100,000 live births.53 The aver- age MMR across all low-income countries (LICs)54 (496) is much higher, and more than half (21) of LICs exhib- it higher MMR model estimates than Haiti (figure 6.1). As such, Haiti appears to be in relatively good stand- ing compared with similar country contexts. However, Haiti’s 2014 total health expenditure (THE) per capi- ta55 ($130.80) is the seventh highest of those of all LICs (which range from $25 to $223.70) and almost 1.5 times higher than the average 2014 THE ($91.30) across all LICs. Furthermore, the 2015 infant mortality rate (IMR) in Haiti is 52.2 deaths per 1,000 live births, which 53 Unless otherwise specified, all MMR, IMR and THE data cited in this paragraph were retrieved from the World Health Organization’s Global Health Expenditure database (see http://apps.who.int/nha/database for the most recent updates) via the 2016 World Development Indicators database at the World Bank. PHOTO CREDIT : UN/MINUSTAH 54 For the purposes of this report, the term low-income countries refers to the 30 countries in the low-income category defined by the World Bank. 55 The total health expenditure is the sum of public and private health expenditures as a ratio of the total population. It covers the provision of health services (preventive and curative), family planning activities, nutrition activities, and emergency aid designated for health, but it does not include the provision of water and sanitation. Data are in international dollars converted using 2011 purchasing power parity (PPP) rates. BETTER SPENDING, BETTER CARE: 60 A LOOK AT HAITI’S HEALTH FINANCING Chapter 6 • EFFICIENCY ANALYSIS 61 Infant Mortality Rate and Maternal Mortality Ratio against Total Health Expenditure per Capita: FIGURE 6.1: LICs and Haiti, 2014/15 100 1600 90 CAF SLE 1400 TCD SLE 80 ZAR MLI 1200 70 SSD BEN Infant deaths per 1,000 live births AFG Maternal deaths per 100,000 live births GIN 1000 60 GNB BFA NER MOZ CAF BDI COM HTI TGO TCD LBR 50 LIC 800 SSD ZWE GMB BDI LBR GMB ZAR MWI 40 ETH GNI MWI ZWE MDG UGA MLI 600 ERI ERI GNB TZA 30 BGD LIC RWA NER MOZ NPL BFA BEN TZA KHM 400 MDG BFA HTI UGA AFG 20 ETH COM RWA NPL 200 10 BGD KHM 0 0 0 50 100 150 200 250 Per capita total health expenditure 2014 (2011 international $) Mortality rate, infant (per 1,000 live births) 2015 [YR2015] Maternal mortality ratio (modeled estimate, per 100,000 live births) 2015 [YR2015] Source: World Bank estimates based on WDI and GHED 2016. Note: LICs = low-income countries. For country codes, see http://www.nationsonline.org/oneworld/country_code_list.htm. is close to the average IMR across all LICs (53.1 deaths Haiti’s disease burden, resources need to be shifted per 1,000 live births) and the 19th highest IMR among from curative to preventive care. The three leading all LICs. By contrast, the 2014 per capita total health ex- causes of disability-adjusted life years (DALYs) in Haiti penditure in the four LICs with MMRs within 15 more are the human immunodeficiency virus (HIV), acute re- or less of Haiti’s–Madagascar (353), Ethiopia (353), spiratory infections, and diarrhea, which could be ad- Togo (368), and Burkina Faso (371)–are much lower at dressed by preventive care interventions. According $43.70, $73, $76.30, and $82.3, respectively. The 2014 to the last National Health Account (NHA) exercise IMRs in LICs with levels of total health expenditure per (2012–13), the allocation to preventive care is 19 capita similar to that Haiti–Rwanda, 125, and Uganda, percent of THE, while the allocation to curative ser- 132.6–are considerably lower at 31.1 and 37.7 infant vices is almost three times as much, 54 percent (fig- deaths per 1,000 live births, respectively (figure 6.1). ure 6.2). Development partners share the responsibili- ty for the way resources are allocated because donors Poor allocative efficiency may be one explanation provided 52 percent of THE in 2012–13 (NHA 2012– for low health sector performance. Considering 13).56 In the future, the Ministry of Public Health and 56 The NHA takes into account the Haitian fiscal year from September to October. For that reason, this number is a bit different than the one shown in figure 4.9 using the Global Health Expenditure database from the World Health Organization (WHO), which shows external funds as percentage of THE to be 59 percent in 2012. BETTER SPENDING, BETTER CARE: 62 A LOOK AT HAITI’S HEALTH FINANCING Total Recurrent Health Expenditures, by FIGURE 6.2: flows, but these activities remain fragmented. The Function: Haiti, 2012–13 main findings of these studies indicate that the gov- 100 ernment of Haiti would need $15 million to run, for ex- 90 ample, the University Hospital of Mirebalais (Baruwa et 23 Percentage of budget allocated 80 al. 2015). Another study shows that an additional $12 million would be needed to run the National University 70 19 Hospital (AEDES 2016). Just for these two hospitals, 60 the total operating amount would represent 40 per- 50 cent of the government’s operational budget allocat- 40 ed to health, which, as noted earlier, is mainly fund- 30 54 ing the government’s payroll. Thus how to finance the 20 running costs of the existing hospital infrastructure re- 10 mains a challenge. 3 0 2012-13 At the primary care level, an increase in operational Governance Preventive services Curative services Other expenditures and the reclassification of some non- Sources: MSPP 2014, 2015a. performing community referral hospitals (hopitaux Note: THE = total health expenditure. communautaire de reference, HCRs)58 as primary care facilities at a lower level should be key pri- Population (Ministère de la Santé Publique et de la orities. Compared with other countries, Haiti has low Population, MSPP) and donors need to consider Haiti’s physical access to the primary care level.59 The country burden of disease when making investment decisions has only 0.3 dispensaries per 10,000 inhabitants, with in the health sector. In view of Haiti’s double burden large variations by department (figure 6.3). This ratio of disease–that is, the coexistence of communicable is well below the norm set by the MSPP in its carte and noncommunicable diseases (NCDs) as the main sanitaire, as well as that of other countries. For exam- causes of death–health prevention and promotion in- ple, the state of Maharashtra in India has one subcen- terventions would yield the highest rate of return on ter (which is comparable to a dispensary in Haiti) per investments because such interventions address both 5,000 inhabitants (Awate 2014), and in Liberia, the infectious diseases and the emerging priorities related government operates one health clinic for each catch- to NCDs. For example, the return on investments on ment of 5,000-10,000 inhabitants (MoHSW 2008). health promotion programs is estimated to be $3–$10 The physical access of Haitians to the second level of for every $1 invested (Coe and de Beyer 2014). primary health care, the health center, is better: Haiti has 1.2 health centers per 30,000 inhabitants, which Although resources should be allocated to preven- is comparable to the situation in other LICs (MoHSW tive care, a major question is how to finance the 2008; Awate 2014). By contrast, the density of com- operational and maintenance costs of the existing munity referral hospitals is very high in Haiti compared hospital infrastructure, which plays a substantial with international benchmarks: 1.4 community re- role in curative care.57 After 2010 earthquake, most ferral hospitals per 150,000 inhabitants versus 1 per hospital construction and rehabilitation work was per- 150,000–250,000 inhabitants in other LICs (MoHSW formed by donors (MSPP 2016). Because of the ur- 2008; Awate 2014; Ujoh and Kwaghsende 2014). The gent situation, little effort was made to understand current ratio of community hospitals to population in the long-term financial implications for running these Haiti also exceeds the norm set by the MSPP (2006). hospitals. Various donors have now begun to formu- The government should explore the potential of con- late business plans to support the MSPP in develop- verting community referral hospitals with poor perfor- ing a more comprehensive picture of hospital financial mance to primary care facilities at a lower level. 57 See table 7 of the NHA 2012–13 (MSPP 2015b). 58 Haiti has 40 community referral hospitals and 65 “small hospitals” (SHs). The latter are the same size as the community referral hospitals according to SPA data, and thus these two hospital categories can be regrouped. If not specifically mentioned, it is assumed that small hospitals are counted with community referral hospitals. 59 Evidence indicates that some institutions in Haiti are not correctly classified. For example, several institutions that are classified as health centers operate as dispensaries. Chapter 6 • EFFICIENCY ANALYSIS 63 FIGURE 6.3: Density of Health Infrastructures: Haiti, 2013 2.5 Formations sanitaires pour 10 000 habitants Norm is 1 dispensary per 10,000 inhabitants; 1 health center per 30,000 2 inhabitants; 1 HCR per 150,000 inhabitants 1.5 1 0.5 0 South North-West North-East Grand’Anse Artibonite North West Nippes South-East Center Mean Density of dispensaries per 10,000 inhabitants Density of health centers per 30,000 inhabitants Density of HCRs/SHs per 150,000 inhabitants (DHs and UHs not included) Source: World Bank estimates based on SPA 2014. Note: The carte sanitaire implies that a dispensary covers a subcommune (population, 10,000), while a health center covers a commune (population, 30,000 on average). There should be one HCR per arrondissement covering a population of between 150,000 and 250,000 (MSPP 2006). HCRs include small hospitals, which have almost the same number of beds as HCRs. DH = departmental hospital; HCR = hôpital communautaire de référence (community referral hospital); SH = small hospital; UH = university hospital. Relationship among Government Health Expenditure, Poverty, Health Supply, and Coverage at FIGURE 6.4: Departmental Level: Haiti, 2012–13 45,000 160 Public health expenditures per 10,000 inh. (HTG) 40,000 140 35,000 120 30,000 100 25,000 Persons 80 20,000 60 15,000 40 10,000 5,000 20 - 0 Ouest Sud-Est Nord Nord-Est Artibonite Centre Sud Grand'Anse Nord-Ouest Nippes Department Public health expenditures per 10,000 inhabitants Poor population per 10,000 inhabitants Medical staff per 10,000 inhabitants Patients who delivered with assistance of qualified staff Sources: World Bank estimates based on BOOST 2016; World Bank 2014; DHS 2012; SPA 2013. Note: Government health spending was reported for fiscal 2012-13 and only included the operational budget. The prevalence of poor per 10,000 inhabitants is from the World Bank’s Haiti Poverty Assessment, which relies on ECVMAS 2012. The density of medical staff was estimated based on SPA 2013, and data on the share of deliveries assisted by qualified staff are from the Demographic and Health Survey (DHS). BETTER SPENDING, BETTER CARE: 64 A LOOK AT HAITI’S HEALTH FINANCING All institutions in Haiti, as well as desirable referral exercise would be to map all facilities, the services they paths between institutions, should be mapped, de- should provide, as well as referral paths. partment by department, to guide investment de- cisions. Furthermore, the shift of resources from hospi- Allocations of financing from the central to the de- tals to the primary care level should be guided by data. partmental level should also be guided by the in- An updated facility mapping or carte sanitaire based troduction of an allocation formula based on eq- on population needs ought to be developed. For ex- uity and efficiency principles. Usually, the MSPP and ample, departments such as Artibonite that have a low development partners determine allocations to de- density of health centers, dispensaries, and community partments based on historical allocations (the previous referral hospitals should be prioritized. However, this year’s budget). Figure 6.4 shows, by department, the does not necessarily mean building new dispensaries. MSPP allocation per 10,000 inhabitants, the number Instead, certain inefficient facilities (such as communi- of poor per 10,000 inhabitants, as well as health in- ty referral hospitals) could be converted to offer health dicators. It illustrates that the distribution of the op- promotion services and primary care services. In this erational budget does not necessarily align with the process, it is critical to agree on a minimum package needs of the population. For example, Artibonite’s an- of services that will be financed and provided at the nual spending per 10,000 inhabitants is 12,000 Haitian primary care level. As shown earlier, very few dispen- gourdes (HTG) versus HTG 14,000 in the North. And saries have the resources needed to offer preventive yet Artibonite has more poor people per 10,000 in- services and provide referral services. Indeed, currently habitants, a lower density of medical staff, and worse there is no budget line for primary health care in the coverage and institutional deliveries than the North. In operational budget of the MSPP. In the future, it will be the future, resource allocations, both by MSPP and de- important to include this line because it is essential to velopment partners, need to be guided by a rational increasing service readiness for delivery of an agreed- allocation formula that takes into account the disease on minimum package of services (see discussion of ser- burden, health systems characteristics, and population vice readiness in chapter 3). A starting point for this of the particular department. Technical Efficiency Solving access problems will require increasing post/dispensary level in, for example, Ethiopia and the productivity of dispensaries. In Haiti, 359 dis- Guatemala (table 6.1). pensaries, staffed by at least one nurse or aid nurse (or both), provide prevention services (SPA 2013). The density of health centers is similar to those in As described earlier, the density of dispensaries per other LICs, but their performance is much weaker 10,000 inhabitants is very low. Although increasing in Haiti. Haiti has 129 health centers with beds (cen- the number of dispensaries is recommended, this tres de santé avec lit CALs) and 298 health centers measure alone will not solve the access problem. An without beds (centres de santé sans lit, CSLs) that pro- analysis of the efficiency of service delivery at the vide diagnostic, curative, and preventive services. CSLs primary care level using a data envelopment analy- and CALs have at least one medical doctor and at least sis (DEA) method (box 6.1) revealed that less than 1 one lab technician to provide diagnostic services, al- percent of dispensaries are efficient in terms of the though not all CSLs offer these services and are ful- number of patient visits for a given number of staff. ly staffed. Haiti has a sufficient density of health cen- In fact, for every 342 dispensaries, only one was effi- ters per 30,000 inhabitants. However, CSLs and CALs cient –that is, accommodated a sufficient number of produce little for the resources they are given, espe- visits for the number of staff available. In Haiti, the cially the CSLs. The analysis found that only 4 percent mean technical efficiency (TE) score for dispensaries of CALs and less than 1 percent of CSLs were using is 0.04, which is much lower than those discovered their resources efficiently: they had the right number of in similar studies of technical efficiency at the health staff and beds (beds only for CALs) for the number of Chapter 6 • EFFICIENCY ANALYSIS 65 BOX 6.1 Definition of Technical Efficiency and DEA Methodology in Haiti In the health sector, technical efficiency consists of achieving a maximum level of consultations or admissions to a health facility with a given level of inputs (Street et al. 2011). There are two cases: (1) input-oriented technical efficiency, which aims at determining the pro- portion of inputs (personnel, other expenses) that must be used to produce a given number of consultations; and (2) output-oriented technical efficiency, which gauges the additional number of consultations possible without having to change the health facility’s number of inputs (Coelli 1996). Linear programming, known as data envelopment analysis (DEA), is a nonparametric method that determines the number of health care facilities included in an efficiency frontier. This method produces a technical efficiency score based on the number of inputs, such as per- sonnel, current expenditures, and results (consultations, hospital admissions). The technical efficiency score ranges from 0 to 1. A score of 1 means that the health facility is on the efficiency frontier and so is efficient. A score below 1 demonstrates poor performance, espe- cially if the score is close to 0. Initially applied in the industrial sector, this methodology is being used increasingly in the health sector to measure the technical efficiency of hospitals or primary health care facilities. In Haiti, several DEA analyses were conducted, all using SPA data sets. An output-oriented model was chosen because we wanted to know how much health facilities could produce with the resources available. Two separate DEA analyses were undertaken: one for dis- pensaries, health centers without bed (centres de santé sans lit, CSLs), and health centers with bed (centres de santé avec lit, CALs)–that is, the primary health care (PHC) level–and one for hospitals or secondary and tertiary care facilities. At the PHC level, separate DEA analysis was conducted for dispensaries, CSLs, and CALs because all three facilities produce different outputs and have various staffing standards: dispensaries focus on preventive visits and have only one nurse or aid nurse; CSLs provide both curative and preventive services; and CALs provide preventive and curative consultations as well as hospitalization. Inputs for dis- pensaries and CSLs included number of medical staff. For CALs, inputs included number of medical staff and beds, and outputs included number of consultations and admissions. Because of missing data in the SPA data set, the sample was 342 dispensaries (out of 359), 265 CSLs (out of 298), and 72 CALs (out of 129). DEA analysis at the hospital level was conducted for a sample of 78 hospitals (out of 121). Various values were missing, in particular for outputs. In addition, several hospitals providing specialized services were excluded–for example, the Doctor without Borders hospital providing intensive care for burn patients or small hospitals providing only deliveries. Inputs included four separate categories of medi- cal staff (not possible at the PHC level because there are fewer staff): medical doctor, nurse, nurse assistant, and laboratory technician. Output included consultations and admissions. Four categories of hospitals were included: HCRs, small hospitals (equivalent to HCRs), departmental hospitals, and university hospitals. Because in university hospitals medical doctors allocate part of their time to teaching, lower technical efficiency scores were expected for these institutions. However, this was not the case in Haiti, where university hospitals, in fact, had some of the highest technical efficiency scores of all institutions. visits and admissions they produced. The mean tech- work to improve hospital performance are needed. nical efficiency score was 0.30 for CALs and 0.09 for Hospitals in Haiti do not perform very well and this is CSLs. However, these scores are weak compared with especially true for departmental and small hospitals those from similar studies on technical efficiency at the with a mean score of 0.36 (table 6.2). Physical access health center level (table 6.1). to secondary care–HCRs and small hospitals (SHs)–is high (figure 6.3). However, the mean technical efficien- Compared with other LICs, Haiti fares poorly in hos- cy score of hospitals is quite low (0.49) compared with pital efficiency, despite the fact that 38 percent of those of other countries (figure 6.5). Only 23 percent of its total health expenditure60 is spent at this level. hospitals in Haiti are efficient –that is, have a TE score Development of a hospital licensing policy and further equal to 1. Haiti’s bed occupancy rate (BOR),61 average 60 The 2012–13 NHA indicates that hospital expenditures represent 42 percent of recurrent health expenditures. By adding capital expenditures to recurrent expenditures, thereby estimating the total health expenditure, hospital expenditures represent 38 percent of THE. 61 BOR is the percentage of official beds occupied by hospital inpatients for a given period of time. In general, the greater the occupancy rate, the greater is the revenue for the hospital. BETTER SPENDING, BETTER CARE: 66 A LOOK AT HAITI’S HEALTH FINANCING TABLE 6.1: Technical Efficiency, Haiti and Other LICs THE at the hospital level often reaches 50 percent (MSH 2001), Haiti spends more than countries at a similar lev- % of sample Country that is not Average score Sample el of economic development. In recent years, Burundi, efficient (<1) Tanzania, and Afghanistan have spent, respective- ly, 23 percent (2015), 26 percent (2012), and 29 per- 96.00%, 0.30, 79 CALs, cent (2013) of THE on hospitals.65 Of greater concern, CALs; 99.24%, CALs; 0.09, Haiti 265 CSLs, 342 is that hospital-level expenditures in Haiti do not seem CSLs; 99.41%, CSLs; 0.04, dispensaries to translate into improved output. The low efficiency is dispensaries dispensaries Burkina 25 PHC certainly due to too many small hospitals, which, even – 0.86 if they were classified as hospitals, lack basic hospital Faso facilities services and equipment. Dealing with this issue will re- Ethiopia 75% 0.57 60 health posts quire an urgent effort to set up a licensing policy, im- Random pede further hospital construction, consolidate existing Ghana 78% 0.88 selection of 86 hospital infrastructure, map out needed hospitals, and health facilities ensure that hospital services are provided in selected 71%, institutions where the needed volume of care can be Guatemala but 53% have 0.78 34 health posts obtained to also improve the quality of hospital care. It a score >0.9 is important as well to explore how hospital manage- Sources: World Bank staff, 2016; Akzali et al. 2008; Sebastian and Lemma ment can be improved. A line of technical assistance 2010; Marshall and Flessa 2011; Hernandez and Sebastian 2013. Note: – = not available; CALs = centres de santé avec lit (health centers with will be needed to implement these important reforms. bed); CSLs = centres de santé sans lit (health centers without bed), LICs = low-income countries; PHC = primary health care. See table C.1 in appendix C for descriptive statistics of the dispensary and health center samples and more details on the data envelopment analysis score. Factors Influencing Technical Efficiency: Ownership and length of stay (ALOS),62 and unit cost per bed day63– Geography three proxies for efficiency of hospital care–also high- light poor hospital efficiency (the data used to estimate these three indicators are taken from the macro-cost- At the PHC level, facilities managed by nongovern- ing hospital study–see appendix C). Of these indicators, mental organizations (NGOs) perform better than Haiti performs best on the ALOS (figure 6.6). However, other facilities. Lessons from how well-performing fa- Haiti’s BOR is 29 percent, which is significantly lower cilities are managed should be transmitted to all fa- than that of other LICs (figure 6.7). The low BOR con- cilities. There is no broad variation in the TE score by tributes to a high unit cost per bed day of $76,64 which ownership at the PHC level, although the NGOs seem is much higher than that of, for example, Cambodia, to perform slightly better in their CSLs and dispensa- Guatemala, and the Philippines, all of which have a ries, while the MSPP scores better in its CALs (see table higher gross domestic product (GDP) per capita (figure C.3 in appendix C). This finding confirms those from a 6.8). The poor efficiency of Haitian hospitals is particu- previous study showing that technical assistance was larly alarming because Haiti spent 38 percent of its total associated with an increase in the utilization of PHC health expenditure at this level (MSPP 2015a). In abso- health services by 35 percent over three years (Zeng lute terms, this figure represents HTG 11,221 million, et al. 2013). However, the correlation between the TE or $260 million in fiscal 2012. Although the share of score and ownership is only statistically significant at 62 ALOS refers to the number of calendar days from the day of patient admission to the day of discharge averaged for all patients during the period. 63 This ratio provides a macro-costing of hospitals by estimating the unit cost of a hospital treatment day (bed day equivalent) and is indicative of the total resource cost per bed. The unit cost per bed equivalent is obtained by dividing total annual costs by the total number of bed day equivalents. Bed day equivalent includes both outpatient and inpatient cases. Outpatient visits were converted to equivalent bed days using the methodology of Shepard, Hodgkin, and Anthony (2000) in which one outpatient visit is equivalent to 0.32 inpatient cases. 64 The unit cost per inpatient day is usually twice as much at a referral hospital (departmental and university hospitals) as at a district/community hospital. Baruwa et al. (2015) found out that in the University Hospital of Mirebalais, the unit cost per inpatient ranged from $41 to $188 and major services cost more than $138 a day. Thus the macro-costing results of the community referral and small hospitals study are aligned with the findings of Baruwa et al. (2015) as the unit cost per bed day is much lower than $138. 65 See National Health Account reports from the Ministry of Health and Social Welfare, Republic of Tanzania (2012); Ministry of Public Health, Afghanistan (2013); and Ministry of Public Health, Burundi (2015). Chapter 6 • EFFICIENCY ANALYSIS 67 TABLE 6.2: Bed Capacity and Technical Efficiency Score, by Hospital Type: Haiti, 2013 Technical Total number Number of Number of Facility Number % of beds efficiency of beds beds (mean) beds (median) score (mean) University hospital (UH) 6 1,249 17 156 107 0.52 Departmental hospital (DH) 6 722 10 90 82 0.36 Community referral hospital (HCR) 32 1,977 27 49 38 0.52 Small hospital (SH) 34 3,256 45 50 26 0.36 Sample 78 7,198 100 59 40 0.49 Source: World Bank estimate based on SPA 2014. Note: Small hospitals do not belong in any official hospital category, such as an HCR, DH, or UH, but can be classified as a HCR having a similar average number of beds. Haiti has 121 hospitals according to the EPSSS data set (2013). They can be broken down as follows: 8 UHs, 6 DHs, 40 HCRs, and 65 SHs. For this analysis, however, data were missing for 43 hospitals, and thus the result in the first column of this table. See table C.2 in appendix C for descriptive statistics of the hospital sample. FIGURE 6.5: Distribution of Technical Efficiency Scores: Hospitals in Haiti and Selected Countries, Various Years 1 0.9 0.9 0.8 Efficiency score (range, 0-1) 0.69 0.71 0.67 0.7 0.61 0.6 0.49 0.5 0.4 0.3 0.2 0.1 0 Haiti (n=78) Ghana (n=128) Eritrea (n=19) Afghanistan (n=68) Namibia nN=30) Brazil (n=30) Sources: Data for Haiti are taken from a sample of 22 hospitals for which the World Bank collected hospital statistics and financial data from February to April 2016; Ghana (data from 2005): Jehu-Appiah et al. 2014; Eritrea (data from 2007): Kirigia and Asbu 2013; Afghanistan (data from 2012): Osmani 2015; Namibia (data from 1998–2001): Zere et al. 2006; Brazil (data from 2003–06): Lobo et al. 2010. the CSL level (p < .10)–see table C.4 in appendix C). Selected secondary care hospitals should be con- There is little variation in the TE score by location at the solidated to achieve economies of scale in hospital CSL and dispensary levels, but CALs located in urban care. As noted earlier, the density of hospitals per capi- and metropolitan areas scored higher than those in ru- ta in Haiti is higher than in other LICs. Haiti has approx- ral areas (see table C.3 in appendix C). This observa- imately 1.14 hospitals per 100,000 inhabitants (SPA tion is confirmed by the correlation between TE score 2013), which is nearly twice the density in LICs–0.08 and location at the CAL level (p < .05)–see table C.4. hospitals per 100,000 inhabitants (WHO 2015). For One possible explanation for this observation is great- hospital beds, this scenario is reversed. Haiti has approx- er demand for health services in urban and metropol- imately 6.4 hospital beds per 10,000 inhabitants, but itan areas than in rural areas. Differences in manage- an average of 21 hospital beds per 10,000 inhabitants ment practices across facility types and categories with are available in LICs (World Bank estimates using the higher TE scores should be studied and mainstreamed SPA data set, 2013; WHO 2015). Fewer hospital beds across all facilities whenever possible. across a larger number of hospitals may reduce the pa- tient flow in Haiti’s health facilities, and this finding may provide a possible explanation for the low hospital TE BETTER SPENDING, BETTER CARE: 68 A LOOK AT HAITI’S HEALTH FINANCING FIGURE 6.6: Average Length of Stay: Haiti and Selected Countries, Various Years 8 7 6 Length of stay (days) 5 4 3 2 1 0 Haiti Cambodia Namibia Eritrea Ghana Afghanistan Malaysia West Bank and Gaza Sources: Haiti: World Bank estimates based on SPA 2014; Cambodia (data from 2007): Collins, Gupta, and Sovannarith 2009; Namibia (data from 2000–2001): Zere et al. 2006; Eritrea (data from 2013): Ministry of Health, Eritrea 2014; Ghana (data from 2009): Salaeh 2013; Afghanistan (data from 2012): Osmani 2015; Malaysia (data from 2010): Nwagbara and Rasiah 2015; West Bank and Gaza (data from 2006–12): Hamidi 2016. FIGURE 6.7: Bed Occupancy Rate: Haiti and Selected Countries, Various Years 80 67 68 Percent of hospital beds occupied 59 60 61 60 38 40 29 20 0 Haiti Eritrea Afghanistan Ghana Malaysia Namibia Cambodia Sources: Haiti: World Bank estimates based on SPA 2014; Eritrea (data from 2013): Ministry of Health, Eritrea 2014; Afghanistan (data from 2012): Osmani 2015; Ghana (data from 2009): Saleh 2013; Malaysia (data from 2010): Nwagbara and Rasiah 2015; Namibia (data from 2000–2001): Zere et al. 2006; Cambodia (data from 2007): Collins, Gupta, and Sovannarith 2009. scores. As the theory of economies of scale highlights,66 (HCRs and small hospitals reduced to CALs) to generate a hospital sector in which small entities must each pro- efficiency gains. The hospitals shown in map 6.1 are col- vide (and afford) distinct diagnostic and administrative or coded to indicate those that are efficient (green) and services is likely to incur higher costs and lower produc- those that are inefficient (red). This mapping exercise tivity. Haiti’s hospital sector is subject to these inefficien- highlights clusters of inefficient hospitals that might be cies. Low patient flow also makes it difficult for medical consolidated to concentrate resources and capacity into staff to maintain a high level of proficiency in their skill a smaller number of higher-performing facilities. In sev- set. This problem can, in time, lead to poor quality of eral departments, such as North-West, Artibonite, and care and decreased patient safety. These findings and West, several secondary hospitals that are inefficient are observations in Haiti’s hospital sector make a strong case located close to each other, and they could potentially for downsizing or consolidating secondary care hospitals be consolidated to increase productivity. 66 This microeconomics theory describes the observation that as hospital operations scale up, the cost per unit of output decreases. Accordingly, a hospital sector with several small and operationally independent hospital units is associated with higher costs per unit of output. Chapter 6 • EFFICIENCY ANALYSIS 69 Relationship between Unit Cost per Bed Day and GDP per Capita: Haiti and Selected Countries, FIGURE 6.8: Various Years 4500 SLV 4000 GTM 3500 GDP per capita (current international US$) 3000 PHL 2500 2000 1500 KHM 1000 HTI 500 0 0 20 40 60 80 100 120 140 160 180 Cost per bed day (current international US$) Sources: Haiti (HTI, data from 2013): World Bank estimates based on SPA 2014; Cambodia (KHM, data from 2005): Suaya et al. 2009; Guatemala (GTM, data from 2006): Suaya et al. 2009; El Salvador (SLV, data from 2005): Suaya et al. 2009; Philippines (PHL, data from 2011): Largo 2012. Note: The unit cost per bed day may reflect the economic status of the country; the wealthier countries tend to be more expensive, possibly because of the higher costs of labor. The unit cost per bed day equivalent is $76.34 on average, which represents 9.2 percent of GDP per capita (high). This is certainly due to a low bed occupancy rate and high costs. NGO-managed hospitals perform better than MSPP NGOs ($61). Furthermore, the relationship between hospitals, and private for-profit hospitals are the the unit cost and ownership is statistically significant. least-performing entities. In line with findings on per- Regression analysis shows that hospitals managed by formance variations by ownership at the dispensary and private for-profit providers yield an increase in the unit clinic levels, NGO-managed hospitals perform better (TE cost per bed day of 109 percent compared with those score 0.6) than those managed by the MSPP (TE score managed by the MSPP (p < 0 .05)–see table C.8 in 0.47), and private for-profit hospitals (TE score 0.41)– appendix C. Low efficiency in private for-profit health see table C.5 in appendix C. The high level of technical facilities might be explained by the high number of assistance provided by NGOs may influence both sys- small facilities that receive a relatively lower number tem management capacity and facility service readiness. of admissions and consultations than other ownership However, this association is not statistically significant at types, which may itself be the result of high user fees at the hospital level. Therefore, further study is needed to admission. Still, private for-profit hospitals exhibit the explain these differences in performance levels. highest BOR (compared with NGO and MSPP facilities) and the highest ALOS, which contributes to the high Private for-profit hospitals are not just the unit cost per bed day. To enhance the performance of least-performing; they also spend more than MSPP private hospitals, these entities should be included in and NGO facilities.67 The MSPP needs to engage with the proposed hospital licensing program. these private entities and include them in the pro- posed hospital licensing program. Overall, hospitals Some departments have poorer technical efficien- run by the private for-profit sector have a higher unit cy scores, and these should immediately be priori- cost ($117) than hospitals run by the MSPP ($48) and tized by the MSPP and its partners. By linking the TE 67 The analysis that collected financial data at the hospital level included a sample of 22 small hospitals and community referral hospitals. See table C.6 in appendix C for descriptive statistics of the study sample. BETTER SPENDING, BETTER CARE: 70 A LOOK AT HAITI’S HEALTH FINANCING MAP 6.1: Efficient and Inefficient Secondary Care Hospitals, Haiti Inefficient hospitals that are in close proximity which could be merged to increase scale economy Efficients inefficients Source: World Bank estimate based on SPA 2014. TABLE 6.3: Technical Efficiency Scores, by Facility Type: Haiti, 2016 Technical efficiency score CALs CSLs Dispensaries Hospitals Total South 0.24 0.05 0.03 0.35 0.67 North-West 0.19 0.02 0.04 0.48 0.73 North-East 0.37 0.07 0.03 0.28 0.75 Grand’Anse 0.33 0.08 0.04 0.34 0.79 Artibonite 0.32 0.17 0.05 0.34 0.88 North 0.27 0.08 0.03 0.53 0.91 West 0.26 0.09 0.05 0.52 0.92 Nippes 0.16 0.04 0.03 0.7 0.93 South-East 0.26 0.05 0.05 0.63 0.99 Center 0.57 0.1 0.09 0.5 1.26 TE mean 0.30 0.09 0.04 0.49 0.92 Source: World Bank estimates based on SPA 2014. Note: Shaded values fall below the TE mean for that respective category. CALs = centres de santé avec lit (health centers with bed); CSLs = centres de santé sans lit (health centers without bed); TE = technical efficiency. score to each health facility, it is possible to assess tech- (0.91). Donors and the MSPP should focus their efforts nical efficiency by department (table 6.3). The overall on these departments because value-for-money could mean TE score is 0.92. Six departments are below this be improved. The South department has an abnormal- mean: South (0.67), North-West (0.73), North-East ly high number of hospitals per 150,000 inhabitants, (0.75), Grand’Anse (0.79), Artibonite (0.88), and North but they produce little (their TE score is 0.35 compared Chapter 6 • EFFICIENCY ANALYSIS 71 TABLE 6.4: Average Annual Wage of Medical Staff: Haiti and Selected Countries, Various Years U.S. dollars Haiti (public) Haiti (NGO) Burkina Faso (public) Rwanda (public) 2013 2013 2014 2011 Medical doctor 10,415.20 15,328.40 9,469.50 – Nurse 5,659.50 6,214.00 5,754.00 5,445.1 GNI (2014) 800 800 700 590 Ratio of salary to GNI per capita Medical doctor 13.0 19.2 13.5 – Nurse 7.1 7.8 8.2 7.9 Source: WB estimates based on World Bank, USAID, and MSPP 2013; Appaix, Henry, and Badjeck 2015; Collins et al. 2011. Note: – = not available. Figures include all monetary allowances, including income from per diems, but exclude other private income and nonmonetary benefits. In Haiti, the average wage of a medical doctor in the public sector was calculated based on 14 observations; for nurses, 12 observations. In the private sector, the average wage of a medical doctor is based on 14 observations; for nurses, 52 observations. The data were collected in three departments (World Bank, USAID, and MSPP 2013). In Burkina Faso, the average is from estimates from seven districts (Appaix 2015). In Rwanda, the average is estimated based on seven health facilities (Collins et al. 2012). GNI = gross national income; NGO = nongovernmental organization. with the mean TE score for hospitals countrywide of lead to wasted resources. The six patients seen per day 0.49). In addition, the TE scores of CALs, CSLs, and dis- by medical staff at the first-level primary health care fa- pensaries in the South are below the overall mean TE cilities is low compared with the numbers in other LICs scores for the same categories, and the same analysis (World Bank 2015a68). The low productivity of medical could be drawn for the North-West. Both the North- staff may also be the reason Haiti fares poorly in inter- East and Grand’Anse are performing very poorly at the national comparisons of the correlation between the dispensary, CSL, and hospital levels, but relatively bet- density of medical staff and key health outputs and ter at the CAL level compared with the national-level outcomes.69 Low productivity is often caused by high averages. Thus the focus should be on assessing the levels of absenteeism, which is linked to income-gener- service readiness of hospitals and merging or upgrad- ating opportunities outside of the facility (WHO 2006). ing some of them. In Artibonite, which has the lowest In Haiti, more than one-third of medical staff have a TE scores, the MSPP and development partners should second job in the Centre, North-West, and North-East prioritize service readiness and process management departments (World Bank, USAID, and MSPP 2013), at the dispensary and hospital levels. and those with a second job spend less time work- ing at the health facility (p < .05)–see table C.9 in ap- pendix C. Public sector wages for doctors and nurses Supply Factors Influencing the in Haiti are comparable to those in LICs70 (table 6.4). Technical Efficiency/Performance of Delays in payments demotivate staff, and health work- Health Facilities ers who experience payment delays are more likely to have a higher level of absenteeism (see table C.9). Not unexpectedly, absenteeism yields a tremendous waste At the PHC level, medical staff see only six patients of resources in the health system. For example, Haiti’s a day. Absenteeism and work outside of the facility 2014 operating budget for health amounted to $68 are key drivers of low human resource productivity and million, 90 percent ($61.2 million) of which was spent 68 See the World Bank’s 2015 public expenditure review in Haiti for further discussion (World Bank 2016a). 69 For example, there is a correlation between the density of medical staff and the proportion of births attended by skilled personnel in LICs (correlation: 0.66, P < .001). And yet, Haiti is doing very poorly (37 percent of births are attended by skilled personnel) compared with other LICs such as Liberia (61 percent), Mali, (57 percent), and Mozambique (54 percent), although it has a higher density of medical staff than these countries: the density of medical staff per 10,000 inhabitants is 2.8 in Liberia, 5.1 in Mali, 4.5 in Mozambique (WHO 2015), and 9.5 in Haiti (World Bank staff estimates based on SPA 2014). 70 A common approach is to benchmark the salaries of health workers against the average gross national income (GNI) per capita (McCoy et al. 2009). In Haiti in 2013, public sector medical doctors earned 13.0 times as much as the GNI per capita, whereas public sector nurses earned about 7.1 times the GNI per capita. By comparison, nurses make 8.2 times the GNI per capita in Burkina Faso and 7.9 times the GNI per capita in Rwanda, and thus a slightly higher salary than in Haiti. Medical doctors earn 13.5 times the GNI per capita in Burkina Faso, again comparable with Haiti, where the ratio was 13.0. BETTER SPENDING, BETTER CARE: 72 A LOOK AT HAITI’S HEALTH FINANCING BOX 6.2 Preliminary Results of Results-Based Financing (RBF) Pilot Since 2013, the MSPP has had a contracting unit in charge of implementing RBF at the national level. The unit, with support from the World Bank and the USAID/LMG (Leadership, Management, and Governance) project, has been piloting RBF in the North-East department since August 2014. In 2016 the contracting unit began to implement RBF in 80 sites across seven departments. RBF funds will be used in two ways: (1) at least 30 percent of RFB funds will improve the functioning and quality of services (training, advanced strategies at the com- munity level, and small investments), and (2) up to 70 percent of the total funds will be used to pay premiums for individual performances. The results of the pilot in the North-East are promising. Service coverage increased for almost all indicators from August 2014 to December 2015 (table B6.2.1). Diarrhea is the second cause of deaths among under children under 5, and the number of treated chil- dren with diarrhea increased by 500 percent. Only antenatal care decreased during the same time frame. It could be that medical staff targeted specific services and did not manage to improve all services. Although higher reporting must have favored those positive results, the results are so large that it is likely that RBF played a role in this improvement. The ongoing impact evaluation of the RBF program will study the impact of the program with more scientific rigor. TABLE B6.2.1: Health Coverage in Six Health Facilities from RBF Pilot in North-East: Haiti, 2014 and 2016 Aug.–Oct. 2014 Oct.–Dec. 2016 % change Diarrhea cases treated 22 132 500% Referral to next level of care 8 100 1,150% Nutritional screening children 6–59 months 987 1,494 51% Complete immunization of children <12 77 136 77% months Institutional delivery 18 26 44% Fourth antenatal care visit completed 40 23 –43% Home postnatal visit during days 0–3 77 201 161% Sources: SPA 2014; World Bank 2016a. on salary payments to staff. If a 50 percent productivity Linking financing for individual staff and facilities loss is assumed for staff being paid for full-time produc- to the production of results through results-based tivity,71 then half of Haiti’s annual expenditure of $61.2 financing (RBF) mechanisms72 is one way to million, or about $30.6 million of its operating budget strengthen accountability for results and increase for health, is being wasted every year. Furthermore, a human resource productivity. Evidence shows that loss of about $30.6 million represents roughly 4.7 per- health sector interventions based on RBF mecha- cent of Haiti’s THE of $650 million (World Bank esti- nisms increase human resource productivity and ac- mates based on NHA 2012–13). A study done by the countability in service delivery (Fritsche, Soeters, and U.S. Agency for International Development (USAID) es- Meessen 2014). Initial piloting of the national RBF timates that absenteeism alone costs the public sector program in Haiti shows very promising results (box $3 million (USAID forthcoming) on a yearly basis. 6.2). Even though the results are preliminary, this payments mechanism demonstrates a measurable 71 A USAID, World Bank, and MSPP study conducted in 2013 in the North-West, North-East, and Center departments revealed that medical staff (a sample of 200 medical doctors, nurses, and aid nurses) worked on average four hours a day and thus 50 percent of the time. 72 Results-based financing (RBF) is defined as the transfer of money or material goods to a recipient conditional on measurable action taken or the realization of a predetermined performance target (Eichler and Levine 2009). Applied to the health sector, RBF has helped to improve the use of maternal and child health services and key functions of the health system in several low-income countries. Experience shows that this approach (1) gives the providers of health services clear signals about the government’s priorities and ensures that institutions continue to place enough emphasis on prevention interventions and on the poor; (2) enables an off-center focus on inputs to the production of tangible results; (3) strengthens monitoring and evaluation systems; (4) strengthens the decentral- ization of decision making; and (5) increases productivity and accountability in service delivery. All these qualities are essential for both improving outcomes for maternal and child health and strengthening the health system (Fritsche, Soeters, and Meessen 2014). Chapter 6 • EFFICIENCY ANALYSIS 73 Administrative Personnel as Share of Total Number of Staff of Primary Health Care Facilities: FIGURE 6.9: Haiti and Selected Countries, Various Years 50 40 40 33 30 29 30 Percent 20 20 10 0 Haiti Afghanistan Rwanda Liberia (dispensaries) Liberia (CSLs/CALs) Sources: MSPP 2014a; World Bank 2016a; Haiti: Cros and Zeng 2014; Afghanistan: Ministry of Health, Afghanistan 2003; Rwanda: Collins et al. 2011; Liberia: Wang, Young and Connor 2009. Note: CALs = centres de santé avec lit (health centers with bed); CSLs = centres de santé sans lit (health centers without bed). impact on human resource productivity in the Haitian resource management, quality assurance, and financial context. incentives (Abzalova et al.1998; Barber, Bonnet, and Bekedam 2004). Under Haiti’s current national RBF mod- The scale-up of RBF in Haiti should be accompa- el, health facilities receive results-based funding from the nied by decentralization of key human resource central level. Through formal contracting, the DDS ties decisions and improved human resource manage- the receipt of payment to the delivery of specified health ment practices. Recruitment practices and large re- service delivery outputs–the basis of the RBF financial in- source investments are not fully decentralized, and centive mechanism. Furthermore, the DDS is obligated health facilities have little to say in this area of decision to uphold regular performance monitoring to confirm making, which requires approval from the departmen- health facility performance. This facilitates increased au- tal health directorates (directions departementales tonomy for health facilities in the use of financial resourc- sanitaires, DDSs) or at the NGO level. Although med- es, and it supports motivation by engaging medical staff ical staff members are appointed by the departmen- more intensively. Through the national RBF initiative, Haiti tal health directorates, management candidates are has already increased facility autonomy levels in several appointed by the MSPP at the central level. However, departments. To scale up this model, the RBF contracting health facilities do manage and spend revenues gener- unit and the MSPP human resources and planning direc- ated by on-site user fee collection. torates must reach a consensus on current health human resource reforms and move forward collaboratively. The existing health facility performance manage- ment systems are weak. On average, only 30 per- Poor working conditions73 lead to low satisfaction cent of facilities hold management meetings, and only and productivity by medical staff. Increasing the 70 percent of health facilities receive supervisory visits nonsalary operational budget would improve service from staff at the departmental level (SPA 2014). In ad- readiness and the overall performance of health work- dition, clinical guidelines are rarely available (World Bank ers. In a survey of medical staff and health managers in 2015a), which may contribute to low staff accountabil- three departments, lack of medicines and equipment ity in quality of care. Global evidence shows that health and limited opportunities for training were the main facility performance is increased by health reforms that reasons for poor motivation of medical staff and non- combine facility autonomy, staff accountability, human functioning of health facilities (World Bank, USAID, and 73 Working conditions include availability of equipment and supplies, infrastructure, support services, regulations at work and lines of authority, as well as decision making. These are important determinants of job satisfaction and the performance of health facilities (Dieleman and Harnmeijer 2006). BETTER SPENDING, BETTER CARE: 74 A LOOK AT HAITI’S HEALTH FINANCING BOX 6.3 The Supply Chain for Drugs in Haiti Haiti has a three-tiered supply chain system. Health facil- FIGURE B6.3.1: Supply Chain for Health Care Facilities ities retrieve products from a regional warehouse (Centre in Haiti Départemental d’Approvisionnement en Intrants, CDAI). CDAIs Frequency stock those products that are specifically related to health pro- of Resupply Payment grams sponsored in-country and distributed free of charge by the Programme des Medicaments Essentiels1 (PROMESS, the es- Aid Agency Pick-upf Annual Day of sential drugs program managed by PAHO) from Port-au-Prince. e.g. UNICEF PROMESS sends CDAIs and health facilities also procure additional products from products donors and private wholesalers. Figure B6.3.1 illustrates this to CDAI Pick-upf Months process. The boxes in blue represent the public sector, the boxes Day of 3 to 4 Every Private in orange the foreign aid sector, and the box in purple the pri- Wholesaler vate for-profit sector. The direction of the arrows indicates which group is responsible for transportation. For example, a dispensa- Pick-upf ry goes to a wholesaler to pick up products. Exceptions include Day of Months Every when programs run by aid agencies drop off stock at regional Aid Agency stores either regularly or as part of a campaign for a specific CDAI e.g. Caritas health program such as family planning. Some health facilities go directly to PROMESS for stock. The private for-profit sector plays a key role in product availability in Haiti. There are three registered local manufacturers, 35 registered wholesalers based Dispensary, CSL, CAL in Port-au-Prince, and 129 authorized private for-profit pharma- cies (PAHO 2012). Over 400 unauthorized pharmacies also op- erate in the metropolitan area (PAHO 2012). There is as well a Patient thriving informal (and thus unregulated) market for both the pro- Source: Johnson, Laverty, and Sjoblom 2014, unless otherwise indicated curement and sale of pharmaceuticals. Note: CAL = centre de santé avec lit (heath center with bed) ; CDAI = Centre Départemental d’Approvisionnement en Intrants; CSL = centre de santé sans lit (health center without bed); PROMESS = Programme des Medicaments Essentiels. MSPP 2013). Although the facility equipment index are low qualified, which is high based on international and an improved source of water are not significantly benchmarks (AEDES Consortium 2016). In public fa- associated with TE scores in Haiti’s PHC sector, positive cilities, administrative staff represent nearly half of the associations exist between the drug index in CALs and workforce (MSPP 2014a). Another study found that electricity in CSLs (see table C.4 in appendix C). As noted administrative personnel represent 40 percent of total earlier, service readiness is very low in Haiti. Improving institutional personnel at the PHC level (Cros and Zeng it would require an increase in the nonsalary recurrent 2014). This ratio also seems high in comparison with budget to allow for the availability of the other inputs those for the LICs (figure 6.9). such as medicines, equipment, and medical supplies that are needed to produce health services. Over the The availability of medicines could also be im- last decade, 90 percent of Haiti’s operational budget proved by providing better supply chain manage- has been allocated to staff salaries (BOOST 2016), ment. Seven different parallel supply chains were es- which is very high compared with the allocation in LICs tablished after the 2010 earthquake, but in recent years (World Bank 2016a). One way to free up resources for they have been slowly consolidated into two main sup- complementary inputs is to tackle the high level of sup- ply chains. However, more could be done to improve port staff on the payroll. At the University Hospital of supply chain management (see box 6.3). Considerable the State of Haiti (Hôpital de l’Université d’État d’Haï- savings could result from improving the coordination ti, HUEH), 87 percent of the operational budget is used of the distribution network and focusing on improving to pay staff (BOOST 2016), and 22 percent of the staff the last-mile distribution.74 One possible approach is 74 Last mile is a term used in supply chain management planning to describe the movement of goods from a transportation hub to a final destination at the facility level. Chapter 6 • EFFICIENCY ANALYSIS 75 to outsource distribution to local transport companies, of maternal and child health services (Eichler, Auxila, which has already been successfully piloted in Haiti. and Pollock 2001; Eichler and Levine 2009; Zeng et al. Some medical products are not distributed by develop- 2013). However, such an approach would require ad- ment partners, and to fill this gap in inventory, health ditional funding to cover the per diems of clinical staff facilities have to rely on a regular cash flow to facilitate as well as transportation. Second, community health outside procurement. However, persistent deficits in agents can play an instrumental role by referring pa- this cash flow necessitate more frequent, smaller trips tients seen at the community level to higher levels of to medical supply vendors and lead to higher distribu- care. Currently, discussions are under way on expand- tion costs overall. Improper storage management and ing the scope of practice for community health work- weak information systems at the Program of Essential ers. Third, as discussed in the access section, removing Drugs (Programme des Medicaments Essentiels, user fees for selected services will help boost the de- PROMESS)75 also reduce the availability of medicines mand for services for the poor. at the facility level. In fact, anecdotal evidence seems to show that drugs originally destined for subsidized Most hospitals have low levels of productivity (ta- distribution at public facilities are intercepted at the ble 6.5). Strengthening the referral system, providing regional depots (Centre Departementaux d’Approvi- subsidized transportation to poor patients following sionnement en Intrants, CDAI), syphoned off, and sold referrals, and subsidizing service fees could help in- to private sector pharmacies, which subsequently sell crease the demand for hospital services. Increasing the same product at higher prices. Stock-outs at the the number of admissions is instrumental to improving CDAI then force public facilities to purchase drugs at efficiency at the hospital level. However, the mecha- market prices. Most important, this practice reduces nisms used for this purpose must not allow patients to patient access to subsidized medicines. skip the primary care level. Three mechanisms will in- crease the demand for hospital services. The first is to strengthen the referral system76 so that patients are re- Demand Factors Impeding Technical ferred to community hospitals from the primary health Efficiency/Performance care level or from community hospitals to departmen- tal or university hospitals. Currently, the RBF program is providing incentives for referrals at the dispensary, Although policies addressing the supply of health health center, and community referral hospital levels. care are key to significantly improving the perfor- Therefore, behavioral change is likely to occur at the mance of health care in Haiti, mechanisms should supply level. At the demand level, queuing systems be put in place to stimulate the demand for health should be redesigned to separate referred patients services. These may include mobile clinics, community from nonreferred patients in order to fast-track refer- agents, and the removal of user fees for selected ser- rals and increase patient awareness of the rationale vices for poor populations, in particular. Health cen- underlying treatment referrals (Jamison et al. 2006). ters, especially those in rural areas, have lower tech- Second, it is well known that transport is the key de- nical efficiency scores because poorer people live in terrent to access to hospital services for the poor in these areas and they cannot afford to visit health facili- LICs (Kloos 1990; Martey et al. 1998) and in Haiti over- ties. Improving the efficiency of PHC facilities would re- all (DHS 2012). Policies should enhance transportation quire mechanisms that stimulate demand. First, mobile systems from health centers to hospitals for referred health clinics could be organized by CSLs and CALs be- services. Such a step will be more cost-effective at the cause two-thirds of mobile clinic and community ser- community or departmental hospital level such as for vices users are poor. Such strategies used in the Health children with complicated cases of diarrhea, patients for the Development and Stability of Haiti (Santé Pour with acute respiratory diseases, and women undergo- le Development et La Stabilité d’Haiti, SDSH) program ing C-sections. The third mechanism is the removal of in Haiti have contributed to increasing the utilization user fees for key selected (and referred) services at the 75 The public supply chain in charge of essential drugs (and managed by the Pan American Health Organization, PAHO). 76 A referral system ensures that patients can receive appropriate, high-quality care for their condition at the lowest cost and in the closest facility possible (Jamison et al. 2006). BETTER SPENDING, BETTER CARE: 76 A LOOK AT HAITI’S HEALTH FINANCING TABLE 6.5: Projected Changes in Admissions and Consultations across Facility Type, Haiti Current no. of Projected no. Current no. of Projected no. of % difference % difference admissions of admissions consultations consultations Sample 1,176 4,524 285% 1,6908 42,210 150% Facility category University hospital 7,716 25,878 235% 56,916 138,827 144% District hospital 2,868 16,187 464% 18,624 74,141 298% Community referral 1,086 3,417 215% 15,168 37,636 148% hospital Small hospital 816 2,865 251% 9,090 33,076 264% Source: World Bank staff estimates, 2016. Note: This table is the result of the DEA hospital analysis based on SPA data set, 2013. This table highlights by how much hospitals would need to increase their pro- ductivity (number of admissions and consultation) using their current resources to be efficient. For example, community referral hospitals (hôpitaux communautaire de référence, HCRs), with their current resources, should be able to triple their number of admissions and double their number of consultations to remain efficient. In other words, HCRs would have to increase their admissions by 215 percent and consultations by 148 percent to reach efficiency. hospital level. Financial losses from user fee removal 2014).77 Other options include voucher and health eq- for selected services could be financed by cross-sub- uity funds. For example, in Cambodia voucher mecha- sidization between the rich and the poor as suggest- nisms provide free C-sections in public hospitals for the ed for Mirebalais referral hospitals (Baruwa and Meline poor (Noirhomme et al. 2007). 77 Baruwa at al. (2015) suggested cross-subsidizing services at the University Hospital of Mirebalais –such as radiology, physiotherapy, and surgical services–be- tween those who can afford them and the poor. Such a policy is valid and should be applied to referred services only to deter the poor from consulting a university hospital for a service that could be treated at the PHC or district hospital level. Chapter 6 • EFFICIENCY ANALYSIS 77 CHAPTER 7 MAIN FINDINGS AND RECOMMENDATIONS Seven prioritized strategic shifts emerged from the key findings of this report, and they are presented in this chap- ter in tandem with recommendations on how to take the next steps toward accelerating and sustaining progress in achieving universal health care (UHC) in Haiti. Shift 1: Prioritize Primary Health Care. Realign resources from hospital to primary health care and cost and prioritize the existing Health Master Plan (Plan Directeur de Santé, PDS) to guide future financing. MAIN FINDINGS Strengthening the delivery of preventive and primary health care (PHC) services would help address the three leading causes of disabil- ity-adjusted life years (DALYs) in Haiti, but only 19 percent of the total health expenditure (THE) is directed to preventive care and more than half (54 percent) to curative care. Although best practice indicates that funding allocations should be tailored to the needs of the population, allocations made at the departmental level in Haiti are instead based on historically set values. Haiti has 1.4 community hospitals for every 150,000 inhabitants. By contrast, it has only 0.3 dispensaries for every 10,000 inhabitants. Compared with the averages for low-income countries (LICs), the facility density in Haiti is much higher for hospitals and lower for dispensaries. Furthermore, dispensaries are disadvantaged by their low service readiness overall. RECOMMENDATIONS A core step in implementation of this strategic shift is to prioritize and cost the Plan Directeur with a focus on primary health care. This prioritization and cost should be according to Haiti’s disease burden and take into account the services included in the essential package of health services (EPHS). Based on these priorities, the Ministry of Public Health and Population (Ministère de la Santé Publique et de la Population, MSPP), with support from development partners, would build an “investment case” that would guide the investments as well as the technical and financial contributions of both the government and development partners to the health sector. The MSPP should oversee implementation of the framework developed in the investment case, and the development partners would play supportive roles in the financial and technical components. The MSPP should adjust the resource allocation methodology to incorporate a formula derived from the health and socioeconomic needs of the poor, pertinent health system characteristics, updated data on the disease burden, and population size. BETTER SPENDING, BETTER CARE: 78 A LOOK AT HAITI’S HEALTH FINANCING Shift 2: Increase Equitable Access to Quality Care. Update and implement a facility mapping tool by reclassifying health facilities to enhance service readiness and facilitate a practical referral network. MAIN FINDINGS Service readiness is very low both in absolute terms and against international standards. Only 32 percent of health facilities have es- sential medicines, and only 31 percent of health facilities have the basic medical equipment. There is almost no budget to pay for drugs and running costs at the health facility level because the MSPP assigns 90 percent of its operating budget to staffing costs. In certain areas, there are no health facilities or services available, whereas in others duplications exist. In addition, some health facilities do not necessarily meet the minimum criteria for their level of service (that is, their nomenclature does not reflect the actual services provided). Thus a reclassification is needed for certain facilities. The referral system is not functional; only 6 percent of referrals are carried out properly. RECOMMENDATIONS The MSPP should develop a facility mapping tool to (1) identify the existing public and private facilities; (2) establish their service read- iness–mostly in terms of staff and inputs; and (3) determine the population coverage of each facility. The first step would build on the existing carte sanitaire. The findings of such a mapping tool will identify service gaps or redundancies and trigger a recategorization of certain facilities. The MSPP should then systematically confirm that all facilities included in the referral network meet minimum criteria in terms of service readiness, which will vary by type of facility. Taking into consideration the investment priorities that would be defined in the Plan Directeur (see Shift 1), certain inefficient commu- nity referral hospitals could be transformed into health centers with increased operational expenditures. In other cases, certain facilities could be converted into primary health care units, or upgraded to hospitals, or given special attention to ensure service readiness. Merged facilities would be better equipped with drugs and medical equipment. The recategorization of facilities should align with the definition of a coherent and effective referral system. That would imply, among other things, considering strategies such as subsidized transportation options for patients to hospitals. In this process, it is critical to agree on a minimum package of services that will be financed and provided at the primary level. Chapter 7 • MAIN FINDINGS AND RECOMMENDATIONS 79 Shift 3: Spend More Wisely on Hospitals. Suspend new hospital construction until the existing infrastructure can be mapped and a hospital licensing program has been developed. Development partners should finance technical assistance for hospitals. MAIN FINDINGS External financing, which was especially high after the 2010 earthquake, fueled hospital (re)construction. However, these capital investments were not accompanied by plans on how to sustain service delivery in the units. Consequently, the MSPP is now struggling to deal with rising operational costs. Hospital construction has not been aligned with the needs or gaps reflected in the carte sanitaire but rather built in an opportunistic fashion. RECOMMENDATIONS The MSPP should consider placing a moratorium on the construction of new hospitals that would begin immediately and remain until a hospital mapping exercise is completed and a hospital licensing program has been established. Reconstruction linked to emergency situations such as that launched by Hurricane Matthew should be allowed. A licensing agency, either managed by the MSPP or outsourced to a third party, should license facilities that meet the minimum criteria. Those that do not should be downgraded or closed down. At the hospital level, this will allow the MSPP to rationalize the number of hospitals and potentially convert some community hospitals to lower-level facilities. Private for-profit facilities should be included in the licensing program. To increase the MSPP’s oversight of private for-profit/nongovernmental organization (NGO) facilities, formal contracts should be under- taken with licensed facilities. Haiti should also encourage its development partners to fund a technical assistance that would guide the development of business plans, which could strengthen the financial sustainability of imminent hospital acquisitions (or programs) by the government. New sources of revenues for hospitals such as luxury wards or contributions from wealthy individuals inside and outside of Haiti as well as diaspora and religious organizations should be explored. Shift 4: Improve Technical Efficiency at PHC Level. Value-for-money in service delivery should be increased by reforming human resources, having better availability and use of inputs and serving more patients, especially at PHC level. MAIN FINDINGS All categories of health facilities have low productivity. Of the low-income countries, Haiti displays one of the lowest technical efficiency scores for all health facilities. Primary care facilities–dispensaries, health centers without bed (centres de santé sans lit, CSLs), and health centers with bed (centres de santé avec lit, CALs)–are particularly inefficient. Key measures of hospital productivity such as the bed occupancy rate (BOR) show that Haiti fares poorly compared with other countries. Low efficiency is explained in part by the low productivity of medical staff (they produce six consultations per day at the PHC level), absenteeism (which contributes to wasting approximately $3 million a year), and moonlighting (in certain departments medical staff spend a third of their time working in a second job outside their facility but are still paid for full-time work). Haiti does not use its available health sector workforce potential. There appears to be a shortage of qualified midlevel staff such as clinical officers and nursing staff. The low utilization and thus demand for health services are another key factor in low productivity. RECOMMENDATIONS To improve the productivity of human resources (HR), certain HR decisions should be decentralized to make health facilities more ac- countable for results, thereby limiting absenteeism and low productivity issues. The pilot of the results-based financing (RBF) program is showing promising results in terms of increasing the productivity of HR. If it continues to sustain these initial results, the program should be scaled up nationwide. The MSPP should also establish regulatory frameworks to strengthen management of human resourc- es for health, which will help reduce dual practice, absenteeism, and the number of ghost workers. BETTER SPENDING, BETTER CARE: 80 A LOOK AT HAITI’S HEALTH FINANCING To improve the availability and access to subsidized drugs, the last-mile distribution problem should be studied further. Storage man- agement and strengthening of the information system at the regional depot (Centre Départemental d’Approvisionnement en Intrants, CDAI) level to avoid leakages of subsidized products should also be implemented. Nationally pooled procurement of medical equipment and commodities could generate significant cost savings, which could in turn help finance more affordable services for the poorest. Additional work is needed to understand the market conditions for pharmaceutical importation, wholesale arrangements, and distribu- tion. The MSPP and donors should focus on mechanisms to improve the demand for the services of health facilities (see Shift 7), especially dispensaries. Such mechanisms could increase their efficiency and their utilization by the poor. The emphasis should be on public hospitals, which had the lowest BOR, but also because the poorest segments more often seek care at public hospitals than private hospitals. Shift 5: Better Use of External Funding. Put in place an adequately staffed and well-functioning donor coordination unit that pursue donor tracking and transition planning to increase impact and enforce adherence to Plan Directeur. MAIN FINDINGS At least a third of the total health expenditure is financed externally. External financing is particularly fragmented; 90 percent is off-budget and channeled through many different implementers. Budget execution rates for external financing is below 80 percent. Because of the fragmentation, the MSPP has limited control of the uses of external financing. There is no regular, established mechanism whereby donors and MSPP can discuss and coordinate their technical and financial contri- butions to the health sector. The existing mechanisms have not yet yielded results in terms of aligning partners in the implementation of the Plan Directeur. A large share of external financing is emergency aid, which is volatile and has tended to focus on hospital construction. New hospitals are not necessarily what Haiti needs in view of its disease burden and existing health infrastructure. These large infrastructure invest- ments often do not take future operational costs into account from the outset, and the MSPP, with its available resources, cannot cover the running costs of many of the institutions that were built (or rehabilitated) after the earthquake. RECOMMENDATIONS The MSPP should oversee the investments of development partners and seek their support for a costed and prioritized investment case (see Shift 1). One way to do this is to create an adequately staffed and well-functioning donor coordination unit that, among other things, would (1) maintain the national database of cooperation projects, and (2) ensure that transition plans (especially because many donors are withdrawing) match health system needs with the available resources. The MSPP should also enforce the need for develop- ment partners to register with the donor coordination unit. In the shorter term, development partners should begin to pool external financing virtually around the essential package of health services (EPHS). Some partners have begun this process for a limited set of services in the context of the RBF program. In the longer term, key donors should work with the MSPP to strengthen the public financial management (PFM) structures of the MSPP or of entities such as the MSPP’s Project Management Unit (Unite de Gestion du Projet, UGP), which is used currently by several donors. They should also work together to develop a joint PFM manual with the objective of creating enough confidence to channel more resources on-budget through a Sector Wide Approach (SWAp) mechanism. Such a modality would imply a harmonization of pro- cedures for PFM and agreement on the level of per diems, salaries etc. That would dramatically reduce the transaction costs of external financing. The same agreement or memorandum of understanding (MoU) should contain details related to emergency funding. The MoU could develop minimum standards for improving the sustainability of emergency financing such as including requirements that major capital investments (hospitals) be supported by long-term sustainability plans. Chapter 7 • MAIN FINDINGS AND RECOMMENDATIONS 81 Shift 6: Increase resources for health. Leverage greater health financing overall by increasing public health expenditure through better tax collection and more sustainable external financing. MAIN FINDINGS There has been a sharp decline in the general government health expenditure (GGHE) over the last two decades in Haiti. The government is heavily reliant on external funding (donor dependency increased particularly after the 2010 earthquake), but exter- nal financing has decreased sharply in recent years, while out-of-pocket (OOP) expenditures have increased. Recurrent health expenditures for vaccine supplies, salaries of human resources for health, and medical products are largely financed by external resources, which are declining rapidly. Although large efficiency gains can be achieved in the health sector (see Shift 4), more resources are needed to sustain and improve health outcomes in the future. The government needs to start planning to increase domestic financing for health to compensate for the drop in external aid and protect the poor from increased OOP expenditures. It also may need to attract sustainable external financing, at least in the short run. For some time now, vaccines in Haiti have been fully financed by donors, and in this respect Haiti is different from most other LICs which provide some of their own domestic financing for vaccines. But increasingly these donors are finding it difficult to continue to fi- nance vaccines in the country without any cofinancing contribution from the government. It is hence urgent for the government to start allocating some of its own funds to vaccines. Similar arguments apply to other items considered by the government to be essential. RECOMMENDATIONS A strong case should be built for the Ministry of Economy and Finance (MEF) to invest in the health sector. It is essential to show im- proved value-for-money, improve budget execution rates, build trust with the MEF, and explain the vision to accelerate progress toward universal health care. As indicated in the public expenditure review (PER), general tax system reforms should be implemented to increase mobilization of revenue for health and other sectors. The MSPP is in the process of developing a long-term health financing strategy. This work should continue, including implementation of such a strategy. The strategy should, among other things, consider unorthodox ways to raise revenue for the health sector. For example, the following should be considered: Putting in place mechanisms to allow migrants to send remittances directly to prepayment mechanisms for health care or other form of earmarking remittances for the health expenses of the recipient. Remittances accounted for about a fifth of Haiti’s gross domestic product (GDP) in recent years. Having the MSPP engage strategically with wealthy individuals inside and outside of Haiti to finance individual facilities or programs. Exploring an earmarked tax on alcohol and tobacco and determining which segments of the population would or would not benefit from an excise tax, as well as feasibility of implementation. Exploring the possibility of receiving more external financing. For example, Haiti could be selected for the next round of countries that will receive support from the Global Financing Facility (GFF) for every woman and child. The GFF does not just provide additional financing, but, more important, it is a partnership that helps rally external partners around an investment case and work on long-term financing strategies to achieve universal health care. Haiti should ensure that domestic financing for health is spent in a manner that addresses key priorities and is also strategically “smart”, leveraging donor financing for essential items like vaccines. For this to happen, a domestic budget line for vaccines needs to be established and maintained, with significant funding made available under this budget line. BETTER SPENDING, BETTER CARE: 82 A LOOK AT HAITI’S HEALTH FINANCING Shift 7: Increase the Affordability of Health Services for the Poor. The feasibility of removing user fees for selected services or target populations -- children under 5 and pregnant women, especially in rural areas -- should be assessed. MAIN FINDINGS Out-of-pocket expenditures make up 35 percent of Haiti’s total health expenditure. Because of the reduction in external funding and the low level of GGHE, almost all health facilities (93 percent) charge user fees. Con- sequently, households are taking on a growing burden for financing the health system, with increases in OOP expenses and catastroph- ic health expenditures (CHEs) observed in recent years. This situation raises affordability issues, and the poorest are priced out of health care. Sixty-three percent of households in the lowest wealth quintile do not consult a health provider when sick because they cannot afford to do so. The lack of affordability may explain the low utilization rates and its impact on the low productivity. RECOMMENDATIONS: Explore the feasibility of removing user fees associated with the delivery of essential health services. In particular, focus on the removal of fees associated with essential health services for pregnant women and children under 5 years of age. Because the poorest use more mobile clinics and services provided by community health workers, more resources should be allocated to expand and strengthen community care, which should be a key part of the prioritization of PHC (see Shift 1). Mechanisms to increase the affordability of health services for the poorest should be part of the investment case (developed from a prioritized and costed Plan Directeur). These mechanisms would include a transportation voucher program or the revival of the equity fund at the facility level to protect the poorest from the direct and indirect costs of health care. Lifting financial barriers should boost use of health services, which if of quality will increase health outcomes that is the ultimate goal of the health system. Chapter 7 • MAIN FINDINGS AND RECOMMENDATIONS 83 APPENDIX A. Domestic Revenues Domestic revenue mobilization has improved in Haiti. The recent improvement in fiscal revenue has been driv- en by the increase in domestic income and sales tax revenue collected in Port-au-Prince (table A.1). In addition, a new tax instrument was introduced in fiscal 2012. The National Education Fund (National Fond d’Education, FNE), which is financed through taxes on international telephone calls and money transfers, supports the Free and Compulsory Universal Enrollment Program (Programme de Scolarisation Universelle Gratuite et Obligatoire, PSUGO). And yet domestic revenue mobilization remains low compared with that of neighboring and other low-in- come countries. Haiti also still has the second lowest tax-to-gross domestic product (GDP) ratio (13.7 percent) of all countries in the region, and its ratio is only slightly better than those of the low-income countries (LICs)–see pan- el a of figure A.1. In 2012, 24 low- and lower-middle-income countries had tax-to-GDP ratios below 15 percent TABLE A.1: Tax Categories as Share of GDP: Haiti, 2009–15 2009 2010 2011 2012 2013 2014 2015 Fiscal revenue 11.2 11.8 12.8 12.8 12.7 12.5 13.7 Domestic taxes 7.4 7.3 8.1 8.6 8 8.5 8.9 Income taxes (Port-au-Prince only) 2.3 2.2 2.5 3 2.6 2.9 3.1 Domestic taxes in provinces 0.5 0.4 0.4 0.5 0.5 0.7 0.7 Excise taxes 0.7 0.5 0.3 0.3 0.3 0.2 0.8 Sales taxes 3.5 3.2 3.7 3.7 3.7 3.6 3.3 Other taxes (Port-au-Prince only, including 0.5 0.9 1.2 1 1 1.1 1.1 discrepancies) Customs duties (including inspection fees) 3.3 4.3 4.5 4.2 3.9 3.4 4.1 Other (including FNE) 0.4 0.2 0.3 0.1 0.8 0.6 0.6 Sources: Ministry of Economy and Finance, Bank of the Republic of Haiti, International Monetary Fund, and World Bank staff calculations. Note: FNE = Fond National d’Education. BETTER SPENDING, BETTER CARE: 84 A LOOK AT HAITI’S HEALTH FINANCING FIGURE A.1: Tax-to-GDP Ratio, Haiti and Selected Countries a. Tax-to-GDP regional comparis 30 Tax revenues (in percentage of GDP) 25 20 15 10 5 0 Guatemala Haiti Dominican Republic Costa Rica El Salvador Nicaragua Honduras Suriname Belize St. Lucia Dominica Barbados Jamaica and Tobago Trinidad b. . Tax-to-GDP regional comparison (natural log of per capita GDP) 35 30 25 20 15 Haiti 10 5 7 8 9 10 11 12 Natural log of per capita GDP Source: Adapted from World Bank 2016a. (World Bank 2016a), which is an arbitrary but often suggested minimum benchmark.78 Haiti’s tax-to-GDP ratio is 1.07 times higher than that of the LICs, but its GDP per capita is 1.36 times higher than the LIC average. Given its economic status, Haiti should have a higher tax-to-GDP ratio, as reflected in panel b of figure A.1 where Haiti is below the fitted line. If Haiti were to increase its tax-to-GDP ratio to 15 percent, it would increase its fiscal revenue by $18 per capita or 2 percent of GDP (see table A.2). 78 In view of administrative and capacity constraints, tax shares of 20 and 25 percent may be difficult to achieve (Heller 2005, 2006; IMF 2011). APPENDIXES 85 TABLE A.2: Tax Efforts: Haiti and Its Comparators, 2015 Additional tax revenue Additional tax revenue Tax (% of GDP) GDP per capita (US$) per capita if tax collection per capita if tax collection were 15% of GDP (US$) were 20% of GDP (US$) LICs 12.8% $605 $23 $58 LAC 17.0% $9,279 – $427 CAM 15.5% $5,840 – $343 Haiti 13.7% $824 $18 $59 Source: World Bank staff estimates based on World Bank 2016a. Note: – = not available; CAM = Central America and Mexico; LAC = Latin America and the Caribbean; LICs = low-income countries. TABLE A.3: Additional Revenues Levied from Earmarked Taxes on Spirits, Haiti Rum Beer Global pretax sales $12,000,000 $33,377,406 Price elasticity of demand for spirits –1.5 –0.3 Potential tax 25% 25% Proceeds from tax $1,312,500 $6,946,673 Generated revenue per capita $0.12 $0.64 Per capita government health spending $7 $7 Per capita government health spending with tax $7.12 $7.64 Growth rate of per capita government health spending 1.71% 9.14% Sources: World Bank staff estimates based on literature review of the price elasticity of spirits and sales of spirits in Haiti (Josephson and Bode 2013); population estimates for Haiti, 2013: World Bank 2016a; 2013 per capita government health spending for Haiti: WHO 2015. In Haiti, taxes earmarked for the health sector could address negative externalities while raising a sub- stantial amount for the sector. Several countries are using taxes on alcohol and cigarettes to reduce the preva- lence rate of tobacco and alcohol use and to raise revenue for the health sector. Despite the prevalence of tobac- co at 8 percent of the population in Haiti, there is no tax on tobacco. Haiti did sign the World Health Organization Framework Convention on Tobacco Control on July 23, 2003, but it has not yet ratified the convention. The tax rates on spirits are 4 percent for those locally produced and 16 percent for imports. However, none of the taxes levied on spirits are earmarked for the health sector. On average, taxes account for 31 percent of the retail price of cigarettes in LICs and 47 percent in the Latin America and the Caribbean (LAC) region (WHO 2015). In Haiti, a 25 percent tax on alcohol earmarked for health could produce an estimated $8.2 million in revenue a year for health financing based on sales estimates for Prestige beer and Barbancourt rum (Table A.3). The proceeds from such a tax would represent a growth rate of 24 percent in government health spending, or $1.64 per cap- ita (Table A.3). Estimates for revenue are based on sales numbers for selected brands of rum and beer because countrywide data on alcohol sales or use are not readily available. Estimates for tobacco sales are unknown, but a large tobacco company based in Haiti accounted for 97.8 percent of the market share in 2012 (Josephson and Bode 2013). Sin taxes could increase the predictability of financing and affect risky behavior, thereby improving health. Earmarking taxes for health raises technical and political issues that warrant a thorough assessment. Earmarking taxes on tobacco and alcohol for the health sector could be instrumental in raising domestic reve- nues for that sector, especially in Haiti where external financing needs to be replaced by domestic resources in the near future. However, success in levying sin taxes requires sufficient administrative capacity and information as well as alignment from tobacco and alcohol corporations and lobbies. That said, administrative capacity is an issue for several potential tax reforms in Haiti. Other valid concerns are that earmarking taxes could reduce the BETTER SPENDING, BETTER CARE: 86 A LOOK AT HAITI’S HEALTH FINANCING TABLE A.4: Summary of Tax Reforms, Rwanda and El Salvador Rwanda El Salvador 1997: Established Rwanda Revenue Authorities to 1990–2000: Modernized tax and customs operations oversee taxation processes Mid-2000: Improved administration of tax structures 2003: Adopted software system to support customs Administrative reform and policies by taking measures to improve fiscal operations, finance processes, and taxpayer audit compliance, mitigate fraud, and enhance efficiency 2011: Implemented mechanism for electronic tax registration 2001: Widened the VAT base and ultimately replaced existing sales taxes with VAT; implemented a new excise tax 1990–2000: Adopted multiple reforms of existing tax and trade policies–in particular, adopted reforms Mid-2000: Implemented legislation that that widened the tax base and replaced sales tax Policy reform strengthened tax collection processes and instituted with VAT penalties for tax evasion, which bolstered tax compliance 2012: Raised the rates applied to the income tax as well as some excise taxes GGHE as % of GGE 19% 18% Tax-to-GDP ratio (%) Close to 15% in 2015 15.40% In Rwanda, the 2008–2012 Economic Development In El Salvador, implementation of the 2007 Law for and Poverty Reduction Strategy I established explicit the Creation of the National Health System set forth Political support long-term goals for increasing the ratio of the public the government’s aim to expand health care expenditure on health to the total public expenditure coverage and reduce inequities in health outcomes. from 12 to 15 percent by 2012. Sources: Nakamura and Williamson 2015; Heredia-Ortiz 2016. Note: GGE = general government expenditure; GGHE = general government health expenditure; VAT = value added tax. discretionary allocation of the Ministry of Public Health and Population (Ministère de la Santé Publique et de la Population, MSSP)–World Bank (2014). But it also would be regressive because the poor and nonpoor would pay the same amount of taxes. However, the benefit is that youth and poor people tend to respond more than others by not starting to smoke or smoking less (Savedoff and Alwang 2015). Earmarked taxes on luxury goods would be more progressive, but they are known to raise less revenue than direct taxes such as those on income (World Bank 2015e). Thus a more in-depth study should be conducted to assess the political feasibility of such reforms. The LICs and the LAC region have already examined tax compliance and expanding the tax base to im- prove resource mobilization. Lessons learned from Rwanda and El Salvador on revenue mobilization for the health sector pinpoint the importance of administrative and policy reforms as well as political reforms. In both Rwanda and El Salvador, high political support, tax compliance, and reforms widening the tax base were key driv- ers for successful tax reforms and broader resource mobilization for the health sector (table A.4). Levying taxes on tobacco and alcohol is a promising way to raise additional revenue for the health sec- tor, but it will require further analysis (table A.5). Replacing the tax on turnover with a value-added tax (VAT) is another option adopted by several low- and middle-income countries such as Rwanda and El Salvador to increase domestic resource mobilization. Tax reform efforts are under way in Haiti. Taxes on luxury goods such as commercial airline flights may be more difficult to implement than taxes on alcohol and tobacco products. Unlike the latter, taxes on luxury goods do not address negative externalities, which can help generate public APPENDIXES 87 TABLE A.5: Summary of Potential Tax Options to Expand Revenue in Haiti Option Pros Cons Feasibility in other countries Haiti Would increase domestic Justified as tackling a Would be regressive spending by 23 percent negative externality Taxes on Argentina, Colombia, El Needed to initiate dialogues Would encourage healthy Would be opposed by tobacco and Salvador, Guatemala, Jamaica, with “business families” lifestyle lobbies alcohol Madagascar, Nepal, Panama owning alcohol companies Would require administrative capacity Ghana (2.5%); VAT covers Would be regressive two-thirds of NHIA revenue. Replacement of TCA by VAT Many Initiated in in El Salvador and VAT is under waya (not Would require Rwanda but not earmarked for earmarked for health) administrative capacity health Cameroon, Chile, Republic of Perhaps taxes on Congo, France, Madagascar, commercial airline flights, Luxury goods Would be limited Would be progressive Mali, Mauritius, Niger, and but this is more difficult to Republic of Korea justify as a tax tied to health Taxes on money Provides cheap, stable Democratic Republic of Congo, transfers and Already used in the source of revenue for Gabon, Ghana, Philippines, mobile devices education sector government Senegal, Uganda and services Source: Cotlear et al. 2015. Note: NIHA = National Health Insurance Authority; TCA = taxe sur le chiffre d’affaires (turnover tax); VAT = value added tax. a. Because deductibility is applied to consecutive sales, including input, the turnover tax erodes the competitiveness of Haitian firms. Technical work is currently under way to determine the transition of the TCA toward a regular VAT (such as by removing deductibility restrictions, building a sound refund administration, or introducing a zero rate for exporters) and a schedule for its implementation. According to a World Bank analysis (World Bank 2016a) of nine low- and lower-income countries, tax revenue as a share of GDP increased from about 13 to 14.6 percent of GDP within three years after introduction of the VAT. and political support for tax reform. Several LICs have considered levying taxes on money transfers and mobile devices and services. However, Haiti already administers such taxes to subsidize services within the education sector, precluding the application of these taxes within the health sector. Finally, Haiti could also raise tax reve- nues by expanding the personal and corporate income tax base. For example, reforming the tax bracket struc- ture applied to income levels in the 91st percentile and up would effectively raise the overall personal income tax rate by 12 percent (World Bank 2016a). BETTER SPENDING, BETTER CARE: 88 A LOOK AT HAITI’S HEALTH FINANCING Statistical Analysis of Access to Health Care Services TABLE B.1: Determinants of Health-Seeking Behavior: Haiti, 2013 Dependent variable: consulted a health Standard Coefficient Odds ratio P > |z| [95% confidence interval] provider when sick (yes=1; no=0) error Area (vs. rural) Urban –0.1593 0.8527 0.0700 0.0520 0.7260 1.0015 Department (vs. Artibonite) Center 0.1601 1.1736 0.1344 0.1620 0.9376 1.4690 Grand’Anse –0.6211 0.5373 0.0784 0.0000 0.4037 0.7152 Nippes –0.9810 0.3749 0.0648 0.0000 0.2672 0.5262 North –1.2403 0.2893 0.0437 0.0000 0.2152 0.3889 North-East –2.0205 0.1326 0.0310 0.0000 0.0838 0.2098 North-West –1.7064 0.1815 0.0332 0.0000 0.1268 0.2598 West –1.3518 0.2588 0.0259 0.0000 0.2127 0.3148 South –0.4592 0.6318 0.0837 0.0010 0.4873 0.8191 South-East –1.0375 0.3543 0.0553 0.0000 0.2610 0.4810 No. of elderly in household (vs. no elderly) One 65 + 0.3126 1.3670 0.1124 0.0000 1.1635 1.6061 Two or more 65+ 0.4245 1.5288 0.2156 0.0030 1.1597 2.0155 No. of children in household (vs. no child) One child <5 0.2610 1.2982 0.0935 0.0000 1.1273 1.4951 Two or more children <5 0.4065 1.5016 0.1371 0.0000 1.2555 1.7958 Gender of household head (vs. man) Female 0.0532 1.0546 0.0707 0.4270 0.9248 1.2026 Employment status of head of household (vs. employed) Unemployed –0.0022 0.9978 0.1033 0.9830 0.8145 1.2223 Inactive 0.4526 1.5725 0.1417 0.0000 1.3178 1.8763 Educational level of head of household (vs. no education) Incomplete primary 0.2427 1.2747 0.1084 0.0040 1.0790 1.5059 Completed primary, secondary incomplete 0.1008 1.1061 0.1076 0.3000 0.9141 1.3384 Completed primary and secondary 0.1420 1.1526 0.1110 0.1400 0.9544 1.3920 Insurance status of head of household (vs. no insurance) Insured 1.2459 3.4643 0.6286 0.0000 2.4276 4.9438 Expenditure quintile (vs. 1st quintile) 2nd quintile 0.2364 1.2666 0.1429 0.0360 1.0154 1.5800 3rd quintile 0.6372 1.8912 0.2090 0.0000 1.5228 2.3487 4th quintile 0.7625 2.1437 0.2485 0.0000 1.7080 2.6906 5th quintile 0.8496 2.3386 0.2937 0.0000 1.8283 2.9914 Constant –2.025382 n.a. 0.1371 0.0000 –2.2941 –1.7567 Sources: World Bank estimates based on ECVMAS 2013. Note: n.a. = not applicable. Quintile 1 is the poorest and quintile 5 the wealthiest. Number of observations = 10,879; Prob > chi2 = 0.0000; Pseudo R-squared = 0.0830. APPENDIXES 89 TABLE B.2: Determinants of Catastrophic Health Expenditures (CHEs): Haiti, 2013 Standard [95% confidence Determinant Coefficient. Odds ratio P > |z| error interval] Area (vs. rural) Urban –0.5074 0.6021 0.1360 0.0250 0.3868 0.9372 Department (vs. Artibonite) Center –1.4308 0.2391 0.0738 0.0000 0.1305 0.4380 Grand’Anse –2.1367 0.1180 0.0519 0.0000 0.0499 0.2794 Nippes –1.1891 0.3045 0.1639 0.0270 0.1061 0.8743 North –2.0607 0.1274 0.0731 0.0000 0.0413 0.3924 North-East –0.5367 0.5847 0.3320 0.3450 0.1921 1.7795 North-West 0.2720 1.3126 0.5106 0.4840 0.6124 2.8137 West –1.2744 0.2796 0.0728 0.0000 0.1678 0.4659 South –1.2389 0.2897 0.1100 0.0010 0.1377 0.6096 South-East –1.4518 0.2341 0.0992 0.0010 0.1021 0.5370 No. of elderly in household (vs. no elderly) One 65 + 0.4912 1.6342 0.3659 0.0280 1.0538 2.5344 Two or more 65+ 0.0652 1.0674 0.4061 0.8640 0.5064 2.2500 No. of children in household (vs. no child) One child <5 0.1304 1.1393 0.2401 0.5360 0.7538 1.7219 Two or more children <5 0.5685 1.7656 0.4227 0.0180 1.1044 2.8228 Gender of household head (vs. man) Female –0.3017 0.7396 0.1355 0.1000 0.5164 1.0592 Employment status of head of household (vs. employed) Unemployed 0.6385 1.8937 0.4937 0.0140 1.1361 3.1565 Inactive 0.7792 2.1797 0.5178 0.0010 1.3682 3.4723 Educational level of head of household (vs. no education) Incomplete primary 0.7376 2.0910 0.4721 0.0010 1.3433 3.2549 Completed primary, secondary incomplete –0.1063 0.8991 0.2510 0.7030 0.5202 1.5540 Completed primary and secondary 0.1199 1.1274 0.2953 0.6470 0.6747 1.8839 Insurance status of head of household (vs. no insurance) Insured 0.1025 1.1079 0.5044 0.8220 0.4539 2.7043 Expenditure quintile (vs. 1st quintile) 2nd quintile 0.2116 1.2357 0.4306 0.5440 0.6241 2.4465 3rd quintile 0.8907 2.4368 0.7860 0.0060 1.2950 4.5854 4th quintile 0.2028 1.2248 0.4268 0.5610 0.6187 2.4247 5th quintile 1.0561 2.8751 1.0374 0.0030 1.4175 5.8315 Went to a hospital (vs. did not go to a hospital) 1.1947 3.3026 0.8665 0.0000 1.9748 5.5232 Health problem type (vs. fever/malaria) Diarrhea –0.6076 0.5447 0.2824 0.2410 0.1971 1.5049 BETTER SPENDING, BETTER CARE: 90 A LOOK AT HAITI’S HEALTH FINANCING Standard [95% confidence Determinant Coefficient. Odds ratio P > |z| error interval] Accident 0.3443 1.4109 0.6558 0.4590 0.5674 3.5089 Dental problem –0.0829 0.9204 0.7367 0.9170 0.1917 4.4183 Skin problem 0.6170 1.8534 0.8235 0.1650 0.7758 4.4277 Eye problem 1.0598 2.8858 1.2047 0.0110 1.2733 6.5404 Hypertension 0.0023 1.0023 0.3757 0.9950 0.4808 2.0896 Typhoid fever –0.4008 0.6698 0.3627 0.4590 0.2317 1.9360 Ulcers –0.2282 0.7959 0.5173 0.7250 0.2227 2.8448 Ear, nose, throat problems 1.5505 4.7139 4.1184 0.0760 0.8506 26.1246 Diabetes –0.8892 0.4110 0.4471 0.4140 0.0487 3.4662 Meningitis 0.2885 1.3344 1.4675 0.7930 0.1546 11.5183 Pregnancy –0.3080 0.7349 0.3608 0.5300 0.2807 1.9239 Other –0.2520 0.7772 0.1637 0.2320 0.5143 1.1745 Health facility type (vs. public dispensary) Public hospital 0.2498 1.2838 0.3411 0.3470 0.7626 2.1610 Community health center –0.2431 0.7842 0.4288 0.6570 0.2685 2.2901 Traditional healer/provider 0.7560 2.1297 0.8029 0.0450 1.0172 4.4588 Private dispensary –0.1419 0.8677 0.2796 0.6600 0.4614 1.6318 Private clinic/polyclinic 0.9976 2.7117 0.7677 0.0000 1.5569 4.7231 Pharmacist/optometrist 0.3763 1.4569 0.6305 0.3850 0.6239 3.4024 Ambulatory provider –0.4544 0.6348 0.2809 0.3040 0.2667 1.5111 Other –0.5810 0.5593 0.2975 0.2750 0.1972 1.5865 Constant –2.2928 n.a. 0.4445 0 –3.1642 –1.4215 Sources: World Bank estimates based on ECVMAS 2013. Note: n.a. = not applicable. Quintile 1 is the poorest and quintile 5 the wealthiest. Number of observations = 1,704; Prob > chi2 = 0.0000; Pseudo R-squared = 0.1762. TABLE B.3: Routine User Fees, by Ownership and Facility Type: Haiti, 2013 Yes No Total % Ownership Government/public 320 22 342 94% NGO and mission/faith-based facilities 321 28 349 92% Private for-profit 207 7 214 97% Facility type All hospitals 109 12 121 90% Health centers 398 28 426 93% Dispensaries 341 17 358 95% Total 848 57 905 94% Source: World Bank estimates based on SPA 2013. Note: Health centers include health centers with bed (centres de santé avec lit, CALs) and health centers without bed (centres de santé sans lit, CSLs). All hospitals include community referral hospitals (hôpitaux communautaire de référence, HCRs), hospitals, departmental hospitals, and university hospitals. NGO = nongovern- mental organization. APPENDIXES 91 C. Methodology and Approach to Analysis of Health Service Efficiency in Haiti Sample of the Macro-Costing Hospital Study: Haiti, 2016 In Haiti, the department health directorates (DDSs) and the Ministry of Public Health monitor information on hospital statistics at departmental and central levels. However, the quality is usually not double-checked. It is only recently that MSPP’s Planning and Evaluation Unit (Unité de Planification et d’Evaluation, UPE) began training data clerks at the hospital level on hospital statistics. As a result, the study team decided to collect hospital statistics on a representative sample of hospitals. In view of Haiti’s budget constraints and the fact that 87 percent of hospitals are community referral hospitals (hôpitaux communautaires de référence, HCRs) or small hospitals (SHs), the team randomly selected a sample of 22 HCRs and SHs across the 10 departments based on their total number of HCRs and SHs (proportion-to-size sampling strategy). Thus there were more HCRs and SHs from the West. There were data limitations because discharge data were missing in a few hospitals and therefore was not input when the team estimated the average length of stay. In addition, both hospital inpatient and length of stay data were missing for several months in a few hospitals. As a result, both the bed occupancy rate and the aver- age length of stay were annualized based on the available data. Data were collected across the 22 hospitals from February to April 2016. However, the data could not be fully collected in two hospitals. TABLE C.1: Descriptive Statistics and Technical Efficiency Scores, Primary Health Care Level: Haiti, 2013 CAL (min-max) CSL (min-max) Dispensary (min-max) Sample 72 265 342 No. of personnel 17 (1–205) 7 (1–49) 2 (1–22) No. of beds 16 (2–174) – – No. of visits 8,242 (156–57,060) 6,122 (12–75,780) 2,755 (12–64,800) No. of admissions 683 (12–15,136) – – Technical efficiency score 0.30 (0.01–1) 0.09 (0.00–1) 0.04 (0.00–1) No. of efficient units 4 1 1 Percentage of efficient units 4% <1% <1% Source: World Bank staff estimates based on SPA 2013. Note: – = not available; CAL = centre de santé avec lit (health center with bed); CSL = centre de santé sans lit (health center without bed). TABLE C.2: Descriptive Statistics of Hospitals: Haiti, 2013 Variable (n=78) Mean Standard deviation Min Max Discharges 2,741 5,192 60 28,800 Outpatient visit 22,343 21,467 12 119,880 Medical doctor 15 15 1 88 Nurse 23 29 1 164 Aid nurse 17 18 – 80 Laboratory technician 7 5 – 23 Bed 61 65 3 400 Source: World Bank estimates based on SPA 2013. Note: – = not available. Of the 121 hospitals, data were missing for 43.Thus the sample was composed of 78 hospitals. BETTER SPENDING, BETTER CARE: 92 A LOOK AT HAITI’S HEALTH FINANCING Technical Efficiency Score of Primary Health Care Facilities, by Department, Ownership, and TABLE C.3: Location: Haiti, 2013 CAL CSL Dispensary Technical efficiency mean 0.30 0.09 0.04 Department Artibonite 0.32 0.17 0.05 Center 0.57 0.10 0.09 Grand’Anse 0.33 0.08 0.04 Nippes 0.16 0.04 0.03 North 0.27 0.08 0.03 North-East 0.37 0.07 0.03 North-West 0.19 0.02 0.04 West 0.26 0.09 0.05 South 0.24 0.05 0.03 South-East 0.26 0.05 0,05 Ownership MSPP 0.35 0.07 0.04 NGO 0.28 0.11 0.05 Private for-profit 0.26 0.08 0.04 Location Rural 0.22 0.10 0.04 Urban 0.40 0.09 0.06 Metropolitan 0.36 0.08 0.06 Source: World Bank staff calculations based on SPA 2014. Note: CAL = centre de santé avec lit (health center with bed); CSL = centre de santé sans lit (health center without bed); MSPP = Ministère de la Santé Publique et de la Population (Ministry of Public Health and Population); NGO = nongovernmental organization. TABLE C.4: Correlation between Technical Efficiency Score and Covariates, Primary Health Care Sector: Haiti CAL CSL Dispensary Department ANOVA 1.39 3.61*** 1.02 Location ANOVA 4.32** 0.61 0.16 Ownership ANOVA 0.51 2.40* 0.64 Drug index Correlation 0.21* 0.06 0.02 Equipment index Correlation 0.05 0.01 0.08 Improved source of water Correlation 0.09 0.09 0.01 Electricity Correlation 0.10 0.10* 0.04 Source: World Bank staff estimates based on SPA 2013. Note: Electricity: combines functional generator and fuel available today; definitions for electricity and improved source of water: SPA 2013; drug and equipment indexes: developed in line with the WHO’s SARA methodology (WHO 2010a). Bivariate statistics were applied in each of the three following PHC facility data sets: CAL, CSL, and dispensary. ANOVA = analysis of variance; CAL = centre de santé avec lit (health center with bed); CSL = centre de santé sans lit (health center without bed); SARA = Service Availability and Readiness Assessment. *p < .10 **p < .05 ***p < .01 APPENDIXES 93 TABLE C.5: Technical Efficiency Scores of Hospitals, by Ownership and Facility Type: Haiti, 2013 Mean Standard deviation Min Max Facility type University hospital 0.52 0.41 0.08 1 Departmental hospital 0.36 0.31 0.11 1 HCR 0.52 0.31 0.08 1 Small hospital 0.48 0.34 0.03 1 Ownership MSPP 0.47 0.33 0.08 1 NGO 0.60 0.36 0.06 1 Private 0.41 0.28 0.03 1 Source: World Bank staff estimates based on SPA 2013. Note: HCR = hôpital communautaire de référence (community referral hospital); MSPP = Ministère de la Santé Publique et de la Population (Ministry of Public Health and Population); NGO = nongovernmental organization. TABLE C.6: Technical Efficiency Scores of Hospitals, by Department, Ownership, and Category: Haiti, 2013 Ownership Category Department Department Community University Departmental Small score MSPP NGO Private referral hospital hospital hospital (n) hospital West 0.42 (34) 0.43 1 0.27 0.41 – 0.67 0.42 North 0.50 (10) 0.52 0.49 0.45 1 0.28 0.52 0.38 North-West 0.45 (6) 0.49 0.40 0.59 – – – – North-East 0.28 (1) 0.28 – – – 0.28 – – Center 0.34 (5) 0.28 1 – 0.22 0.23 0.74 – Artibonite 0.23 (7) 0.23 0.07 0.67 – 0.11 0.23 0.06 South 0.26 (7) 0.23 0.63 0.17 – – 0.23 0.27 Grand’Anse 0.24 (3) 0.17 0.68 – – – 0.24 – Nippes 0.80 (3) 0.80 – – 1 0.55 – – South-East 0.63 (2) 0.63 – – – 0.26 – 1.00 Source: World Bank estimates based on SPA 2013. Note: – = not available; MSPP = Ministère de la Santé Publique et de la Population (Ministry of Public Health and Population); NGO = nongovernmental organization. BETTER SPENDING, BETTER CARE: 94 A LOOK AT HAITI’S HEALTH FINANCING TABLE C.7: Descriptive Statistics of the Macro-Costing Hospital Sample: Haiti, 2016 Mean ± standard deviation Median HTG 32,572,841 ± HTG 30,958,809 ($678,600 Annual expenditures HTG 20,476,426 ($426,592) ± $644,975) No. of staff 93 ± 62 88 Share of administrative and support staff 43.65% ± 10% 43.5% No. of beds 26 ± 16 25 No. of admissions 998 ± 784 785 No. of external consultations 18,104 ± 11,597 15,485 Bed occupancy rate 29.85% ± 16.83% 28% Average length of stay (days) 3.19 ± 2.09 2.8 Unit cost per bed day equivalent HTG 3,664 ± HTG 2,9922 ($76.34 ± $ 60.88) HTG 3,058 ($63.61) Recovery rate 36.23% ± 36.64% 38% Source: World Bank staff estimates based on data collected in 22 hospitals. Note: HTG = Haitian gourde. TABLE C.8: Determinants of Unit Cost per Bed Day Equivalent: Haiti, 2016 Log of unit cost Coefficient Standard error t value Ownership (MSPP) NGO 0.113 0.291 0.39 Private 1.089 0.418 2.58** Share of cost recovery –0.704 0.313 –2.25 Region (West) North –0.419 0.437 –0.96 South 0.001 0.383 0.00 Bed occupancy rate –0.564 0.648 –0.87 Average length of stay (days) 0.045 0.080 0.57 Share of outpatient departments –0.021 0.045 –0.48 Share of direct labor cost –1.747 1.432 –1.22 (compared with overhead expenses) Constant 5.012 1.398 3.58** Source: World Bank staff estimates based on SPA 2013. Note: The dependent variable is the log of the unit cost. MSPP = Ministère de la Santé Publique et de la Population (Ministry of Public Health and Population); NGO = nongovernmental organization. *p < .10, **p < .05, ***p < .01; R-squared, 0.70. APPENDIXES 95 TABLE C.9: Regression Analysis, Dependent Variable: Number of Hours Worked per Day, Haiti Dependent variable: number of hours worked per day Variable Coefficient Standard error Department (omitted variable: North-West) North-East –0.293* 0.156 Geography (omitted variable: urban) Rural -–0.367 0.235 Facility type (omitted variable: dispensary) Health center without bed (CSL) –0.398 0.205 Health center with bed (CAL) –0.248 0.222 Community referral hospital (HCR) -0.504 0.234 Job category (omitted variable: medical doctor) Nurse 0.064 0.210 Aid nurse 0.039 0.190 Professional status (omitted variable: civil servant) Contracted –0.064 0.164 Delay in salary (omitted variable: had delayed salary) Did not have delay in salary 0.226* 0.123 Second job –0.23* 0.125 Lack of medicines (omitted variable: not an obstacle) Obstacle to providing health services –0.064 0.178 Lack of equipment (omitted variable: not an obstacle) Obstacle to providing health services –0.278 0.217 R-squared 0.16 No. of observations (no. of medical staff) 122 Source: World Bank staff estimates based on human resource assessment conducted by Leadership, Management, and Governance project, a collaboration of the World Bank, USAID, and MSPP (2013). Results report a regression analysis. Note: CAL = centre de santé avec lit (health center with bed); CSL = centre de santé sans lit (health center without bed; HCR = hôpital communautaire de référence (community referral hospital). *p < .10 **p < .05 ***p < .01. BETTER SPENDING, BETTER CARE: 96 A LOOK AT HAITI’S HEALTH FINANCING BIBLIOGRAPHY Abzalova, R., C. Wickham, A. Chukmaitov, and T. Rakhipbekov. 1998. Baruwa, E., S. Faye, C. Yao, and W. Beausejour. 2015. Hôpital “Reform of Primary Health Care in Kazakhstan and the Effects on Universitaire de Mirebalais (HUM) Costing Study. Bethesda, MD: Primary Health Care Worker Motivation: The Case of Zhezkazgan Health Finance and Governance Project, Abt Associates. Region.” Major Applied Research 5, Working Paper No. 3, Baruwa, E., and M. Meline. 2014. Developing Haiti’s First Health Partnerships for Health Reform Project, Abt Associates, Bethesda, Financing Strategy. Bethesda, MD: Health Finance and Governance MD. Project, Abt Associates. Adhikari, S. R., N. M. Maskay, and B. P. Sharma. 2009. “Paying for Brinda, E. M., A. R. Andres, and U. Enemark. 2014. “Correlates of Hospital Based Care of Kalaazar and Economic Consequences.” Out-of-Pocket and Catastrophic Health Expenditures in Tanzania: Health Policy Plan 24: 12939. Results from a National Household Survey. BMC International AEDES Consortium. 2016. “Preparation de la future organization et ac- Health and Human Rights 5 (14): 5. compagnement de l’ouverture du nouvel HUEH: Hôpital de l’Uni- Buigit, S., R. Ettarh, and D. Amendah. 2015. “Catastrophic Health versité d’État d’Haïti. Synthese du projet.” Port-au-Prince, Haiti. Expenditure and Its Determinants in Kenya Slum Communities.” Aguilar Rivera, A. M., K. Xu, and G. Carrin. 2006. “The Bolivian International Journal for Equity in Health 14: 46. Health System and Its Impact on Health Care Use and Financial Burgert, Clara R., and Debra Prosnitz. 2014. “Linking DHS Household Risk Protection.” Discussion Paper No. 7, “Cluster Evidence and and SPA Facility Surveys: Data Considerations and Geospatial Information for Policy» (EIP), Health System Financing Department, Methods.” DHS Spatial Analysis Report No. 10. http://dhsprogram. World Health Organization, Geneva. com/pubs/pdf/SAR10/SAR10.pdf. Akazili J., M. Adjuik, S. Chatio, D. Kanyomse, A. Hodgson, M. Aikins, Buss, T. F. 2013. “Foreign Aid and the Failure of State Building in Haiti and J. Gyapong. 2008. “What Are the Technical and Allocative under the Duvaliers, Aristide Preval, and Martelly.” Working Paper Efficiencies of Public Health Centers in Ghana?” Ghana Medical 2013/104, WIDER. Journal 42 (4): 149–55. Cahn, D., and R. Clifford. 2013. “Health, Safety and Environmental Alexandre, P. K., G. Saint-Jean, L. Crandall, and E. Fevrin. 2005. Issues in Haiti.” Environmental Safeguards Unit Discussion Paper “Prenatal Care Utilization in Rural Areas and Urban Areas of Haiti.” No. IDB-DP-343, Inter-American Development Bank, Washington, Revista Panamericana de Salud Pública 18 (2): 84–92. DC, December. Alphonse, Roberson. 2016. “Seuls 5 % des dépenses publiques al- Calvin, D., J. Lohi, and E. R. Bulmer. 2014. “Haiti Systematic Country louées à la santé en Haïti, déplore la Banque mondiale.” Le Diagnostic. Trade Background Note.” SCD Background Paper, Nouvelliste, July 19. http://lenouvelliste.com/lenouvelliste/arti- World Bank, Washington, DC. cle/161110/Seuls-5-des-depenses-publiques-allouees-a-la-sante- Cavagnero, E., G. Carrin, K. Xu, and A. M. Aguilar-Rivera. 2006. en-Haiti-deplore-la-Banque-mondiale#sthash.1ZHAICBG.dpuf. “Health Financing in Argentina: An Empirical Study of Health Care Altaras, R. 2009. “Increasing Financial Access to Primary Health Care Expenditure and Utilization.” Innovations in Health Financing, for the Rural Poor: A Study on Changes in Health Service Use and Working Paper Series 8, Fundación Mexicana para la Salud/ Costs Following the Removal of User Fees in the Grande Anse, Instituto Nacional de Salud Pública. Haiti.” Médecins du Monde–France, with support from the Cavagnero, E., B. Daelmans, N. Gupta, R. Scherpbier, and A. Shankar. European Commission, Directorate–General of Humanitarian Aid 2008. “Assessment of the Health System and Policy Environment (ECHO). as a Critical Complement to Tracking Intervention Coverage for Appaix, O., P. E. Henry, and A. Badjeck. 2015. “Evaluation Independante Maternal, Newborn, and Child Health. Lancet 371 (9620): 1284–13. du Programme de Pays: Burkina Faso 2011–2015.” United Nations CHI (Haiti Child Health Institute), ICF International, and MSPP (Ministry Population Fund (UNFPA), New York, December. of Public Health and Population, Haiti). 2014. Haiti Service Araujo, J. M., M. Brueckner, M. Clvijo, E. Vostroknutova, and K Wacker. Provision Assessment 2013. Fairfax, VA: ICF International. 2014. “Benchmarking the Determinants of Economic Growth in Coe, G., and J. de Beyer. 2014. “The Imperative for Health Promotion in Latin America and the Caribbean.” Report 91015-LAC, World Universal Health Coverage.” Global Health Science and Practice Bank, Washington, DC. 2 (1): 10–22. Awate, P. 2014. “Primary Health Care System in India–Structure and Coelli, T. J. 1996. “A Guide to DEAP: Version 2.1: A Data Envelopment Services.” PowerPoint presentation by Pradip Awate, assistant di- Analysis (Computer) Program.” Coordination Nationale de la rector of health services, Maharashtra, published online July 28. Sécurité Alimentaire (2011), Enquête Nationale de la Sécurité Barber S., F. Bonnet, and H. Bekedam. 2004. “Formalizing Under-the- Alimentaire. Table Payments to Control Out-of-Pocket Hospital Expenditures in Coello, Barbara, Gbemisola Oseni, Tanya Savrimootoo, and Eli Weiss. Cambodia.” Health Policy and Planning 19 (4): 199–208. 2014. Rural Development in Haiti: Challenges and Opportunities. Barro, R., and J. Lee. 2000. “International Data on Educational Washington, DC: World Bank. Attainment Updates and Implications.” NBER Working Paper Collier, P. 2007a. The Bottom Billion: Why the Poorest Are Failing and 7911, National Bureau of Economic Research, Cambridge, MA. What Can Be Done about It. New York: Oxford University Press. APPENDIXES 97 ____. 2007b. Post-Conflict Recovery: How Should Policies Be Heller, P. 2005. “Understanding Fiscal Space.” IMF Policy Discussion Distinctive. Oxford: Center for the Study of African Economies. Paper, Fiscal Affairs Department, International Monetary Fund, Collins, D., J. L. Mukunzi, Z. Jarrah, C. Ndizaye, P. Kayobotsi, C. Washington, DC. Mukantwali, B. Nzeyimana, and M. Cros. 2011. “Management ____. 2006. “The Prospects of Creating ‘Fiscal Space’ for the Health Sciences for Health.” Submitted to U.S. Agency for International Sector. Health Policy and Planning 21 (2): 75–79. Development by the Integrated Health System Strengthening Heredia-Ortiz, E. 2016. “DAI. Mobilizing Tax Revenue and Prioritizing Project, Rwanda Health Service Costing: Health Centre Analysis. Health Spending in El Salvador–A Case Study.” USAID/Health Collins, David, Prateek Gupta, and Em Sovannarith. 2009. “Cost Finance and Governance Program, U.S. Agency for International Projections for the Complementary Package of Activities for Referral Development, Washington, DC. Hospitals: Ministry of Health, Royal Government of Cambodia.” Basic Hernandez, A. R., and M. S. Sebastian. 2013. “Assessing the Technical Support for Institutionalizing Child Survival (USAID/BASICS) for the Efficiency of Health Posts in Rural Guatemala: A Data Envelopment U.S. Agency for International Development (USAID), Arlington, VA. Analysis.” Global Health Action, no. 7. Cotlear, Daniel, Somil Nagpal, Owen Smith, Ajay Tandon, and Rafael Herrera, J., N. Lamaute-Brisson, D. Milbin, F. Roubaud, C. Saint-Macary, Cortez. 2015. Going Universal: How 24 Developing Countries C. Torelli, and C. Zanuso. 2014. “L’Evolution des conditions de Are Implementing Universal Health Coverage Reforms from vie en Haïti entre 2007 et 2012. La réplique sociale du séisme.” the Bottom Up. Washington, DC: World Bank. http://www- Haitian Institute of Statistics and Data Processing, DIAL, Paris and wds.worldbank.org/external/default/WDSContentServer/ Port-au-Prince. WDSP/IB/2015/09/10/090224b0830cc779/1_0/Rendered/PDF/ Hughes, J., A. Glassman, and W. Gwenigale. 2012. ”Innovative Financing Going0universa0s0from0the0bottom0up.pdf. in Early Recovery: The Liberia Health Sector Pool Fund.” Working Cros, M., and W. Zeng. 2014. “Measuring Efficiency of Primary Health Paper 288, Center for Global Development, Washington, DC. Care Facilities in Haiti Using Data Envelopment Analysis (DEA).” IDB (Inter-American Development Bank). 2013. “Correspondence with Poster presentation at the 3rd Global Symposium on Health Multilateral Investment Fund from the Secretary on September 4, Systems Research in Cape Town, September 30–October 3. 2013. Subject: Haiti. Nonreimbursable Technical-Cooperation for Deininger, Klaus, and Paul Mpuga. 2004. “Economic and Welfare Effects the Project Carte Avantage Santé–Expansion of Basic Healthcare of the Abolition of Health User Fees: Evidence from Uganda.” Policy Services to Low-income Populations through a Pre-paid Health Research Working Paper 3276, World Bank, Washington, DC. Services Card.” Washington, DC. Dieleman, M., and J. W. Harnmeijer. 2006. “Improving Health Worker IHE (Institut Haïtien de l’Enfance) and ICF International. 2014. Performance: In Search of Promising Practices.” Evidence and Évaluation de la Prestation des Services de Soins de Santé (EPSSS), Information for Policy, Department of Human Resources for Haïti 2013. Rockville, MD: IHE and ICF International. Health, World Health Organization, Geneva. IMF (International Monetary Fund). 2011. Revenue Mobilization in Durham, J., Michael, M., Hill, P. S., and E. Paviignani. 2015. “Haiti and Developing Countries. Washington, DC: IMF. the health marketplace: the role of the private, informal market in ____. 2016. World Economic Outlook. Washington, DC: IMF, April. filling the gaps left by the state.” BMC Health Serv Res. 15: 424. INEC (Instituto Nacional de Estadísticas y Censos) and MINSA (Ministerio Published online 2015 Sep 28. de Salud). 2002. Encuesta Nacional de Salud y Demografía Echevin, Damien. 2011a. “Livelihoods and the Allocation of Emergency 2001. Calverton, MD: Macro International, October. Assistance after the Haiti Earthquake.” Policy Research Working INSAE (Institut National de la Statistique et de l’Analyse Économique) Paper 5851, World Bank, Washington, DC. and ICF International. 2013. Enquête Démographique et de Santé ____. 2011b. “Vulnerability and Livelihoods before and after the Haiti du Bénin 2011–2012. Calverton, MD: INSAE and ICF International. Earthquake. Policy Research Working Paper 5850, World Bank, INSD (Institut National de la Statistique et de la Démographie) and ICF Washington, DC. International. 2012. Enquête Démographique et de Santé et à Eichler, R., P. Auxila, and J. Pollock. 2001. “Output-Based Health Care: Indicateurs Multiples du Burkina Faso, 2010. Calverton, MD: Paying for Performance in Haiti. Public Policy in the Private Sector.” INSD and ICF International. Note 236, World Bank, Washington, DC. Jamison, Dean T., Joel G. Breman, Anthony R. Measham, George Eichler, R., and R. Levine, eds. 2009. Performance Incentives for Global Alleyne, Mariam Claeson, David B. Evans, Prabhat Jha et al. 2006. Health: Potential and Pitfalls. Center for Global Development. Disease Control Priorities in Developing Countries. 2nd ed. Baltimore: Brooking Institution Press. Washington, DC: World Bank and Oxford University Press. https:// ENFOLIEN. 2015. “Développement du Système de l’Assurance de openknowledge.worldbank.org/handle/10986/7242. l’OFATMA–Les Orientations Stratégiques.” http://www.enfolien. Jehu-Appiah, C., S. Sekidde, M. Adjuik, J. Akazili, D. D’Almeida, F. Nyonator, net/index.php?option=com_content&view=article&id=122%3A- R. Baltussen et al. 2014. “Ownership and Technical Efficiency of documentaire-developpement-du-systeme-dassurance-de-lofat- Hospitals: Evidence from Ghana Using Data Envelopment Analysis.” ma-les-orientations-strategiques&catid=37%3Aorganismes-aut- Cost Effectiveness and Resource Allocation 12: 9. onomes&Itemid=87&showall=1. Jha, Prabhat, and Richared Peto. 2014. “Global Effects of Smoking, Fritsche, G. B., R. Soeters, and B. Meessen. 2014. Manual de pago por of Quitting and of Taxing Tobacco.” New England Journal of desempeno. Washington, DC: World Bank. Medicine 370: 60–68. Gertler, P., and J. Hammer. 1997. “Strategies for Pricing Publicly Johnson, B. G., M. Laverty, and M. Sjoblom. 2014. “Assessment: Haiti. Provided Health Services.” Policy Research Working Paper 1762, Field Validation of RBF–Supply Chain Bottleneck Assessment Tool, World Bank, Washington, DC. July 21–28, 2014.” William Davidson Institute at University of Hamidi, S. 2016. “Measuring Efficiency of Governmental Hospitals in Michigan and World Bank Group, Washington, DC. Palestine Using Stochastic Frontier Analysis.” Cost Effectiveness Joseph, G., I. C. Mohnsam da Silva, F. C. Wehrmeister, A. J. D. Barros, and and Resource Allocation 14: 3. C. G. Victora. 2016. “Inequalities in the Coverage of Place of Delivery BETTER SPENDING, BETTER CARE: 98 A LOOK AT HAITI’S HEALTH FINANCING and Skilled Birth Attendance: Analyses of Cross-Sectional Surveys in McPake, B., N. Brikci, G. Comet, A. Schmidt, and E. Araujo. 2011. 80 Low and Middle-Income Countries. Reproductive Health 13: 77. “Removing User Fees: Learning from International Experience to Josephson, E. and L. Vinyals. 2012. “An Assessment of the Haitian Support the Process.” Health Policy and Planning 26: ii104–ii117. MSPP’s Readiness to Establish a Contracting Function.” Institute MIF (Multilateral Investment Fund). 2012. Access to Basic Services for Collaborative Development, Bethesda, MD. Agenda Review. Washington, DC: MIF. Josephson, E., and M. Bode. 2013. Fiscal Space Analysis for Haiti. ____. 2013. “MIF and DASH to Facilitate Prepaid Health Services Washington, DC: World Bank. for Low Income Populations in Haiti.” Project announcement, Kingdom of Morocco. 2014. “Revue des dépenses publiques. Secteur September 17. http://www.fomin.org. de la santé.” Report no. 86115-MA, World Bank, Washington, DC. Ministry of Economics and Finance, Haiti. 2011. Manuel de Procedures Kirigia, J. M., and E. Z. Asbu. 2013. “Technical and Scale Efficiency D’Elaboration du Budget. Port-au-Prince: Direction General du of Public Community Hospitals in Eritrea: An Exploratory Study.” Budget, Ministry of Economics and Finance. Health Economics Review 3: 6. Ministry of Health, Afghanistan. 2003. A Basic Package of Health Kloos, H. 1990. “Utilisation of Selected Hospitals, Health Centres and Services for Afghanistan 2003/1382. Kabul: Transitional Islamic Health Stations in Central, Southern and Western Ethiopia.” Social Government of Afghanistan. Science and Medicine 31: 1003–19. Ministry of Health, Eritrea. 2014. “Eritrea Health Sector Strategic Lagomarsino, G., A. Garabrant, A. Adyas, R. Muga, and N. Otoo. 2012. Development Plan (HSSDP 2012–2016). Mid-Term Review. 3rd “Moving towards Universal Health Coverage: Health Insurance.” November–19th November 2014. Final Report.” http://www.na- Lancet 380 (9845): 933–43. tionalplanningcycles.org/sites/default/files/planning_cycle_reposi- Lamaute-Brisson, N. 2013. “Social Protection Systems in Latin America tory/eritrea/final_mtr_report_eritrea.pdf. and the Caribbean: Haiti.” United Nations Economic Commission Ministry of Health, Federal Republic of Ethiopia. 2010. “Health Sector for Latin America and the Caribbean (ECLAC) Project Document. Development Program IV.” October. ____. 2015. “Promotion and Social Protection of Childhood and Ministry of Health, Uganda, and Macro International. 2008. “Uganda Adolescence in Haiti.” United Nations Economic Commission for Service Provision Assessment Survey 2007.” Kampala: Ministry of Latin America and the Caribbean (ECLAC) Project Document. Health and Macro International. http://dhsprogram.com/pubs/ Largo, F. M. 2012. “A Macro-Costing Analysis of Hospitals in the pdf/SPA13/SPA13.pdf. Philippines.” University of San Carlos, Cebu, Philippines. Ministry of Health and Social Welfare, Republic of Tanzania. 2012a. Le Moniteur journal officiel de la République d’Haïti (Official Gazette “Directorate of Policy and Planning, Health Sector Public of Haiti). 2014. Budget national de la République d’Haïti pour l’ex- Expenditure Review, 2010/11.” Tanzania and Health Systems ercice 2013–2014, 169è Année–Spécial No. 2 (National Budget of 20/20, Abt Associates, Dar es Salaam, July. the Republic of Haiti for 2013–2014, 169th Year–Special Issue No. ____. 2012b. “National Health Accounts 2010 with Sub-accounts for 3), October 1, 2014. Port-au-Prince: Les presses nationales d’Haïti. HIV and AIDS, Malaria, Reproductive and Child Health.” Health ____. 2015. Présentation Générale des Ressources Budgétaires, Systems 20/20, Dar es Salaam, May. 170é Année–Spécial No. 2 (General Presentation of Budget for Ministry of Public Health, Afghanistan. 2013. “Afghanistan National Resources 170th Year–Special No. 2), March 27, 2015. Port-au- Health Accounts with Subaccounts for Reproductive Health 2011– Prince : Les presses nationales d’Haïti. 2012.” Health Policy Project, October. Le Nouvelliste. 2013. “Profile of Dr. Ronald Laroche, Digicel Ministry of Public Health, Burundi. 2015. “Comptes de la Sante au Entreprenuer of the Year Finalist.” November 22. http://lenou- Burundi: 2012–2013” (National Health Accounts in Burundi: velliste.com/lenouvelliste/article/124252/Profile-of-Dr-Ronald- 2012–2013). Bujumbura. Laroche-Digicel-Entrepreneur-of-the-Year-finalist.html. MoHSW (Ministry of Health and Social Welfare, Republic of Liberia). Li, Y., Q. Wu, L. Xu, D. Legge, Y. Hao, L. Gao, N. Ning et al. 2012. “Factors 2008. “Basic Package of Health and Social Welfare Services for Affecting Catastrophic Health Expenditure and Impoverishment from Liberia”. Government of Liberia, June. http://liberiamohsw.org. Medical Expenses in China: Policy Implications of Universal Health ____. 2015. ”Investment Plan for Building a Resilient Health System in Insurance.” Bulletin of the World Health Organization 90: 664–71. Liberia: 2015 to 2021.” Version 15, Government of Liberia, April. Lobo, M. S. C., Y. A. Ozcan, A. C. M. Silva, M. P. E. Lins, and R. Fiszman. http://pages.au.int. 2010. “Financing Reform and Productivity Change in Brazilian MSH (Management Sciences for Health). 2001. “Using Performance- Teaching Hospitals: Malmquist Approach.” Central European Based Payments to Improve Health Programs.” The Manager 10. Journal of Operations Research 18 (2): 141–52. MSPP (Ministry of Public Health and Population, Haiti). 2006. “Le Marshall, P., and S. Flessa. 2011. “Efficiency of Primary Care in Rural Paquet Minimum de Service.” Port-au-Prince. Burkina Faso. A Two-Stage DEA Analysis.” Health Economics ____. 2009. Haiti Global Fund Facility Survey 2008. Geneva: Global Review 1: 5–10. Fund. Martey, J., J. Djan, S. Twum, E. Browne, and S. Opoku.1998. “Referrals ____. 2012. “Politique Nationale de la Santé.” Port-au-Prince. for Obstetric Complications from Ejisu District, Ghana.” West ____. 2014a. “Evaluation des Ressources Humaines es Institutions African Journal of Medicine 17: 58–63. Sanitaires du Secteur Public (ERHIS 1).” Publication not yet released. Masiye, F., J. M. Kirigia, A. Emrouznejad, L. G. Sambo, A. Mounkaila, ____. 2014b. “Manuel de politiques, norms et procedures de D. Chimfwembe, D. Okello et al. 2006. “Efficient Management of Gestion des Ressources Humaines.” Office de Mangement et des Health Centres Human Resources in Zambia.” Journal of Medical Ressources, MSPP, Port-au-Prince, December. Systems 30 (6): 473–81. ____. 2015a. “Les Grandes Realisations 2014–15 Tome 1 & 2.” McCoy, D., S. Chand, and D. Sridhar. 2009. “Global Health Funding: Port-au-Prince. How Much, Where It Comes From and Where It Goes.” Health ____. 2015b. “Rapport des Comptes Nationaux de Sante: 2012– Policy and Planning 24 (6): 407–17. 2013.” Unité d’Etudes et de Programmation (UEP). BIBLIOGRAPHIE 99 ____. 2016. “Les Grandes Realisations 2014–15 Tome 3.” Ramachandran, R., and J. Walz. 2012. “Haiti: Where Has All the Money Port-au-Prince. Gone?” CGD Policy Paper 004, Center for Global Development, MSPP (Ministry of Public Health and Population, Haiti), IHE (Institute of Washington, DC. http://www.cgdev.org/content/publications/ Finance), and ICF International. 1994. Demographic and Health detail/1426185. Survey. Calverton, MD: MSPP, IHE, and ICF International. Saleh, K. 2012. A Health Sector in Transition to Universal Coverage in ____. 2000. Demographic and Health Survey. Calverton, MD: MSPP, Ghana. Washington, DC: World Bank. IHE, and ICF International. ____. 2013. The Health Sector in Ghana: A Comprehensive ____. 2006. Demographic and Health Survey. Calverton, MD: MSPP, Assessment. Directions in Development: Human Development. IHE, and ICF International. Washington, DC: World Bank. ____. 2013. 2012 Haiti Mortality, Morbidity, and Service Utilization Savedoff, W. 2016. Tobacco Taxes: Good for Public Health, Good for Survey: Key Findings. Calverton, MD: MSPP, IHE, and ICF Fiscal Revenues. Washington, DC: World Bank. International. Savedoff, W., and A. Alwang. 2015. “The Single Best Health Policy in MSPP (Ministry of Public Health and Population, Haiti), IHSI (Haiti Institute the World: Tobacco Taxes.” CGD Policy Paper, Center for Global of Statistics and Informatics), WHO (World Health Organization), Development, Washington, DC. and USAID (U.S. Agency for International Development. 2008. Schneider, P. 2007. “Provider Payment Reforms: Lessons from Europe Haiti National Health Accounts, HIV and AIDs, Tuberculosis, and and America for South Eastern Europe.” HNP Discussion Paper, Malaria Subaccounts 2005–2006. Port-au-Prince: MSPP. World Bank, Washington, DC, October. Murray, C. J., B. Shengelia, N. Gupta, S. Moussavi, A. Tandon, and M. Sebastian, M. S., and H. Lemma. 2010. “Efficiency of the Health Thieren. 2013. “Validity of Reported Vaccination Coverage in 45 Extension Programme in Ethiopia: A Data Envelopment Analysis.” Countries.” Lancet 362: 1022–27. BMC International Health and Human Rights 10 (16). Nakamura, Y., and A. Williamson. 2015. “Government Health Shepard, D. S., D. Hodgkin, and Y. E. Anthony. 2000. Analysis of Spending and Tax Reform in Rwanda 2000–2013: A Case Study.” Hospital Costs: A Manual for Managers. Geneva: World Health Results for Development Institute, U.S. Agency for International Organization. Development, Washington, DC. Singh, Raju and M. Barton-Dock. 2015. Haiti: Toward a New Narrative. National Coordinating Agency for Population and Development Systematic Country Diagnostic. Washington, DC: World Bank. (NCAPD) [Kenya], Ministry of Medical Services (MOMS) [Kenya], Stierman, E, F. Ssengooba and S. Bennett. 2013. “Aid alignment: a lon- Ministry of Public Health and Sanitation (MOPHS) [Kenya], Kenya ger term lens on trends in development assistance for health in National Bureau of Statistics (KNBS) [Kenya], and ICF Macro. Uganda.” Globalization and Health 2013, 9:7. 2011. Kenya Service Provision Assessment Survey 2010. Nairobi: Street, A., J. O’Reilly, P. Ward, and A. Mason. 2011. “DRG-Based NCAPD, MOMS, MOPHS, KNBS, and ICF Macro. http://dhspro- Hospital Payment and Efficiency: Theory, Evidence, and gram.com/pubs/pdf/SPA17/SPA17.pdf. Challenges.” In Diagnosis Related Groups in Europe. Moving to- Noirhomme, M., et al. 2007. “Improving Access to Hospital Care for wards Transparency, Efficiency and Quality in Hospitals, edit- the Poor: Comparative Analysis of Four Health Equity Funds in ed by R. Busse, A. Geissler, W. Quentin, and M. Wiley, 93–114. Cambodia.” Health Policy and Planning 22: 246–62. Copenhagen: European Observatory on Health Systems and Nwagbara, V. C., and R. Rasiah. 2015. “Rethinking Health Care Policies Series, WHO Regional Office for Europe. Commercialization: Evidence from Malaysia.” Globalization and Su, T. T., B. Koaita, and S. Flessa. 2006. “Catastrophic Household Health 11: 44. Expenditure for Health Care in a Low Income Society: A Study OFATMA (Office d’Assurance Accidents du Travail, Maladie et from Nouna District, Burkina Faso.” Bulletin of the World Health Maternité). 2015. “Programme Assurance Maladie et Maternité.” Organization 84: 217. Powerpoint presentation, May 2015. Suaya, J. A., D. S. Shepard, J. B. Siqueira, C. T. Martelli, L. C. S. Lum, Okwere, P., A. Tandon, S. Sparkes, J. Mclaughlin, and J. G. Hoogeveen. L. H. Tan, S. Kongsin et al. 2009. “Cost of Dengue Cases in 2010. “Fiscal Space for Health in Uganda.” Working Paper 186, Eight Countries in the Americas and Asia: A Prospective Study.” Africa Human Development Series, World Bank, Washington, DC. American Journal of Tropical Medicine and Hygiene 80: 846–55. Osier, D., S. D’Almeida, M. O. George, J. M. Kirigia, A. O. Mensah, and Ujoh, F., and F. Kwaghsende. 2014. “Analysis of the Spatial Distribution L. H. Kainyu. 2005. “Technical Efficiency of Public District Hospitals of Health Facilities in Benue State.” Nigeria Scientific and and Health Centers in Ghana: A Pilot Study.” Cost Effectiveness Academic Publishing 4 (5): 210–18. and Resource Allocation 3: 9. UNFPA (United Nations Population Fund). 2016. “Universal Access Osmani, A R. 2015. “Technical and Scale Efficiency of District Hospitals to Reproductive Health: Progress and Challenges.” https://www. in Afghanistan: A Data Envelopment Analysis Approach.” https:// unfpa.org/sites/default/files/pubpdf/UNFPA_Reproductive_ www.researchgate.net. Paper_20160120_online.pdf. Page, A.-L., I. Ciglenecki, E. R. Jasmin, L. Desvignes, F. Grandesso, J. Urrutia, R. P., D. Merisier, M. Small, E. Urrutia, N. Tinfo, and D. K. Polonsky et al. 2015. “Geographic Distribution and Mortality Risk Walmer. 2012. “Unmet Health Needs Identified by Haitian Women Factors during the Cholera Outbreak in a Rural Region of Haiti, as Priorities for Attention: A Qualitative Study.” Reproductive 2010–2011.” PLoS Neglected Tropical Diseases 9 (3): e0003605. Health Matters 20 (39): 93–103. PAHO (Pan American Health Organization). 2012. Regional Outlook Valdmanis, V. G., M. D. Rosko, and R. L. Mutter. 2008. “Hospital and Country Profiles: Health in the Americas. Washington, DC: Quality, Efficiency, and Input Slack Differentials.” HSR: Health PAHO. Services Research 43: 5, Part II. Powell-Jackson, T., K. Hanson, and D. McIntyre. 2012. “Fiscal Space for Verner, Dorte. 2008. “Making Poor Haitians Count: Poverty in Rural and Health: A Review of the Literature.” Working Paper 1, Resilient and Urban Haiti Based on the First Household Survey for Haiti.” Policy Responsive Health Systems (RESYST). Research Working Paper 4571, World Bank, Washington, DC. BETTER SPENDING, BETTER CARE: 100 A LOOK AT HAITI’S HEALTH FINANCING Wagenaar, B. H., B. A. Kohrt, A. K. Hagaman, K. E. McLean, and B. N. ———. 2016. World Health Statistics 2016. Geneva: WHO. Kaiser. 2013. “Determinants of Care Seeking for Mental Health WHO (World Health Organization) and World Bank. 2013. “Monitoring Problems in Rural Haiti: Culture, Cost, or Competency.” Psychiatric Progress towards Universal Health Coverage at Country and Global Services 64 (4): 366–72. Levels: A Framework.” Joint WHO/World Bank Group Discussion Walsh, S., K. Norr, G. Sankar, and H. Sipsma. 2015. “Newborn Cord Paper 1. Care Practices in Haiti.” Global Public Health 10 (9). ____. 2015. Tracking Universal Health Coverage: First Global Wang, H., D. Young, and C. Connor. 2009. Synthesis of Health Monitoring Report. Geneva: WHO. Financing Studies and Alternative Health Financing Policies World Bank. 1999. Curbing the Epidemic: Governments and the in Liberia. Bethesda, MD: Health Systems 20/20 Project, Abt Economics of Tobacco Control. Washington, DC: World Bank. Associates. ____. 2004. Benin Enhancing the Effectiveness of Public Spending. A Wang W., R. Winter, L. Mallick, L. Florey, C. Burgert-Brucker, and E. Review of Three Sectors. Washington, DC: World Bank. Carter. 2015. The Relationship between the Health Service ____. 2005a. Burkina Faso: The Budget as Centerpiece of PRSP Environment and Service Utilization: Linking Population Data to Implementation, Public Expenditure Review. Africa Region 2005. Health Facilities Data in Haiti and Malawi. DHS Analytical Studies Washington, DC: World Bank. No. 51. Rockville, MD: ICF International. ____. 2005b. Natural Disaster Hotspots: A Global Risk Analysis. White, K., et al. 2006. “Health Seeking Behavior among Pregnant Disaster Risk Management Series No. 5. Washington, DC: World Women in Rural Haiti.” Health Care for Women International 27: Bank. 822–38. ____. 2008. Haiti–Public Expenditure Management and Financial WHO (World Health Organization). 2001. “Macroeconomics and Accountability Review (PEMFAR): Improving Efficiency of the Health: Investing in Health for Economic Development.” Report of Fiscal System and Investing in Public Capital to Accelerate the Commission on Macroeconomics and Health. Geneva: WHO. Growth and Reduce Poverty. Public Expenditure Review (PER). ———. 2002. World Health Survey: A Household Questionnaire. Washington, DC: World Bank. Evidence and Information for Policy. Geneva: WHO. ____. 2010a. Haiti Earthquake PDNA (Post-Disaster Needs ———. 2004. “The Impact of Health Expenditure on Households and Assessment): Assessment of Damage, Losses, General and Options for Alternative Financing.” Report EM/RC51/4, Regional Sectoral Needs. Washington, DC: World Bank. http://docu- Committee for the Eastern Mediterranean, WHO, Cairo. ments.worldbank.org/curated/en/355571468251125062/Haiti- ____. 2006. The World Health Report 2006: Working Together for earthquake-PDNA-Post-Disaster-Needs-Assessment-assessment- Health. Geneva. of-damage-losses-general-and-sectoral-needs. ____. 2008. “Essential Health Packages: What Are They For? What ____. 2010b. Public Expenditure Tracking and Service Delivery Do They Change?” WHO Service Delivery Seminar Series: DRAFT Survey–Education and Health in Honduras. Latin America and Technical Brief No. 2, Geneva. the Caribbean Regional Office, World Bank. ———. 2010a. Macroeconomics and Health: Investing in Health ____. 2011. United Republic of Tanzania Public Expenditure Review for Economic Development. Report of the Commission on 2010. Washington, DC: World Bank. Macroeconomics and Health. Geneva: WHO. ____. 2012. “Health Equity and Financial Protection Datasheets: Latin ———. 2010b. Monitoring the Building Blocks of Health Systems: America and Sub-Saharan Africa 2012.” http://www.worldbank. A Handbook of Indicators and Their Measurement Strategies. org/povertyandhealth. Geneva: WHO. ____. 2014. Reflections for Evidence-Based Policy Making. ———. 2010c. WHO Technical Manual on Tobacco Tax Washington, DC: World Bank. http://documents.worldbank.org/ Administration. Geneva: WHO. curated/en/2014/01/24142063/investing-people-fight-pover- ———. 2010d. World Health Report 2010: Health Systems Financing: ty-haiti-reflections-evidence-based-policy-making. The Path to Universal Health Coverage. Geneva: WHO. ____. 2015a. “The Balanced Scorecard: Basic Indicators of Haiti’s ____. 2011. The Abuja Declaration: Ten Years On. Geneva: WHO. Health System.” Unpublished manuscript, World Bank, ____. 2013a. “Enforcing Bans on Tobacco Advertising, Promotion and Washington, DC. Sponsorship.” In WHO Report on the Global Tobacco Epidemic. ____. 2015b. Better Spending, Better Services: A Review of Public Geneva: WHO. Finances in Haiti. Public Expenditure Review (PER), Vol. 1. ———. 2013b. World Health Report 2013: Research for Universal Washington, DC: World Bank. http://documents.worldbank.org/ Coverage. Geneva: WHO. curated/en/2015/09/24984747/haiti-towards-greater-fiscal-sus- ———. 2013c. World Health Statistics 2013. Geneva: WHO. tainability-equity-discussion-public-finance. ———. 2014a. “Haiti.” NHA Report 2005–2006. http://www.who. ____. 2015c. Doing Business 2016: Going Beyond Efficiency. int/nha/country/hti/en/. Washington, DC: World Bank. ———. 2014b. “Haiti: Health Profile.” Geneva, May. ____. 2015d. Doing Business Economy Profile 2016: Haiti. ____. 2014c. Initiative Pour un Monde Sans Tabac. Taxation des pro- Washington, DC: World Bank. duits du tabac. Geneva: WHO. ____. 2015e. “Innovative Health Financing Initiatives. How Much Can ____. 2014d. Initiative Pour un Monde Sans Tabac. Taxation du ta- They Contribute To Financial Sustainability of Health Systems?” bac: Un financement novateur des systèmes de santé. Geneva: Health Financing Global Solution Area HNP GP Management Team WHO. Meeting, February. ———. 2014e. “Trends in Maternal Mortality: 1990 to 2013. Estimates ____. 2015f. “Republic of Haiti: Towards Greater Fiscal Sustainability by WHO, UNICEF, UNFPA, the World Bank and the United Nations and Equity: A Discussion of Public Finance.” GMFDR–Latin Population Division.” Geneva. America, Report No. ACS14253, World Bank, Washington, DC. ———. 2015. World Health Statistics 2015. Geneva: WHO. ____. 2015g. World Development Indicators 2015. Washington, DC. BIBLIOGRAPHIE 101 ____. 2016a. Better Spending, Better Services: A Review Wright, J. 2015. Health Finance and Governance Project. Essential of Public Finances in Haiti. Public Expenditure Review Package of Health Services Country Snapshot: Haiti. Bethesda, (PER), Vol. 2. Washington, DC: World Bank. http://doc- MD: Health Finance and Governance Project, Abt Associates. uments.worldbank.org/curated/en/2016/06/26527828/ Xu, Ke. 2005. “Distribution of Health Payments and Catastrophic better-spending-better-services-review-public-finances-haiti-vol-2. Expenditures Methodology.” Discussion Paper No. 2, World Health ____. 2016b. World Development Indicators 2016. Washington, DC: Organization, Geneva. World Bank. Yang, F. 2013. “Health Beliefs and Contraception Use in Leogane, Haiti: World Bank and International Finance Corporation, Latin America A Qualitative Study.” Duke University Libraries, Durham, NC. http:// and the Caribbean Region. 2012. “Interim Strategy Note for dukespace.lib.duke.edu/dspace/handle/10161/7326. the Republic of Haiti for FY13-14. Report No. 71885-HT.” Haiti Yoriko, Nakamura, and April Williamson. 2016. “Results for Country Management Unit Latin America and the Caribbean Development Institute. Government Health Spending and Tax Region, September 27. Reform in Rwanda 2000–2013: A Case Study.” USAID/Health World Bank and Observatoire National de la Pauvreté et de l’Exclu- Finance and Governance Program, U.S. Agency for International sion Sociale. 2014. Investing in People to Fight Poverty in Haiti: Development, Washington, DC. Reflections for Evidence-based Policy Making. Washington, DC: Zeng W., M. Cros, K. Dilley, and D. Shepard. 2013. “Impact of World Bank. Performance-Based Financing on Primary Health Care Services in World Bank, USAID (U.S. Agency for International Development), and Haiti.” Health Policy and Planning 28: 596–605. MSPP (Ministry for Public Health and Population, Haiti). 2013. Zere, E., T. Mbeeli, K. Shajngula, C. Mandlhate, K. Mutirum, B. “Study on Human Resources and Health Spending in the Nord-Est, Tjivambi, and W. Kapenambili. 2006. “Technical Efficiency Plateau Central et Nord-Ouest Health Departments. An Exercise of District Hospitals: Evidence from Namibia Using Data Conducted to Estimate the Amount of the Incentive Based on Envelopment Analysis.” Cost Effectiveness and Resource Performance under the Performance-Based Financing (PBF).” June. Allocation 4: 5. BETTER SPENDING, BETTER CARE: 102 A LOOK AT HAITI’S HEALTH FINANCING