UNIVERSAL HEALTH COVERAGE STUDY SERIES NO. 38 Please cite this paper as: Smith, O. 2018. “Sri Lanka: Achieving Pro-Poor Universal Health Coverage without Health Financing Reforms ”. Universal Health Coverage Study Series No. 38, World Bank Group, Washington, DC. 2 UNIVERSAL HEALTH COVERAGE STUDY SERIES No. 38 Sri Lanka: Achieving Pro-Poor Universal Health Coverage without Health Financing Reforms Owen Smith The World Bank, Washington, DC, 2018 3 4 Contents Preface to the second round of the Universal Health Coverage Study Series ................................ 7 Acknowledgments........................................................................................................................... 8 About the Author ............................................................................................................................ 9 Executive Summary ...................................................................................................................... 10 Introduction ................................................................................................................................... 12 General Overview: Supply Side.................................................................................................... 14 General Overview: Health Financing ........................................................................................... 17 Interpreting UHC in Sri Lanka: People, Services, and Financial Protection ................................ 21 Covering People ........................................................................................................................ 21 Covering Services ...................................................................................................................... 24 Financial Protection ................................................................................................................... 25 Pending Agenda ............................................................................................................................ 27 What Lessons for the Rest of the World? ..................................................................................... 29 Annex ............................................................................................................................................ 34 References ..................................................................................................................................... 36 Figures Figure 1a Infant Mortality Rate (per 1,000 live births)………………………………………….13 Figure 1b Government Health Expenditure (% of GDP)………………………………………..13 Figure 2a Sri Lanka’s Government Health Spending is Relatively Low as a Share of GDP…...17 Figure 2b … but Closer to the Average as a Share of Total Government Spending……………17 Figure 3 Reliance on OOP in Sri Lanka is Typical for its Income Level…………………….....18 Figure 4 What Do People Pay Out-of-Pocket For?......................................................................19 Figure 5a Estimated Per Capita Government Spending by Province, 2013…………………….21 Figure 5b Government Health Expenditures by Provider Type, 2013………………………….21 Figure 6 Utilization Patterns…………………………………………………………………….22 Figure 7 Average Time Spent at Facilities per Visit……………………………………………23 Figure 8 Out-of-Pocket Payments by Decile, 2015–16………………………………………...26 Figure 9 Incidence of Catastrophic Health Spending, 2015-16………………………………...27 Figure A.1 District-Level Health Care Delivery System…………………………….…………34 Figure A.2 Provincial-Level Health Care Administration……………………………………...34 Figure A.3 Central-Level Health Administration Structure…………………………………….35 Table Table 1. Tracer Universal Health Coverage Indicators ................................................................ 24 5 Abbreviations GDP Gross Domestic Product HIES Household Income and Expenditure Survey MDGs Millennium Development Goals MoH Ministry of Health NCDs Noncommunicable Diseases NHS National Health Service OOP Out-of-Pocket UHC Universal Health Coverage 6 Preface to the second round of the Universal Health Coverage Study Series All over the world countries are implementing pro-poor reforms to advance universal health coverage. The widespread trend to expand coverage resulted in the inclusion of the “achieving universal health coverage by 2030” target in the Sustainable Development Agenda. Progress is monitored through indicators measuring gains in financial risk protection and in access to quality essential health-care services. The Universal Health Coverage (UHC) Studies Series was launched in 2013 with the objective of sharing knowledge regarding pro-poor reforms advancing UHC in developing countries. The series is aimed at policy-makers and UHC reform implementers in low- and middle-income countries. The Series recognizes that there are many policy paths to achieve UHC and therefore does not endorse a specific path or model. The Series consists of country case studies and technical papers. The case studies employ a standardized approach aimed at understanding the tools –policies, instruments and institutions- used to expand health coverage across three dimensions: population, health services and affordability. The approach relies on a protocol involving around 300 questions structured to provide a detailed understanding of how countries are implementing UHC reforms in the following areas: • Progressive Universalism: expanding population coverage while ensuring that the poor and vulnerable are not left behind; • Strategic Purchasing: expanding the statutory benefits package and developing incentives for its effective delivery by health-care providers; • Raising revenues to finance health care in fiscally sustainable ways; • Improving the availability and quality of health-care providers; and, • Strengthening accountability to ensure the fulfillment of promises made between citizens, governments and health institutions. By 2017, the Series had published 24 country case studies and conducted a systematic literature review on the impact of UHC reforms. In 2018 the Series will publish an additional15 case studies, A book analyzing and comparing the initial 24 country case studies is also available: Going Universal: How 24 Developing Countries are Implementing UHC Reforms from the Bottom Up . Links to the Series and the book are included below. Daniel Cotlear, D. Phil. Manager and Editor Universal Health Coverage Study Series Links: http://www.worldbank.org/en/topic/health/publication/universal-health-coverage-study-series http://www.worldbank.org/en/topic/universalhealthcoverage/publication/going-universal-how- 24-countries-are-implementing-universal-health-coverage-reforms-from-bottom-up 7 Acknowledgments The author would like to thank Deepika Attygalle, Daniel Cotlear, Somil Nagpal, Aparnaa Somanathan, and participants at the UNICO workshop held in December 2016 for their valuable comments, and Eoghan Brady, Zelalem Debebe, David Newhouse, and Ani Silwal for timely data work. 8 About the Author Owen Smith is a Senior Economist with the World Bank’s Health, Nutrition and Population Global Practice. He is currently based in New Delhi, India, and was previously based in Colombo, Sri Lanka. Since joining the World Bank in 2005, he has worked on health financing, health policy, and social protection issues in the Europe and Central Asia and South Asia regions. Prior to joining the World Bank, he worked at a consulting firm and as an economist at the Canadian Ministry of Finance. He is a co-author of the World Bank report “Going Universal: How 24 countries are implementing universal health coverage reforms from the bottom up.” 9 Executive Summary Sri Lanka’s health system has a long track record of strong performance. For at least 50 years it has achieved much better outcomes in maternal and child health and infectious disease control than would have been predicted by its income level. Health financing indicators also indicate that the health system is both pro-poor and efficient. There are few if any other low or middle income countries that have simultaneously achieved strong health outcomes, good financial protection, and low cost. Many countries are lauded for achieving two out of three, but few can claim to have done as well as Sri Lanka on all fronts, especially considering that it is still classified as a lower- middle income country. The objective of this case study is to describe the main features and achievements of Sri Lanka’s high-performing health system, and to distill lessons for the rest of the world. The cornerstone of Sri Lanka’s Universal Health Coverage (UHC) agenda has been supply-side efforts to ensure strong service delivery. It has provided universal, free access to government- provided health care services to its population since the 1930s. Preventive health care services are provided through a well-planned network of facilities across the country, each of which is led by a Medical Officer of Health. Their responsibilities include maternal and child health and infectious disease control. The separate curative care network offers comprehensive services but is less well- organized, as there is no referral system and many patients bypass lower-level facilities in favor of secondary and tertiary care. The private sector also plays an important role in Sri Lanka’s health system equilibrium, especially in the context of outpatient curative care, where government doctors can supplement their salaries during off-duty hours and patients willing to pay out-of- pocket can receive more convenient, personalized care. Sri Lanka has undertaken relatively few health financing reforms over the years. There is no purchaser-provider split: financing and delivery are fully integrated with the national and provincial governments both funding and operating public facilities across the country. Unlike many countries, it has not opted for a demand-side financing approach to UHC which seeks to identify and target the poor. Line-item budgets predominate. There is no output, performance, or results-based financing of health care providers, no conditional cash transfers, and no social health insurance. The prevailing input-based approach to inter-fiscal transfers and provider payment has contributed to cost containment but also a somewhat unbalanced pattern of resource allocation across provinces and facility types. Sri Lanka’s government health spending as a share of Gross Domestic Product (GDP) is relatively low by regional standards, and is currently constrained by a very low level of overall government revenues. Out-of-pocket (OOP) spending accounts for about 40 percent of total health expenditures. Evidence on utilization patterns suggests that access to care in the government sector is pro-poor. The bottom 40 percent is more likely to use public outpatient care than the top 40 percent, while inpatient use is quite equal across all groups. While there is no explicit targeting, utilization patterns suggest that there is implicit targeting of the poor, largely because the better-off opt out due to the “consumer experience” of accessing health care in the public sector. In particular, government operating hours are less convenient, waiting times are much longer, and provider 10 choice is more limited. This is one of the weak links in the public system’s performance. However, evidence suggests that the quality of care across public and private sectors is similar. Sri Lanka provides a comprehensive range of health care services, but does not have an explicit benefit package defining what is available to the population. Almost 100 percent coverage has been achieved for services such as ante-natal care, skilled birth attendance, and DPT3 immunization. While a range of noncommunicable disease (NCD) services are available, including cardiology and oncology, in reality there may be significant rationing of advanced care in the form of waiting lists or limited availability of the required specialized human resources, equipment, and drugs. A significant share of total OOP spending is associated with payments for lab tests and drugs for public sector consultations. But overall utilization rates are similar to those in many high-income countries. OOP spending accounts for about 40 percent of total health expenditures, but financial protection is good, largely because most OOP is incurred by the rich. About half of all OOP is incurred by the richest decile, and the better-off also spend a larger share of their total household expenditures on health. Standard indicators of financial protection indicate that Sri Lanka is performing well relative to its peers. Looking forward, the pending agenda for Sri Lanka’s health system stems largely from the challenges associated with an aging population. Addressing NCDs is a more complex task than delivering on the maternal and child health agenda, and will entail new approaches to service delivery. It is not evident yet that Sri Lanka is performing well in terms of managing NCDs, nor that the health system as it is currently configured is well positioned to do so. The reforms needed to rise to this challenge will require strengthening all aspects of the health system – financing, human resources, pharmaceuticals, and information systems – to meet the needs of a more costly and complex service delivery system. In brief, the very positive narrative that has accompanied Sri Lanka’s health system performance over several decades may become increasingly untenable in the years ahead unless this agenda is successfully addressed. Sri Lanka’s high-achieving health system raises the obvious question of what lessons, if any, it can offer to the rest of the world. These are likely to be most relevant in lower-income countries, since its successes have largely been achieved with respect to the low-income health agenda (addressing maternal and child health) rather than the middle-income agenda (managing non- communicable diseases and a more complex health system). One obvious implication of Sri Lanka’s experience is that it challenges the common wisdom about the appropriate timing of health financing reforms, and in particular that provider payment reform should be pursued without delay by all low-income countries to somehow improve “incentives” or “efficiency”. A second important lesson is that achieving excellent maternal and child health outcomes can be done with low levels of spending. The common claim that large health budgets or even a high share dedicated to primary care are essential to improve outcomes do not find support in Sri Lanka’s experience. Third, Sri Lanka’s health system equilibrium adds considerable nuance to the often stale debate about the appropriate role of public and private sectors. At the same time, there are also several ways in which Sri Lanka appears to differ from other countries, potentially limiting the applicability of its experiences to other settings. These include better governance and the status of women. 11 Introduction Sri Lanka’s health system has a long track record of strong performance. For at least 50 years it has achieved much better outcomes in maternal and child health and infectious disease control than would have been predicted by its income level. In fact, Sri Lanka represents the global frontier for indicators such as infant mortality: no country with a similar income level has better outcomes (figure 1). It started the Millennium Development Goal (MDG) era in 1990 with a lower level of child mortality than many East and South Asian countries achieved by the end of the era in 2015. Its maternal, under-five and neonatal mortality rates are already less than half the respective Sustainable Development Goal targets for 2030. Communicable diseases such as malaria, polio, tetanus, and measles have been eliminated or are at near elimination state. Total fertility is at the replacement rate of 2.1. There is also nearly 100 percent coverage for services such as antenatal care, institutional deliveries, and childhood immunizations. Health financing indicators also compare favorably. Sri Lanka’s good health has been achieved at low cost: government health spending has generally been around 1.5 percent of GDP. It is also a relatively equitable health system that provides good financial protection. As a result of these achievements, for over 30 years Sri Lanka has been recognized as a star performer in global health (Halstead, Walsh, and Warren 1985). There are few if any other low- or middle-income countries (see box 1) that have simultaneously achieved strong health outcomes, good financial protection, and low cost. Many countries are lauded for achieving two out of three, but few can claim to have done as well as Sri Lanka on all three, especially in view of its status as a lower-middle-income country. Thailand and Turkey are commonly cited as UHC success stories, but they are upper-middle-income countries and spend much more on health as a share of GDP than Sri Lanka (without significantly better outcomes). Malaysia has also done well on all three criteria, but it is almost three times richer than Sri Lanka in per capita GDP terms. Singapore has good health at low cost but it is a high-income city state, and financial protection has been a weak link in its health system performance. On the surface, Cuba also appears to have done well, but there is uncertainty about its data. The nearby Indian state of Kerala has strong health outcomes and low spending, but OOP spending accounts for two- thirds or more of total expenditures. Figures 1a and 1b show some comparisons. Sri Lanka’s excellent performance has been achieved despite (or perhaps because of) the fact that few major reforms have been undertaken over many decades. Sri Lanka has provided universal, free access to government-provided health care services to its population since the 1930s. There is no purchaser-provider split: financing and delivery are fully integrated with the government both funding and operating public facilities across the country. Within this system there is no explicit targeting of priority population groups and no explicit benefits package. Health care workers are government employees. There is no output, performance, or results-based financing, and no social health insurance. Hospitals have limited autonomy in terms of having “decision rights” over key inputs such as human resources, and there is no accreditation system. Nor has Sri Lanka adopted demand-side interventions such as conditional cash transfers. In brief, while many countries have opted for a demand-side approach to pursuing UHC by identifying the poor (Cotlear et al. 2015), Sri Lanka has focused its efforts on the supply side. 12 Figure 1a Infant Mortality Rate Figure 1b Government Health Expenditure (per 1,000 live births) (% of GDP) This case study describes the main features and achievements of Sri Lanka’s high-performing health system, to distill lessons for the rest of the world. UNICO case studies focus on a particular health coverage program. In Sri Lanka, the selected health coverage program is the government’s national health service (NHS) that was established in the 1930s and has remained broadly the same since then. Thus, the entire government health system is taken as the health coverage program, without narrowing the focus to a specific program. The fact that no specific program or subsystem focused on the poor was ever created is part of the storyline. As will be discussed, this has not prevented Sri Lanka from achieving a pro-poor health system design. The relationship between the government system and the private health sector in Sri Lanka will also be explored in the case study. Looking forward, the pending agenda related to addressing the health needs of an aging population is also discussed. Box 1 Sri Lanka at a Glance Sri Lanka is a lower-middle-income country with a population of 21 million. By official definitions, the population is about 80 percent rural, although other measures of “agglomeration” suggest that the true figure is closer to 60 percent. Its gross national income per capita in 2015 was about US$3,800, more than twice that of its South Asian neighbors Bangladesh, India, and Pakistan. Human development outcomes are also significantly better than elsewhere in the region. In many respects, Sri Lanka’s more natural country comparators are in East Asia, not South Asia. Sri Lanka enjoyed rapid economic growth for over a decade, and this has led to a significant decline in poverty. The headcount fell from over 22 percent to just over 4 percent between 2002 and 2015 based on the national poverty line. Despite these impressive gains, a significant share of the population remains vulnerable (just above the poverty line). Economic growth has slowed more recently, and prospects going forward face significant obstacles. Public debt is about 75 percent of GDP. Foreign direct investment has been relatively low, and there are structural challenges facing key export sectors such as tea and garments. Nevertheless, Sri Lanka is a country with significant economic potential, based on its strategic location, tourism industry, and other attributes. 13 General Overview: Supply Side The focus of this case study is Sri Lanka’s government health system, which provides services free of charge to the entire population at public facilities. The system is funded by general tax revenues, and these resources are managed by the Ministry of Health (MoH) or its nine provincial counterparts. There is integrated financing and delivery: the MoH (both central and provincial) simultaneously funds and operates health facilities across the country. Thus, Sri Lanka’s government health system resembles the classic Beveridge “national health service” or NHS model most closely associated with postwar Great Britain. However, in Sri Lanka the coexistence of the government health system with a significant private sector (described below) that is funded through out-of-pocket payments and largely staffed by off-duty government doctors makes it distinct from the classic NHS model. The health sector was devolved in 1987, with a significant role for nine provincial MoHs. The MoH is responsible for stewardship functions such as policy making, development of guidelines, program monitoring and technical oversight, the purchase and distribution of drugs and consumables, human resources training and deployment (for provinces as well as the center), and the operation of tertiary and a few other selected hospitals. Nine Provincial MoHs, each with a department headed by a Provincial Director of Health Services and 26 health districts each led by a Regional Director of Health Services, are responsible for primary and secondary levels of curative care and all preventive services. There is a comprehensive network of government health facilities, divided into separate preventive and curative care networks. This includes about 650 hospitals (50 under MoH and 600 under provincial management), about 500 outpatient facilities (provincially managed), and 325 Medical Offices of Health area offices providing preventive care including all maternal and child health and public health issues. A clear separation between preventive and curative service provision at the local level is a unique feature of the Sri Lankan health system (see Annex). Government facilities account for approximately 50 percent of all outpatient visits, 95 percent of all hospital stays, and nearly 100 percent of preventive health care. Geographic access to government services is very good. Households are on average just 2.5 kilometers from a maternity clinic, 4 kilometers from a government dispensary, and 6.5 kilometers from a hospital. About 93 percent of the population is within 15 kilometers of a hospital. Outpatient services are open until 4pm, but emergency care is available 24/7. The existence of widely accessible care is aided by the fact that Sri Lanka has a relatively high population density. Excluding city states and small island states, it is the 15th most densely populated country in the world, and sixth among countries with a population over 20 million (but neighboring India and Bangladesh are denser). Preventive health care services are provided through a well-planned network of facilities across the country, each of which is led by a Medical Officer of Health. Each Medical Officer of Health covers a well-defined geographic area with a population ranging from 50,000 to 100,000. The Medical Officer of Health also coordinates the smaller curative care institutions and other local bodies in the area. Medical Officer of Health teams include Public Health Nursing Sisters, Public 14 Health Inspectors, and Public Health Midwives. The responsibilities of Medical Offices of Health include ante- and postnatal care, child growth monitoring, immunization, family planning, health promotion, nutrition counselling, communicable disease prevention, school and environmental health, food safety, and any health issues related to disaster management. All Medical Offices of Health services are provided free of charge. Given their central role in providing maternal and child health services, Medical Offices of Health also deserve a significant amount of the credit for Sri Lanka’s spectacular MDG-related health outcomes. Public Health Midwives, particularly, have played a critical role in this success story. There are 6,000 Public Health Midwives island-wide, each covering a population of 2,000 to 4,000. They provide basic family planning services, antenatal care, and both home-based and clinic care for pregnant women and children under five. Overall, public health services are underpinned by strong management and reliable public financing (Das Gupta et al. 2013). The separate curative care network offers a comprehensive list of services but is less well organized. There is no referral system and many people bypass lower-level facilities in favor of secondary and tertiary care. As noted in the Ministry of Health’s Annual Health Bulletin, “In Sri Lanka a referral system is not enforced. Hence, patients bypass small medical institutions, particularly those in the rural areas that have only minimal facilities for patient care. This leads to under-utilization of small institutions and overcrowding in the bigger institutions.” This pattern is particularly problematic for addressing NCDs, for which a strong doctor-patient relationship and integrated care are particularly beneficial. There is a notable absence of a primary care model based on family physicians or general practitioners. Recently, there has been an effort to introduce healthy lifestyle centers to help address NCDs, but the population has yet to embrace this model. Each year only about 10 percent of the adult population visits a healthy lifestyle center. In brief, the preventive side of Sri Lanka’s health system which provides most maternal and child health services has been very successful, while the curative side remains a work in progress (even though it has made significant contributions of its own, such as in the area of safe deliveries). Whether Sri Lanka continues to be viewed as a high achiever in 10 or 20 years will depend largely on how well it responds to the NCD challenge. Sri Lanka has a well-performing health workforce. Approximately 120,000 staff are employed by the government’s health system, about half under the central MoH and half under the provinces. Many countries struggle with health worker motivation and absenteeism, but these issues do not appear to be a significant problem in Sri Lanka, even though government salaries are relatively low and there is no pay-for-performance. In effect, Sri Lanka’s health care workforce appears to be “intrinsically motivated,” a characteristic that has proved elusive in many other health systems. Fair, transparent human resource policies and good infrastructure help in this regard. While the size of the health workforce is broadly similar to other Asian countries, it falls far short of levels in advanced health systems. Sri Lanka has almost one doctor and two nurses per 1,000 population. These ratios are over three times higher in OECD countries. While there has been steady growth in key health staff in recent years, health workforce statistics indicate severe shortages in some staff categories such as pharmacists, physiotherapists, and medical laboratory technologists, as well as certain medical specialties such as dentists, cardiologists, oncologists, geriatricians, and others. There is also a need to develop a family physician model for primary care, an agenda now gaining momentum. Health workforce distribution is biased toward urban 15 areas, especially Colombo and Kandy districts, while more isolated areas such as Mullaitivu and Monaragala are short‐staffed. The basis on which human resources for health are distributed across the country is somewhat ad hoc. There is also scope to better use the existing workforce through task-shifting. The government directly procures pharmaceuticals, but not in sufficient quantities to meet population needs. The State Pharmaceutical Corporation purchases drugs for the public sector based on global tendering, with an emphasis on generic and bulk purchasing. Cross‐country affordability studies also point to low prices in Sri Lanka (Mendis et al. 2007), although these comparisons may not reflect the full cost of publicly procured drugs, which would include wastage, staff costs, and sunk capital costs for distribution. About 29 percent of pharmaceutical spending is funded through the public sector and mostly covers drugs dispensed through government hospitals. These are all provided free of charge. However, the government is not able to purchase sufficient quantities of drugs to meet population demand due to budget constraints. Essential drugs are given high priority. In practice, nonessential drugs that are accorded lower priority include those for chronic NCDs. Availability of drugs in the government sector could be improved, especially at lower levels. The result is that the population regularly pays out-of- pocket for drugs, including when seeking care at public facilities. Recent policy initiatives to address this challenge offer some promise. While the focus is on the government system, the private sector also plays a critical role in Sri Lanka’s health system “equilibrium.” It is largely funded by OOP spending and is staffed for the most part by off-duty government doctors engaging (legally) in dual practice. It is responsible for half of all outpatient contacts but fewer than 10 percent of inpatient stays. On the demand side, it provides an important outlet for the excess demand that arises in the public sector due to its free provision of a good service. For outpatient services, it offers more convenient hours (government outpatient departments are closed by midafternoon) and shorter waiting times (a visit to a private doctor is less than half as long as a visit to the outpatient department of a government hospital). In addition, by opting for the private sector, a patient can exercise a choice over the doctor they will see, whereas in the government facility they are typically assigned to whomever is available when they reach the front of the queue. For inpatient services, the private sector provides advanced care with less waiting time to those who can afford it. On the supply side, the private sector provides an opportunity for government doctors to earn extra money (their salaries are below market rates) without resulting in a large-scale exodus of medical professionals from the public sector. Doctors also benefit from the prestige and the high service volumes that characterize service in government facilities. Despite widespread dual practice, absenteeism in the public sector is not a major problem. However, the potential for conflicts of interest—for example, in the context of patient referrals back and forth between the public and private sectors—is an understudied issue. Overall, supply-side efforts to ensure strong service delivery have been the cornerstone of Sri Lanka’s UHC agenda. The health system’s achievements provide strong support to the notion that supply-side reforms are crucial to UHC, and health financing reforms on their own are not enough. The next section reviews the health financing landscape in Sri Lanka. 16 General Overview: Health Financing Sri Lanka has undertaken relatively few health financing reforms over the years. This is significant because the road to UHC is sometimes equated with the implementation of major health financing reforms. This section briefly describes Sri Lanka’s “old-fashioned” health financing system with respect to resource mobilization, pooling, and purchasing. The two major sources of health financing in Sri Lanka are the government budget and out-of- pocket payments, which together account for over 90 percent of total health expenditures. There is no social health insurance other than a very small contributory scheme for civil servants called Agrahara. Voluntary prepayment schemes account for about 6 percent of total health spending. External aid is a minor financing source. Total health expenditure per capita in 2013 was about US$97. Sri Lanka’s government health spending as a share of GDP is relatively low by regional standards, but closer to the average when measured as a share of the government budget. The government health budget of 1.6 percent of GDP is closer to the low spenders among other Asian countries at a similar income level (figure 2a and 2b). It has remained around this level for many years. In some countries, this may be a sign of low prioritization given to health within the government budget, but this is not the case in Sri Lanka, where it absorbs about 8.4 percent of the budget, only slightly lower than average among Asian comparators. Figure 2a Sri Lanka’s Government Health Figure 2b:…but Closer to the Average as a Spending is Relatively Low as a Share of GDP… Share of Total Government Spending Source: World Development Indicators. Source: World Development Indicators. Note: Figures are 2010-14 averages. Note: Figures are 2010-14 averages. 17 Low government revenues are the main reason for Sri Lanka’s relatively low level of health spending. Government revenues as a share of GDP have fallen from nearly 20 percent two decades ago to about 12 percent today, one of the lowest rates in the world and on a par with a much less developed country like Bangladesh. There are a number of structural challenges to strengthen overall revenue mobilization in Sri Lanka, including an excessive reliance on indirect taxes (for example, VAT, customs). Receipts from direct taxes (for example, particularly on income) are low. This is a major impediment not only for greater spending on health, but on a wide range of development priorities, including education, infrastructure, and others. A number of factors have contributed to Sri Lanka’s low-cost system. One reason for low health spending in Sri Lanka is relatively low (below market rate) salaries in the government sector. Indeed, this is true of the entire civil service, of which doctors are a part. Sri Lanka also procures drugs at a low price due to international competitive tenders. Thus, low prices of major inputs to the delivery of health care have played an important role. Finally, the use of line-item budgeting has also helped ensure cost control. Out-of-pocket spending accounts for about 40 percent of total health expenditures in Sri Lanka, which is about average for its income level (figure 3). This includes costs for diagnostics and drugs even when visiting public facilities, the use of the private sector for about half of all outpatient services, and occasional high-cost spending at private hospitals, mainly by the rich. The breakdown of OOP payments by households based on 2013 data is shown in figure 4. Figure 3 Reliance on OOP Spending in Sri Lanka is Typical for its Income Level Source: World Development Indicators. 18 Figure 4 What Do People Pay Out-of-Pocket For? Source: World Bank calculations based on Household Income and Expenditure Survey (HIES) 2015/16. Additional revenue mobilization will be essential over the long term to address the needs of an aging population, but at present there are relatively few options for additional fiscal space. While the outlook for economic growth is promising, the low level of tax revenues severely constrains the extent to which growth will translate into additional spending. A reprioritization within the budget in favor of health would help, but as shown in figure 2a and 2b, the current budget share allocated to health is close to the regional average. Overseas development assistance plays only a small role. There is scope for achieving efficiency gains in all health systems, but Sri Lanka’s health system is already quite efficient, and therefore this is unlikely to generate significant additional resources. An additional potential source of fiscal space, which has received occasional attention in public policy circles, would be to introduce a social health insurance system that raises funds through a payroll tax. The formal sector accounts for about 37 percent of total employment in Sri Lanka. Depending on the design specifics, this could have significant implications for system-wide equity and efficiency. However, no concrete proposals on this issue have been forthcoming. In principle, Sri Lanka has a single health financing “pool” within the government health system. Any individual can seek care at any government facility in the country, implying a single pool for sharing revenues among the population (that is, healthy and sick, young and old, rich and poor). In reality, geography does matter for where people seek care, and therefore any imbalances in provincial allocations can be an impediment to cross-subsidization. Almost 75 percent of the government health budget is managed by the MoH (Ministry of Finance Annual Report 2015). These resources are allocated largely according to historical patterns related to infrastructure and staffing. Provincial budgets are based on three different transfers from the center to provinces as determined by the Finance Commission. By far the largest is a block grant intended for recurrent spending at provincial hospitals (mainly personal emoluments but also maintenance). This is determined largely by allocations in previous years reflecting approved civil servant cadres, and is therefore not needs-based. The other two inter-fiscal transfers follow a formula that is intended to 19 reflect need, but these are much smaller. Thus, overall the mechanism for allocating MoH funds among provinces is not well aligned with population need. The allocation of resources to health care providers in Sri Lanka is input-based. The MoH and its provincial counterparts allocate the budget for operational costs to each hospital largely based on its historical budget, which is closely related to the number of beds and staff rather than on demand, disease burden, or productivity. The MoH makes decisions regarding human resources, adjustments to bed numbers, or to other hospital services as required to meet population health needs, including for provincial hospitals. Since human resources account for about half of total government health spending, the allocation of staff across facilities is de facto a major aspect of resource allocation in Sri Lanka, but this is done in a relatively ad-hoc manner. Thus, the allocation of resources to health care facilities is not tied to the actual delivery of health care services or population needs. Input-based financing has, however, served Sri Lanka well in the sense that it has been a reliable cost-containment strategy. Some nonclinical services (for example, housekeeping) are contracted out, and there is a recent initiative to contract private hospitals for provision of some advanced care (for example, cardiac surgery). The result of the prevailing input-based approach to inter-fiscal transfers and facility funding is a somewhat unbalanced pattern of resource allocation. A disproportionate share of government health spending is allocated to certain provinces and higher-level facilities (figure 5a and 5b). For example, poorer provinces such as Eastern and Sabaragamuwa have the lowest per capita allocations in the country. With regard to facilities, a significant majority of spending is directed to secondary and tertiary hospitals. Primary health care in the curative system accounts for less than 15 percent of the budget, yet this is where the largest health gains could be achieved through cost‐effective chronic disease management. It is also notable that only 10 percent (and by some estimates even less) of the health budget is spent on the preventive system that has largely been responsible for Sri Lanka’s excellent maternal and child health outcomes. Thus, contrary to common perception, Sri Lanka does not spend a large share of its budget on preventive and primary care. The implication is not that hospitals were prioritized, but rather that good-quality primary care (for example, for maternal and child health) can be provided cheaply. 20 Figure 5a Estimated Per Capita Figure 5b Government Health Expenditures Government Spending by Province, 2013 by Provider Type, 2013 Source: MoH 2016. Source: MoH 2016. Interpreting UHC in Sri Lanka: People, Services, and Financial Protection The goal of universal health coverage is to ensure that everyone has access to the health care they need without suffering financial hardship. This definition implies that there are three dimensions of coverage: the population (who is covered?), services (what is covered?), and cost sharing (what proportion of costs are covered?). These are commonly illustrated by the “UHC cube” diagram. This section assesses Sri Lanka’s health system in terms of each of these three dimensions of UHC. Covering People The foundation of Sri Lanka’s equitable health system is the long-standing policy of providing free universal health care for the entire population. For the most part, and unlike many other countries, this commitment is upheld. There are no copayments in the government sector and virtually no informal payments. A brief experiment with user fees started and ended in the 1970s. There is no system of explicitly targeting specific population groups and there is no enrolment requirement to obtain care at government facilities. Residency implies access to any and all government facilities in the country. Indeed, even noncitizens (for example, migrant laborers) are typically provided with free care, although there is no explicit policy on this, making Sri Lanka more generous on the “people” side of the UHC cube than many high-income countries. There is a community targeting system to determine eligibility for Sri Lanka’s cash transfer program for the poor (“Samurdhi”), but this has not been applied in the health system. 21 Evidence on utilization patterns suggests that access to care in the government sector is pro-poor. Panels A and B of figure 6 show the share of the population that recently used outpatient and inpatient services across the public and private sectors. The bottom three quintiles are more likely to have used outpatient services than the top two quintiles. Inpatient service utilization is equal across quintiles. (Note that these data capture whether an individual used services, not how many times.) The richer quintiles are more likely to use the private sector, especially for outpatient care. The top quintile is about three times as likely as the bottom quintile to do so. As noted, government health services account for about half of all outpatient care, and over 90 percent of inpatient care. The utilization patterns show that it is the better-off segment of the population that is opting out of government services in favor of private providers, especially in urban areas. If public and private care is added together, the rich use more services. Outpatient utilization patterns suggest that there is implicit targeting of the poor, largely because of the “consumer experience” of accessing health care in the public sector. An outpatient visit to a government hospital typically lasts at least twice as long as a consultation with a private doctor due to waiting time, and outpatient services close by 4pm (figure 7). In addition, patients see whichever doctor is available when they reach the front of the queue. Thus, one way to interpret the existence of widespread legalized dual practice is that if a citizen would like to see the (public) doctor of their choice, at a more convenient time, with less waiting time, they must pay out-of- pocket for it by going to see that doctor in their private practice. Put differently, the private sector can be interpreted as an extension of the public sector, but with a different provider payment mechanism (fee-for-service instead of salary), a different co-payment policy (100 percent instead of 0 percent), and at a different time of day (after 4pm instead of before). The cost of private consultations is relatively cheap for all but the poor, in large part because they are competing with a free public sector and therefore must offer a low-cost service to be competitive. Figure 6 Utilization Patterns Panel A Panel B Source: HIES 2015/16. 22 Figure 7 Average Time Spent at Facilities per Visit Source: World Bank 2017. This picture is somewhat different for inpatient stays, however. Over 90 percent of hospital admissions are in government hospitals. Thus, few people are opting out of government hospital care, and those who do are likely to be relatively rich and seeking services that are either not available in public hospitals or not without a significant wait (for example, there is reportedly a long waiting list to receive bypass surgery in the public sector). Thus, self-selection out of the government system typically happens for different reasons (and on a very different scale) for outpatient and inpatient care. The implications of these utilization patterns for equity are uncertain. To the extent that they also translate into better health outcomes for those who use the private sector (for example, because they are less likely to forego a test due to the inconvenience, or more likely to get a treatment that is not available in a government facility without a significant wait), there would be implications for the equity of Sri Lanka’s health system. But it is difficult to establish that this is the case with available evidence. One study found that the average consultation time for outpatient visits was about 3.1 minutes in public facilities and 7.8 minutes in private facilities. However, the same study found similar quality of care in the public and private sectors, which is perhaps not surprising since the doctors are largely the same (Rannan-Eliya et al. 2015). For inpatient care, if inequality arises due to the pattern shown in figure 6, it would be between the very rich and everyone else, rather than the poor and the rest of the population. Overall, Sri Lanka’s health system is pro-poor, and except for the convenience factor it does not appear to be the case that “health services for the poor are poor health services,” a claim sometimes voiced about other countries. A global study using benefit incidence analysis found Sri Lanka’s government health system to be the second most pro- poor out of 66 countries studied (Wagstaff et al. 2014). An important historical exception to the pro-poor nature of Sri Lanka’s health system was the status of the population living in the Estates (plantation) sector. This is the (mainly) tea-growing area that is concentrated in the central hills. They were disenfranchised around the time of independence and did not receive government health services until the 1970s. The Estates accounts 23 for about 5 percent of the population and has suffered the highest poverty rates, but social indicators are now starting to catch up with the rest of Sri Lanka. Covering Services Sri Lanka offers a wide range of services, but does not have an explicit benefits package defining the services available to the population. While a defined benefits package is a key feature of health systems with insurance and output-based payment, it is sometimes adopted in other systems as a means to strengthen accountability, by making it clear to providers what they are expected to deliver, and to patients what they can expect to receive. Sri Lanka does have a comprehensive implicit benefits package, with a full range of services (including even cosmetic surgery and costly oncology drugs for at least some patients) provided to the population. However, investment decisions (for example, for high-tech equipment) are not based on robust health technology assessments (for example, cost-effectiveness analysis, and so forth). Sri Lanka’s relatively high service coverage is apparent in cross-country data. Almost 100 percent coverage has been achieved for services such as antenatal care, skilled birth attendance, and DPT3 (diphtheria, pertussis, and tetanus) immunization. However, coverage of key NCD services lags behind (table 1). Table 1. Tracer Universal Health Coverage Indicators Sri Lanka Lower-Middle (%) Income (%) Prevention and Health Promotion Family planning 68 46 Antenatal care 99 86 Skilled birth attendance 99 74 DPT3 immunization 99 86 Tobacco nonuse 85 78 Access to improved water 94 83 Access to improved sanitation 92 59 Treatment Antiretroviral therapy 19 29 Tuberculosis 86 82 Hypertension 21 27 Diabetes 10 11 Source: World Development Indicators. However, not all services are readily available. There is significant rationing, whether in the form of waiting lists or limited availability of the required specialized human resources, equipment, and drugs to provide certain services. To take one example, it has been reported in recent years that over 5,000 people are on the waiting list to receive cardiac surgery. As noted, lab tests and drugs account for about one-third of total OOP spending, and at least some share of this is related to 24 public sector consultations. For example, a 2016 survey found that lab costs paid out of pocket in association with admissions to divisional, base, and provincial hospitals range, on average, between 300 Sri Lankan rupees and 600 Sri Lankan rupees (US$2 to US$4), and considerably higher at the National Hospital. Recent policy initiatives have aimed to address this issue. There is also a geographic dimension to rationing. There are shortages of specialists in the government system, especially outside Western Province. For example, except for general surgery, obstetrics and gynecology, pediatrics, and anesthesiology, the other eight provinces typically have fewer than 10 specialists to cater to their populations of between 1 million and 2.5 million people. The population is free to travel anywhere in the country to seek care, but not everyone does so (especially, perhaps, the poor). More broadly, the maldistribution of specialists is reflective of budget allocations that tend to favor higher-level facilities (secondary and tertiary hospitals). While implicit rationing happens, it is also notable that utilization rates of health care services in Sri Lanka are very high. In the government sector, there are close to three outpatient visits per capita annually, with roughly the same in the private sector. Together these rates are similar to those in advanced systems. The system manages high utilization in part through short consultation times, as noted. There is no referral system. The population can and does bypass lower levels of care as a matter of course. With respect to “effective” coverage, existing studies suggest that the quality of outpatient and inpatient care in the government sector in Sri Lanka is good. This includes the key domains of history taking, examination, investigations, and management (Rannan‐Eliya et al. 2015). Private sector patients receive better quality mainly in nonclinical areas such as interpersonal satisfaction. Financial Protection Out-of-pocket spending accounts for about 40 percent of total health expenditures, but financial protection is quite good. This is largely because most OOP spending is incurred by the rich (figure 8). In total, nearly half of all OOP spending is incurred by the richest decile, who spend almost three times more per month than even the second-richest decile. The better-off also spend a larger share of their total household expenditures on health. Meanwhile, the entire bottom (poorest) 40 percent of the population spends on average less than US$4 per month, or less than 2.5 percent of their total household expenditures on health. 25 Figure 8 Out-of-Pocket Payments by Decile, 2015-16 Source: HIES 2015/16. Standard indicators of financial protection indicate that Sri Lanka is performing well. The incidence of catastrophic out-of-pocket payments (that is, health expenditures that account for a large share of total household spending) is modest by international standards (figure 9). For example, 6.4 percent of households had health expenditures exceeding 10 percent of total household spending in 2015/16, compared to 7.7 percent in the Philippines, 13.9 percent in India, 14.2 percent in Bangladesh, and 19.6 percent in Vietnam. Catastrophic health expenditures in Sri Lanka are more frequent among the richer quintiles: about 4 percent of the richest decile faces catastrophic health expenditures. Impoverishing OOP spending (that is, pushing a household below the poverty line) remains rare in Sri Lanka. Only 0.4 percent of Sri Lankan households were pushed into poverty by their payments for health in 2015, compared to 1.5 percent in the Philippines, 1.7 percent in Vietnam, and 3.5 percent in Bangladesh. This estimate for Sri Lanka is based on the current national poverty line, which is generally viewed as low, and is under revision. Given relatively high utilization rates, it is not the case in Sri Lanka (as in some countries) that good financial protection reflects non-use of services. Sri Lanka is an exception to the global finding that catastrophic OOP spending becomes more common once the OOP share exceeds about 25 to 30 percent of total health spending (Xu et al. 2010). There are two reasons. First, Sri Lanka’s OOP spending does not translate into high catastrophic and impoverishing expenditures because such a large share is concentrated among the rich. Second, the level of OOP spending (about 1 to 1.5 percent of GDP) is lower than in many countries. In other words, the 40 percent OOP share appears high in part because government health spending is itself low. A country like Vietnam has a similar OOP share but higher incidence of catastrophic expenditures because total health spending (and thus OOP) is significantly higher. (Hypothetically, if salaries and drug prices in Sri Lanka were much higher to the extent that the 26 government health budget was twice as large, the OOP share would appear lower without materially changing anything). Figure 9 Incidence of Catastrophic Health Spending, 2015-16 Source: HIES 2015/16. While it is preferable that most OOP spending is incurred by the better-off and not the poor, there are still good reasons to reduce OOP spending over the longer term. First, the relatively well-off should also have some protection against the risk of high medical costs. In economic jargon, financial protection is important due to market failures in insurance (for example, adverse selection), and not only due to equity concerns. Second, OOP spending by the rich represents a foregone opportunity to pool health care resources in a way that would facilitate cross- subsidization of the poor by the rich. For example, if income taxes (which mostly fall on the rich) were increased, it would help generate more resources for the health sector, and thus could improve equity by enhancing access among all quintiles. Pending Agenda The pending agenda for Sri Lanka’s health system stems largely from the challenges associated with an aging population. Sri Lanka is in the advanced stages of a demographic and epidemiological (disease) transition. Population aging will happen quickly—the share of the population over 60 will roughly double between 2015 and 2035. Chronic NCDs, including diabetes, heart disease, stroke, asthma, cancer, kidney diseases and mental health, already account for about 80 percent of the disease burden (IHME 2015). A significant share of NCD deaths occur prematurely, before the age of 70. These changes are rooted in an aging population but also broader development trends, including a growing economy, urbanization, and lifestyle changes such as low levels of physical activity, poor diet, and alcohol use. Increased motorization is contributing to a rising burden of road traffic injuries and deaths. Fortunately, Sri Lanka has taken a proactive approach to tobacco control, most notably through high cigarette taxes, and therefore does not face a tobacco-related disease burden as large as many (less foresighted) middle-income countries. The trend of key health indicators such as life expectancy suggests that Sri Lanka is struggling to make 27 progress against NCDs. While the average lifespan is relatively good at almost 75 years (slightly above what Sri Lanka’s income level would predict), progress has been slow over the past 10 to 15 years. Addressing NCDs is a more complex task than delivering on the maternal and child health agenda, and will require new approaches to service delivery. There is a noticeable absence of a strong primary care system for NCDs, including general practitioners. As noted, the population commonly bypasses lower-level facilities and seeks care directly at secondary and tertiary hospitals (from doctors whom they may or may not have ever seen). This pattern will become increasingly unsustainable. Rising to this challenge will require strengthening all aspects of the health system—financing, human resources, pharmaceuticals, and information systems. It is notable that in many respects Sri Lanka’s health system does not “look like” most health systems in high-income countries. For example, the information system is still paper-based; drug procurement is still done by the public sector; provider payment is input-based; facilities do not have autonomy over decision making; and primary care does not rely on a general practitioner model. None of these are common in advanced health systems. Thus, the pending agenda can be seen as the need to gradually converge with the health system design typically found in those settings. This still leaves open many design issues; for example, Australia, Canada, and the United Kingdom all offer alternative models but none rely primarily on insurance. Here we briefly discuss what this might entail for financing, human resources, pharmaceuticals, and information systems. The major health financing issue at present is how to increase the health budget in a manner that improves NCD outcomes. Making progress on this agenda will have to precede many of the other pending reforms described below. In 2015, the government announced a commitment to roughly double the share of GDP that it spends on health, up to about 3 percent. The needs are many: years of underinvestment have resulted in human resources shortages, weak infrastructure, and shortages of key inputs for selected NCDs, including lab services, drugs, and advanced care. But as noted, fiscal space options at present are somewhat limited. Aside from increased funding of the health sector, consideration could also be given to reallocating resources (across services, programs, provinces, facilities, and so forth) or changing the financing modality more fundamentally (for example, output-based financing, contracting out pharmaceuticals or lab services, and so forth). Service delivery reforms to better address NCDs will also be important for efficiency. The Sri Lankan health system’s human resources constraints will take longer to address than its budget shortages. These constraints include understaffing (advanced health systems have over three times as many doctors and nurses per capita), an acute shortage of certain specialists (cardiology, oncology, geriatrics, and so forth) and other health professionals (for example, pharmacists), and maldistribution (especially between Western Province and the rest of the country). The establishment of a family physician model would appear to be especially important. Although potentially controversial, measures such as task-shifting between doctors and nurses could help raise the productivity of the existing workforce while longer-term challenges are being addressed. 28 With regard to pharmaceuticals, the current approach to procurement ensures low prices but sometimes at the expense of quality. More efficient systems to shorten the procurement cycle are needed. The supply chain could also be strengthened to get more from the existing drug budget or more fundamental changes could be adopted. Over the longer term, converging with advanced health systems would mean contracting with private sector retail pharmacies to manage drug procurement and distribution. The government would continue to cover drugs to ensure access at little or no cost. Smart contracting could help ensure that the current low‐price environment is maintained while taking advantage of private sector capabilities in logistics and distribution. The health information system requires significant investment to meet the need for more aligned health planning, results management, and budgeting. The current system is paper- based, with limited use of e-health. The information systems for preventive health (communicable diseases and maternal and child health), hospital inpatient services, and other individual disease reporting systems are not linked. These systems generate a large volume of data, but it is not timely and not adequately used for strategic analysis, policy making, or resource allocation. Patient reporting information systems in the public sector require major improvements. Available data on indoor patients and services provided through the public health system cannot be disaggregated, as reporting is done on the basis of groups of illnesses and patient episodes. Thus, repeat visits for the same illness episode (a useful indicator of quality of care) cannot be identified. Although the MoH is currently introducing e-IMMR (electronic indoor morbidity and mortality return, or electronic data reporting of all inpatients in hospitals under the International Classification of Diseases, or ICD-10), the information available on clinic patients, the majority of whom have NCDs, is limited to the number of patients, without any data on age, gender, or diagnosis. There is little information on NCD-related health outcomes. In addition, only the total numbers of patients seen at different hospital levels is currently reported for outpatient department patients, and no information is received from private providers. The introduction of unique identification numbers for the entire population would be helpful. There is now a commitment to introduce patient health records. The government issues a detailed report of administrative data called the Annual Health Bulletin. It has recently produced its own National Health Accounts. Overall there is a weak feedback loop between data and policy making. The information system agenda, however, may be less pressing than those related to financing, human resources, and pharmaceuticals. Finally, Sri Lanka can expect to have a more complex health system in the future, and the government will need to invest in its capacity to manage this complexity. This will be true particularly in areas such as hospital management, contracting and payment systems, ensuring quality of care, and analytical capacity to undertake operational research on large volumes of information. In brief, health reform will require strong state capacity to implement complex policies and programs if they are to achieve their intended goals. What Lessons for the Rest of the World? Sri Lanka’s successful health system raises the obvious question of what lessons, if any, it can offer to the rest of the world. It is a cliché to say that each country has a unique context, but differences need not be an excuse to ignore or reject other country experiences, since there are also 29 many similarities. This section identifies some tentative lessons for other countries, and then notes some of the ways that Sri Lanka may indeed be different. It bears mentioning that Sri Lanka itself is not unique in its approach. A similar model prevails in the health systems found in Hong Kong and Malaysia (although they are much richer) and Fiji (much smaller), among others. From the outset, it should be noted that any lessons that Sri Lanka’s experience can offer the rest of the world are likely to be most relevant in lower-income countries. This is because its successes have largely been achieved with respect to addressing maternal and child health and infectious diseases—that is, the low-income health agenda. It is much less clear that it has made large strides on the middle-income agenda, which entails addressing NCDs and managing a costlier and more complex health system. One obvious implication of Sri Lanka’s experience is that it challenges the common wisdom about the importance of certain health financing reforms in low and even lower-middle income countries. The road to universal health coverage is sometimes equated with an imperative to undertake health financing reforms – such as the launch of insurance to mobilize revenues and provider payment reform to improve incentives – without delay. Sri Lanka’s performance suggests that these are by no means necessary, at least in lower-income countries, and that a lot can be achieved with effective supply-side management of human resources, drugs, and facilities to ensure effective service delivery. Reformers in other countries who believe that health financing reforms are essential to achieve better outcomes should proceed with their agenda on the basis that an exclusive supply-side focus “won’t work here” rather than asserting that “it never works anywhere.” An often-neglected issue is the appropriate timing of health financing reforms. While nearly all advanced health systems have moved away from input-based financing and introduced a purchaser-provider split, this leaves open the question of when to embark on health financing reform, given the capacity and governance constraints that make such reforms harder to implement in low-income settings. As Sri Lanka will soon achieve upper-middle-income status, its experience suggests the appropriate timing of certain health financing reforms might be later than is sometimes thought. The relevant time period is not trivial: even at robust growth rates, a country will typically spend 20 to 30 years classified as a lower-middle-income country (that is, gross national income per capita between US$1,025 and US$4,035). More generally, this suggests that health system “models” matter less than having good governance, political leadership, and strong management all working in favor of a stronger health system. The presence of these key ingredients can help achieve strong outcomes with or without health financing reform. Their absence will undermine any system no matter its design. Sri Lanka’s supply-side focus has worked well in two out of three broad categories of service delivery. The first is maternal and child health and infectious disease control. These services are usually the top priority in low-income countries for reasons of ethics, equity, and externalities. They require a strong community and grassroots focus, but demand for care is perhaps facilitated by their “acute” nature. The second category is inpatient hospital care. This is important due to the potentially catastrophic costs for poor and vulnerable households. Here the implementation challenges are perhaps mitigated by the fact that a few dozen hospitals are easier to manage than several hundred primary care units. The third broad category of services is outpatient care for 30 NCDs. Sri Lanka has not yet proven successful in this area. It is arguably the most difficult type of service delivery, both because NCDs are chronic in nature (which strongly affects the demand for care) and because success requires personalized care delivered close to home by a large number of service providers. Although Sri Lanka has not had an explicit policy as such, its experience suggests that prioritizing basic hospital care before outpatient chronic NCD management may be a well-justified approach because the former is somewhat easier to manage than the latter. Another important lesson from Sri Lanka’s experience is that achieving excellent maternal and child health outcomes can be (and should be) very cheap. This observation is based partly on the fact that government health spending has remained low for many years (around 1.5 percent of GDP). But more importantly, only a relatively small fraction of this spending is allocated to maternal and child health. Sri Lanka’s MDG achievement is largely due to the services provided by the network of preventive care providers managed by Medical Officers of Health (although a high rate of institutional deliveries by curative care providers has also played an important role). This accounts for only about 10 percent of the small overall health budget, or less than 0.2 percent of GDP. In many countries, it is claimed that health sector underperformance is due to inadequate financing, whether overall or the share allocated to primary care. But Sri Lanka’s track record suggests that this is unlikely to be the case, even while acknowledging that it probably faces lower service delivery costs than other countries due to its population density and literate population. For example, the backbone of Sri Lanka’s strong MDG outcomes is its 6,000-strong cadre of public health midwives, whose entire payroll is less than US$1 per capita. It is also notable that costs have been kept low despite high utilization rates, thanks in large part to line-item budgets and the allowance of dual practice. Sri Lanka has achieved an equitable health system without explicit targeting of the poor. A common trend in UHC programs worldwide in recent years has been to use targeting mechanisms to identify poor households and enroll them in a health coverage program. While this approach offers the potential for improved targeting of health benefits, it may also result in errors of inclusion and exclusion, as well as creating barriers to access due to potentially cumbersome enrollment procedures. Sri Lanka’s health system has not attempted to make use of targeting methods, but has achieved pro-poor results largely through self-targeting that is the result of the better off seeking more convenient, personalized care in the private sector. The reliance of Sri Lanka’s rich on OOP-financed care in the private sector is noteworthy. The UHC agenda rightly focuses on the poor, but how the rich get care has important implications for any health system. In some countries, the better-off are enrolled in a social insurance system that segments their contributions in a separate (and often better funded) pool, provides access to better providers, and undermines their political support for the mainstream government system. In a few others, employer-based private insurance plays a similar role. In both cases, cross-subsidization may be very limited. In yet others, the government budget is pro-rich because of a focus on urban hospitals instead of rural areas. In Sri Lanka, if the rich want something “better” than the government system, they have to pay out of pocket. Although this represents a foregone opportunity to cross-subsidize between rich and poor, it is arguably the “least bad” way to satisfy demand for better care by the rich, compared to the alternatives seen elsewhere. 31 Finally, Sri Lanka’s experience adds nuance to the often stale debate about the respective roles of the public and private sectors in health care delivery. A thriving, OOP-financed private sector is not always a “problem to be solved.” On the demand side, it has helped to absorb the excess demand for government services by those with the ability to pay. On the supply side, while dual practice by public sector doctors is often seen in a negative light, it has arguably played a positive role in Sri Lanka by supplementing incomes while keeping doctors available in the public system. Since the private sector is primarily used for low-cost outpatient visits—for which market failures are far less significant than for either inpatient care or preventive/public goods (both of which are dominated by the government in Sri Lanka)—the negative implications for catastrophic OOP spending and negative externalities are minimized. But one reason why private sector outpatient care offers good value to patients is because it must compete with a robust (and free) public system. Thus, one of the most important ways that the public sector can help ensure a positive contribution from the private sector is not first and foremost through “regulation” (although that can help too), but rather by participating in the market itself and offering a reliable, low cost public option. This intermediate role as a “market anchor” is more pragmatic than the two more extreme (but often advocated) alternatives of giving up on the potential of public provision on the one hand or denouncing and harassing the private sector on the other. There are also several ways in which Sri Lanka appears to differ from other countries, potentially limiting the applicability of its experiences to other settings. First, it is apparent that the status of women in Sri Lanka is relatively good, especially compared to its South Asian neighbors. This has been noted by visitors at least as far back as the 19th century (Caldwell 1986). The female literacy rate is high and this is strongly correlated with good maternal and child health outcomes globally. A survey in India in 2004–05 found that 73 percent of women reported that they needed permission from their husband to visit a health care facility, ranging from about 50 percent in the southernmost states to over 90 percent in certain northern states. The same figure is not available in Sri Lanka, but responses to similar questions confirm that women appear to be more empowered than elsewhere in South Asia (Department of Census and Statistics 2009). This has obvious implications for care-seeking behavior and, ultimately, health outcomes. Research also suggests that a prominent political voice for women helps ensure public funding for the health sector. Sri Lanka also appears to have stronger governance, as indicated by a different balance of power between politicians, providers, and the public. The medical association is powerful and strongly resists political meddling in the health system. Political patronage has not corrupted the posting and transfer system. Doctor absenteeism is rare. Majoritarian rule has favored effective public service delivery in the rural heartland rather than co-opting of the government by the urban elite. The population will readily exert its citizen power by, for example, voicing its disapproval to politicians if a scheduled antenatal clinic does not take place. These forces are collectively referred to as good accountability (World Bank 2004). They go much deeper than just having a democracy, since many other democratic countries have not achieved such success. Good governance also explains why, for example, Sri Lanka has also performed better than many of its peers with respect to education sector outcomes (enrolment, literacy, and so forth). Replicating Sri Lanka’s health system successes will be difficult in countries with weak governance, but it is not clear that the resulting policy implication should be “implement health financing reforms” as opposed to “improve governance.” 32 Other distinct Sri Lankan characteristics may also play a role. As noted, it has a high population density (especially compared to Sub-Saharan Africa, for example), which helps facilitate good access at low cost. Other explanations are more speculative. For example, 30 years of civil war might have imposed budget discipline on the health sector and discouraged reform. The role of religion and other social factors is discussed at length elsewhere (Caldwell 1986). Whether the Sri Lankan experience can be replicated in other countries or not, it offers a challenge to the conventional wisdom about the path to universal health coverage. Regardless of the route taken, hopefully more countries can repeat its successes. 33 Annex Figure A.1 District-Level Health Care Delivery System Figure A.2 Provincial-Level Health Care Administration 34 Figure A.3 Central-Level Health Administration Structure Box A1 Hospital Categorization • PRIMARY CARE HOSPITALS (MANAGED BY PROVINCES) Only medical officers are available and no specialized units manned by consultants. Central dispensaries Maternity homes Divisional hospitals type B and C. • SECONDARY CARE HOSPITALS (MANAGED BY PROVINCES) Four specialized units (Medicine, Surgery, Pediatrics, and Gynecology & Obstetrics) are available. No units for subspecialties such as ENT, Cardiology, Oncology, Eye, and so forth. Divisional hospitals type A Base hospitals Type A and B. • TERTIARY CARE HOSPITALS (MANAGED BY CENTRAL LINE MINISTRY, EXCEPT DISTRICT GENERAL HOSPITALS, BY PROVINCES) Separate units for main four specialties and for subspecialties are available. District General hospitals Provincial General Hospitals Teaching hospitals National hospitals Special hospitals. 35 References Caldwell, J. 1986. “Routes to Low Mortality in Poor Countries.” Population and Development Review 12 (2): 171–220. Cotlear, D., S. Nagpal, O. Smith, A. Tandon, and R. Cortez. 2015. “Going Universal: How 24 countries are implementing UHC reforms from the bottom up.” World Bank, Washington DC. Das Gupta, M., K. Dalpatadu, C. Shanmugarajah, and H. Herath 2013. “Multisectoral Preventive Health Services in Sri Lanka.” Policy Research Working Paper No. 6558, World Bank, Washington, DC. Department of Census and Statistics and Ministry of Healthcare and Nutrition (MoH). 2009. “Sri Lanka Demographic and Health Survey 2006–07.” Department of Census and Statistics and Ministry of Healthcare and Nutrition (MoH), Colombo, Sri Lanka. Halstead, S. B., J. Walsh, and K. Warren, eds. 1985. Good Health at Low Cost: Proceedings of a Conference held at the Bellagio Conference Center, Bellagio, Italy, April 29–May 2. New York: Rockefeller Foundation. Institute of Health Metrics and Evaluation. 2015. “Global Burden of Disease.” http://www.healthdata.org/gbd. Mendis, S., K. Fukino, A. Cameron, R. Laing, A. Filipe, O. Khatib, J. Leowski, and M. Ewen 2007. “The availability and affordability of selected essential medicines for chronic diseases in six low‐ and middle‐income countries,” Bulletin of the World Health Organization 85 (4) (April): 279–288. Ministry of Health. 2016. “National Health Accounts 2013.” Colombo. http://www.health.gov.lk/moh_final/english/public/elfinder/files/publications/NHA/Sri% 20Lanka%20National%20Health%20Accounts%202013.pdf. Rannan-Eliya, R., and L. Sikurajapathy. 2008. “Sri Lanka: Good Practice Expanding Health Care Coverage.” In Good Practices in Health Financing. World Bank: Washington DC. Rannan-Eliya, R., N. Wijemanne, I. Liyanage, J. Jayanthan, S. Dalpatadu, S. Amarasinghe, and C. Anuranga. 2015. “The quality of outpatient primary care in public and private sectors in Sri Lanka – how well do patient perceptions match reality and what are the implications?” Health Policy and Planning 30: 59–74. Wagstaff, A., M. Bilger, L. Buisman, and C. Bredenkamp (2014). “Who Benefits from Government Health Spending and Why?”. World Bank Policy Research W orking Paper No. 7044. Washington DC. World Bank. 2004. World Development Report 2004: Making Services Work for Poor People. Washington, DC: World Bank. World Bank. 2017. “Health Financing System Assessment: Sri Lanka.” World Bank, Washington, DC. 36 World Development Indicators. 2017. http://data.worldbank.org/data-catalog/world-development- indicators. Xu, K., P. Saksena, M. Jowett, C. Indikadahena, J. Kutzin, and D. Evans. 2010. “Exploring the Thresholds of Health Expenditure for Protection against Financial Risk.” Background Paper 19, World Health Report 2010, World Health Organization, Geneva. 37